computing lives - FTP Directory Listing

›± ¨ª ¨ § ±± § ±± ±´¨ ⁄±Æ ø ø¨ ± ¨Æ´ ¨ ± ¥ªø ª ¥ µ ªÆª ÕªøÆ ´ª § “ª®¨ –ø ª 

¨¨ ÊÒÒ©©©Ú ± ´¨ªÆÚ±Æ 

ÀÃÃ¤Œ ›ÿfl—ÕÙ –Ú Í 

fl“fl‘« “Ÿ fl“ 

fl“› ¤“Ã Õ›Œ –ÃÙ –Ú ÈÍ 

œÀ¤ÕÃ —“ “Ÿ 

“ Õ fi ‘ Ã«Ù –Ú ËÏ 

›± ¨ª ¨ § ±± § ±± ±´¨ ⁄±Æ ø ø¨ ± ¨Æ´ ¨ ± ¥ªø ª ¥ µ ªÆª 

ÕªøÆ ´ª § “ª®¨ –ø ª

A 

Computer Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page M S BE 

aG 

F 

Editor in Chief 

Carl K. Chang 

Iowa State University 

____________ 

chang@cs.iastate.edu 

Associate Editor 

in Chief 

Sumi Helal 

University of Florida 

__________ 

helal@cise.ufl.edu 

Area Editors 

Computer Architectures 

Steven K. Reinhardt 

AMD 

Databases and Information 

Retrieval 

Erich Neuhold 

University of Vienna 

Distributed Systems 

Jean Bacon 

University of Cambridge 

Graphics and Multimedia 

Oliver Bimber 

Johannes Kepler University Linz 

High-Performance Computing 

Vladimir Getov 

University of Westminster 

Information and 

Data Management 

Naren Ramakrishnan 

Virginia Tech 

Multimedia 

Savitha Srinivasan 

IBM Almaden Research Center 

Networking 

Ahmed Helmy 


Software 

Robert B. France 

Colorado State University 

David M. Weiss 

Iowa State University 

Editorial Staff 

Scott Hamilton 

Senior Acquisitions Editor 

_____________ 

shamilton@computer.org 

Judith Prow 

Managing Editor 

___________ 

jprow@computer.org 

Chris Nelson 

Senior Editor 

James Sanders 

Senior Editor 

COMPUTER 

Associate Editor in Chief, 

Research Features 

Kathleen Swigger 

University of North Texas 

__________ 

kathy@cs.unt.edu 

Associate Editor in Chief, 

Special Issues 

Bill N. Schilit 

Google 

schilit@computer.org 

____________ 

Column Editors 

AI Redux 


Virginia Tech 

Education 

Ann E.K. Sobel 

Miami University 

Embedded Computing 

Tom Conte 

Georgia Tech 

Green IT 

Kirk W. Cameron 

Virginia Tech 

Industry Perspective 

Sumi Helal 


IT Systems Perspectives 

Richard G. Mathieu 

James Madison University 

Invisible Computing 

Albrecht Schmidt 

University of Duisburg-Essen 

The Known World 

David A. Grier 

George Washington University 

The Profession 

Neville Holmes 

University of Tasmania 

Security 

Jeffrey M. Voas 

NIST 

Contributing Editors 

Lee Garber 

Bob Ward 

Design and Production 

Larry Bauer 

Design 

Olga D’Astoli 

Cover Design 

Kate Wojogbe 

Computing Practices 

Rohit Kapur 

rohit.kapur@synopsys.com 

________________ 

Perspectives 

Bob Colwell 

bob.colwell@comcast.net 

_____________ 

Web Editor 

Ron Vetter 

vetterr@uncw.edu 

___________ 

Software Technologies 

Mike Hinchey 

Lero—the Irish Software 

Engineering Research Centre 

Web Technologies 

Simon S.Y. Shim 

San Jose State University 

Advisory Panel 

Thomas Cain 

University of Pittsburgh 

Doris L. Carver 

Louisiana State University 

Ralph Cavin 

Semiconductor Research Corp. 

Dan Cooke 

Texas Tech University 

Ron Hoelzeman 



Virginia Tech 

Ron Vetter 

University of North Carolina, 

Wilmington 

Alf Weaver 

University of Virginia 

Administrative 

Staff 

Products and 

Services Director 

Evan Butterfield 

Manager, 

Editorial Services 

Jennifer Stout 

2010 IEEE Computer 

Society President 

James D. Isaak 

cspresident2010@jimisaak.com 

__________________ 

CS Publications Board 

David A. Grier (chair), David 

Bader, Angela R. Burgess, Jean- 

Luc Gaudiot, Phillip Laplante, 

Dejan Milojičić, Linda I. Shafer, 

Dorée Duncan Seligmann, Don 

Shafer, Steve Tanimoto, and 

Roy Want 

CS Magazine 

Operations Committee 

Dorée Duncan Seligmann 

(chair), David Albonesi, Isabel 

Beichl, Carl Chang, Krish 

Chakrabarty, Nigel Davies, Fred 

Douglis, Hakan Erdogmus, Carl 

E. Landwehr, Simon Liu, Dejan 

Milojičić, John Smith, Gabriel 

Taubin, Fei-Yue Wang, and 

Jeffrey R. Yost 

Senior Business 

Development Manager 

Sandy Brown 

Senior Advertising 

Coordinator 

Marian Anderson 

Circulation: Computer (ISSN 0018-9162) is published monthly by the IEEE Computer Society. IEEE Headquarters, Three Park Avenue, 17th Floor, New York, NY 

10016-5997; IEEE Computer Society Publications Office, 10662 Los Vaqueros Circle, PO Box 3014, Los Alamitos, CA 90720-1314; voice +1 714 821 8380; fax +1 

714 821 4010; IEEE Computer Society Headquarters, 2001 L Street NW, Suite 700, Washington, DC 20036. IEEE Computer Society membership includes $19 for 

a subscription to Computer magazine. Nonmember subscription rate available upon request. Single-copy prices: members $20.00; nonmembers $99.00. 

Postmaster: Send undelivered copies and address changes to Computer, IEEE Membership Processing Dept., 445 Hoes Lane, Piscataway, NJ 08855. Periodicals 

Postage Paid at New York, New York, and at additional mailing offices. Canadian GST #125634188. Canada Post Corporation (Canadian distribution) publications 

mail agreement number 40013885. Return undeliverable Canadian addresses to PO Box 122, Niagara Falls, ON L2E 6S8 Canada. Printed in USA. 

Editorial: Unless otherwise stated, bylined articles, as well as product and service descriptions, reflect the author’s or firm’s opinion. Inclusion in Computer 

does not necessarily constitute endorsement by the IEEE or the Computer Society. All submissions are subject to editing for style, clarity, and space. 

A 


aG 

F

A 


aG 

F 

Almost $90 million in server room 

research. Now yours is FREE! 

Implementing Energy 

Efficient Data Centers 

“ Implementing Energy 

Efficient Data Centers” 

White Paper #114 

$ 65 00 FREE! 

Power er and an and 

Cooling Coo oolin ling i g Ca Capacity Ca Capacity CThe pa pacity acityAdvantages 

of Row 

Management Ma for Data and Centers Rack-Oriented 

Five ve Basic Steps for Efficient 

Cooling Architectures 

Space pace Organization within 

High-Density gh-Density for Enclosures 

Data Centers 


An Improved Architecture 

for High-Efficiency, 

High-Density Data Centers 


“ An Improved Architecture 

for High-Efficiency, 

High-Density 

Data Centers” 


$ 120 00 FREE! 

Time well spent... 

The Advantages of Row Row 

and Rack-Oriented 

Cooling Architectures res 

for Data Centers 



“ The Advantages of 

Row and Rack-Oriented 

Cooling Architectures 

for Data Centers” 


$ 85 00 FREE! 

Deploying ng High-Density 

High-Density 

Zones in n a a Low-Density 

Low-Density 

Data Center nter 

White White Paper Paper #130 #130 

“ Deploying High-Density 

Zones in a Low-Density 

Data Center” 


$ 00 

115 FREE! 

Power and Cooling 

Capacity Management 

for Data Centers 


©2010 Schneider Electric Industries SAS, All Rights Reserved. Schneider Electric and APC are owned by Schneider Electric, or its affiliated companies in the United States and other countries. 

 

“ Power and Cooling 

Capacity Management 

for Data Centers” 


$ 00 225 

Download FREE APC White Papers to avoid 

the most common mistakes in planning 

IT power and cooling. 

FREE! 

Have a plan for your data center. 

 

 

cases, turnover and budget cuts resulted in no plan at all. 

Get the answers you need and avoid headaches tomorrow. 

 

 

 

yourself money and headaches tomorrow. 

 

www.apc.com 

Choose and download any three white papers 

within the next 90 days for FREE! 

Visit www.apc.com/promo Key Code s145w x6165 

A 


aG 

F

A 


aG 

F 

CONTENTS 

ABOUT THIS ISSUE 

Growing interest in cloud computing also raises the 

question of cost/benefit. In this issue, we feature an 

article describing a federated, open-source cloud platform 

that lowers the entry barriers to systems and 

applications research communities, an economic model 

for determining whether to lease or buy storage, and an 

analysis of energy consumption and battery life in 

mobile devices offloading computation to the cloud. We 

also look at puzzle-based learning, ethical codes of conduct 

in the legal realm, and the linkage of software 

development and business strategy. 

COMPUTING PRACTICES 

20 Puzzle-Based Learning for 

Engineering and Computer 

Science 

Nickolas Falkner, Raja Sooriamurthi, 

and Zbigniew Michalewicz 

To attract, motivate, and retain students and 

increase their mathematical awareness and 

problem-solving skills, universities are 

introducing courses or seminars that explore 

puzzle-based learning. The authors introduce 

and define this learning approach, describe 

course variations, and highlight early student 

feedback. 

PERSPECTIVES 

29 Using Codes of Conduct 

to Resolve Legal Disputes 

Peter Aiken, Robert M. Stanley, Juanita 

Billings, and Luke Anderson 

In the absence of other published standards of 

care, it is reasonable for contractual parties to 

rely on an applicable, widely available code of 

conduct to guide expectations. 

COVER FEATURES 

35 Open Cirrus: A Global 

Cloud Computing Testbed 

Arutyun I. Avetisyan, Roy Campbell, Indranil 

Gupta, Michael T. Heath, Steven Y. Ko, 

Gregory R. Ganger, Michael A. Kozuch, 

David O’Hallaron, Marcel Kunze, Thomas T. 

Kwan, Kevin Lai, Martha Lyons, Dejan S. 

Milojicic, Hing Yan Lee, Yeng Chai Soh, 

http://computer.org/computer 

For more information on computing topics, visit the Computer Society Digital Library at www.computer.org/csdl. 

Ng Kwang Ming, Jing-Yuan Luke, 

and Han Namgoong 

Open Cirrus is a cloud computing testbed 

that, unlike existing alternatives, federates 

distributed data centers. It aims to spur 

innovation in systems and applications 

research and catalyze development of an 

open source service stack for the cloud. 

44 To Lease or Not to Lease 

from Storage Clouds 

Edward Walker, Walter Brisken, 

and Jonathan Romney 

Storage clouds are online services for leasing 

disk storage. A new modeling tool, 

formulated from empirical data spanning 

many years, lets organizations rationally 

evaluate the benefit of using storage clouds 

versus purchasing hard disk drives. 

51 Cloud Computing for Mobile 

Users: Can Offloading 

Computation Save Energy? 

Karthik Kumar and Yung-Hsiang Lu 

The cloud heralds a new era of computing 

where application services are provided 

through the Internet. Cloud computing can 

enhance the computing capability of mobile 

systems, but is it the ultimate solution for 

extending such systems’ battery lifetimes? 

RESEARCH FEATURE 

57 Linking Software 

Development and 

Business Strategy 

through Measurement 

Victor R. Basili, Mikael Lindvall, Myrna 

Regardie, Carolyn Seaman, Jens Heidrich, 

Jürgen Münch, Dieter Rombach, 

and Adam Trendowicz 

The GQM + Strategies approach extends the 

goal/question/metric paradigm for 

measuring the success or failure of goals and 

strategies, adding enterprise-wide support 

for determining action on the basis of 

measurement results. An organization can 

thus integrate its measurement program 

across all levels. 

A 


aG 

F

A 


aG 

F 

IEEE Computer Society: http://computer.org 

Computer: http://computer.org/computer 

computer@computer.org 

___________ 

IEEE Computer Society Publications Office: +1 714 821 8380 

Cover image © Lolaferari Dreamstime.com 

6 The Known World 

Utter Chaos 

David Alan Grier 

10 32 & 16 Years Ago 

Computer, April 1978 and 1994 

Neville Holmes 

NEWS 

13 Technology News 

Will HTML 5 Restandardize the Web? 

Steven J. Vaughan-Nichols 

16 News Briefs 

Linda Dailey Paulson 

MEMBERSHIP NEWS 

66 IEEE Computer Society 

Connection 

69 Call and Calendar 

COLUMNS 

76 AI Redux 

Probabilistic Analysis of an Ancient 

Undeciphered Script 

Rajesh P.N. Rao 

81 Software Technologies 

When TV Dies, Will It Go to the Cloud? 

Karin Breitman, Markus Endler, Rafael 

Pereira, and Marcello Azambuja 

Flagship Publication of the IEEE 

Computer Society 

April 2010, Volume 43, Number 4 

84 Invisible Computing 

Questioning Invisibility 

Leah Buechley 

87 Education 

Cutting Across the Disciplines 

Jim Vallino 

90 Security 

Integrating Legal and Policy Factors in 

Cyberpreparedness 

James Bret Michael, John F. Sarkesain, 

Thomas C. Wingfield, Georgios Dementis, 

and Gonçalo Nuno Baptista de Sousa 

96 The Profession 

Computing a Better World 

Kai A. Olsen 

DEPARTMENTS 

4 Elsewhere in the CS 

71 Computer Society Information 

72 Career Opportunities 

74 Advertiser/Product Index 

93 Bookshelf 

Reuse Rights and Reprint Permissions: Educational or personal use of this 

material is permitted without fee, provided such use: 1) is not made for profit; 

2) includes this notice and a full citation to the original work on the first page of 

the copy; and 3) does not imply IEEE endorsement of any third-party products 

or services. Authors and their companies are permitted to post their IEEE-copyrighted material on 

their own Web servers without permission, provided that the IEEE copyright notice and a full citation 

to the original work appear on the first screen of the posted copy. 

Permission to reprint/republish this material for commercial, advertising, or promotional purposes 

or for creating new collective works for resale or redistribution must be obtained from the IEEE by 

writing to the IEEE Intellectual Property Rights Office, 445 Hoes Lane, Piscataway, NJ 08854-4141 

or pubs-permissions@ieee.org. _______________ Copyright © 2010 IEEE. All rights reserved. 

Abstracting and Library Use: Abstracting is permitted with credit to the source. Libraries are permitted 

to photocopy for private use of patrons, provided the per-copy fee indicated in the code at 

the bottom of the first page is paid through the Copyright Clearance Center, 222 Rosewood Drive, 

Danvers, MA 01923. 

IEEE prohibits discrimination, harassment, and bullying. For more information, visit www.ieee.org/ 

______________________ 

web/aboutus/whatis/policies/p9-26.html. 

A 


aG 

F

A 


aG 

F 

4 

ELSEWHERE IN THE CS 

Computer Highlights Society Magazines 

The IEEE Computer Society offers a lineup 

of 13 peer-reviewed technical magazines 

that cover cutting-edge topics in computing 

including scientific applications, design and 

test, security, Internet computing, machine intelligence, 

digital graphics, and computer history. Select articles from 

recent issues of Computer Society magazines are highlighted 

below. 

The March/April issue of Software features an overview 

of tools and technologies for improved collaboration titled 

“Collaboration Tools for Global Software Engineering,” by 

Filippo Lanubile, Christof Ebert, Rafael Prikladnicki, and 

Aurora Vizcaíno. Effective tool support for collaboration 

is a strategic initiative for any company with distributed 

resources, no matter whether the strategy involves offshore 

development, outsourcing, or supplier networks. Software 

needs to be shared, and appropriate tool support is the only 

way to do this efficiently, consistently, and securely. 

No single solution has yet emerged to unify communications 

among the various and proliferating messaging 

systems available today. In “Understanding Unified Messaging,” 

in the January/February issue of IT Pro, Clinton M. 

Banner of Alcatel-Lucent surveys the issues and solutions— 

both proprietary and open—that are in development. 

“Integrating Visualization and Interaction Research to 

Improve Scientific Workflows,” by the University of Minnesota’s 

Daniel F. Keefe, identifies three goals—improving 

accuracy, linking multiple visualization strategies, and 

making data analysis more fluid—that serve as a guide 

for future interactive-visualization research targeted 

COMPUTER 

at improving visualization tools’ impact on scientific 

workflows. 

The article, which appears in the March/April issue of 

CG&A, also offers four examples that illustrate the potential 

and challenges of integrating visualization research and 

interaction research: 3D selection techniques in brain visualizations, 

interactive multiview scientific visualizations, 

fluid pen- and touch-based interfaces for visualization, and 

modeling human performance in interactive visualizationrelated 

tasks. 

The Cell Broadband Engine is a heterogeneous chip 

multiprocessor that combines a PowerPC processor core 

with eight single-instruction, multiple-data accelerator 

cores and delivers high performance on many computationally 

intensive codes. “Application Acceleration with the 

Cell Broadband Engine” by Guochun Shi, Volodymyr Kindratenko, 

Frederico Pratas, Pedro Trancoso, and Michael 

Gschwind appears in the January/February issue of CiSE. 

Demand-response systems provide detailed powerconsumption 

data to utilities and those angling to assist 

consumers in understanding and managing energy use. 

Such data reveals information about in-home activities that 

can be mined and combined with other readily available 

information to discover more about occupants’ activities. 

“Inferring Personal Information from Demand-Response 

Systems” by Mikhail Lisovich, Deirdre Mulligan, and Stephen 

Wicker appears in the January/February issue of S&P. 

The March/April issue of Intelligent Systems focuses 

on new paradigms in business and market intelligence. 

The emergence of user-generated Web 2.0 content offers 

Published by the IEEE Computer Society 0018-9162/10/$26.00 © 2010 IEEE 

A 


aG 

F

A 


aG 

F 

opportunities to listen to the voice of the market as articulated 

by a vast number of business constituents. An 

overview in this issue of IS includes five short articles by 

distinguished experts on today’s trends in business and 

market intelligence. Each article presents a unique, innovative 

research framework, computational methods, and 

selected results and examples. 

In the world of Web 2.0, Internet 2, and open systems, 

most learning is still done in traditional classrooms. As 

education costs continue to far outpace inflation, what is 

e-learning’s role? In the March/April issue of Internet Computing, 

Stephen Ruth of George Mason University addresses 

these questions in “Is E-Learning Really Working? The 

Trillion Dollar Question.” 

Spectacles is a hardware/software platform developed 

from off-the-shelf components and ready for market. It 

includes local computation and communication facilities, 

an integrated power supply, and modular system building 

blocks such as sensors, voice-to-text and text-to-speech 

components, localization and positioning units, and microdisplay 

units. Its see-through display components are 

integrated into eyeglass frames. “Wearable Displays—for 

Everyone!” by Alois Ferscha and Simon Vogl appears in the 

January-March issue of PvC. 

The January/February issue of Micro continues a 

seven-year tradition of featuring top picks from computer 

architecture conferences. Guest editor Trevor Mudge of 

the University of Michigan participated in a program committee 

of 31 computer architects from both industry and 

academia. The committee reviewed 91 submissions, selecting 

13 papers for abridgment in the magazine. Topics range 

from practical prefetching to application-domain-specific 

computing and nonvolatile memory. 

Current media-synchronization techniques range from 

the very theoretical to the very practical. In “Modeling 

Media Synchronization with Semiotic Agents,” researchers 

from the City University of Hong Kong and the University of 

Reading describe a modeling technique for mapping theoretical 

models to system implementations. The technique 

combines agent technology with semiotics to offer a sound 

theoretical framework for expressing and manipulating 

media-synchronization attributes in real-world applications. 

Read more in the January-March issue of Multimedia. 

reduced 70% 

reduced 70% 

As CMOS technology scales down to the nanometer 

range, variation control in semiconductor manufacturing 

becomes ever more challenging. Introducing D&T’s March/ 

April special issue in “Compact Variability Modeling in 

Scaled CMOS Design,” guest editors Yu Cao of Arizona State 

University and Frank Liu of IBM Austin Research Laboratory 

preview five articles that address these challenges. 

An article by Fred Brooks, renowned computer scientist 

and author of The Mythical Man-Month, opens the first issue 

of Annals in 2010. In “Stretch-ing Is Great Exercise—It Gets 

You in Shape to Win,” Brooks recounts the history of the 

IBM Stretch Project (1955-1961). Although the company lost 

$35 million on the project in 1960, Brooks describes how 

Stretch drove technologies that enabled the rapid development 

of IBM’s successful 7000-series computers and the 

architectural innovations of its System/360 product family. 

Editor: Bob Ward, Computer; bnward@computer.org 

____________ 

http://computer.org/cn/ELSEWHERE 

April Theme: 

AGILITY AND 

ARCHITECTURE 

APRIL 2010 

A 


aG 

F 

5

A 


aG 

F 

6 

THE KNOWN WORLD 

Utter Chaos 

As far as I can tell, they live 

lives of utter chaos. The 

three of them, all young 

physicians who seem 

no older than 16, seize 

control of the locker room every 

morning. They apparently finish a 

rather raucous game just as I arrive 

to begin my swim. 

They typically appear with an 

explosion of noise, a clash of locker 

doors, and a litter of sweaty athletic 

clothes that bear the seals of expensive 

and exclusive institutions of 

higher education. They talk as if no 

one else is listening, as if they were 

standing in the most private seminar 

room of the local hospital, albeit 

standing buck naked, dripping wet, 

and yelling for no good reason. 

Surprisingly, they talk not of 

women or sports or one of the few 

topics of male discourse but of medicine. 

They pay inordinate attention 

to every corner and fold of their 

own bodies and speculate on what 

signs of good or ill they find. If you 

believe their words, these young doctors 

are the most medicated human 

beings on the planet. They spend an 

unusual amount of time talking about 

the medicines that they prescribe 

for themselves in much the same 

way that a small child will discuss 

the color of a recently encountered 

COMPUTER 

David Alan Grier, George Washington University 

No matter what we may believe, scientific practice often begins with 

a season when we don’t know what we know. 

puppy. One can be grateful that these 

physicians know the effects of their 

own medicine but simultaneously 

worried that these pharmaceuticals 

will impair their judgment when they 

recommend some treatment. 

For the most part, these young 

men talk about their cases and the 

treatments that they recommend. 

Although most of the medical terms 

are a complete mystery to the rest of 

us, we can all tell that these are young 

doctors who don’t know yet how to 

be professional physicians. They’re 

a little too offended when someone 

questions their judgment. They give 

far too many details about their 

patients and make it far too easy for 

the rest of us to identify, or at least 

imagine, the individual in question. 

These young physicians, in spite 

of their enthusiasm, their elite educations, 

and their practical experience 

with drugs, are still learning how to 

organize and discipline their practice. 

They’re living a medicine of 

utter chaos. Their conversation cycles 

around the various bodily systems, 

maps every symptom to every possible 

cause imaginable, and identifies 

extraordinary lists of information that 

they would like to extract from their 

patients. 

Even with this chaos, they’re still 

in love with their profession and still 

excited to encounter new problems. 

Yet they don’t realize that they’re 

only now learning to discipline their 

professional activity, to build a framework 

that will guide their work. Well 

before they reach the age of 30, they 

will learn to organize their professional 

knowledge into a structure that 

will guide the rest of their lives. 

UNINHIBITED SPECULATION 

We usually live so completely 

within our professional knowledge 

that we remember the effort required 

to discipline our thoughts only when 

we approach a topic that is new to us. 

New ideas can dislocate that mental 

structure, that vague semantic network, 

which gives us a foundation in 

the field. 

Over the past three months, I have 

had many opportunities to demonstrate 

that a career in computer 

science can still leave you grossly 

ignorant of many subjects and completely 

disoriented as you try to learn 

the basics of a new topic. The search 

tools of modern digital libraries give 

you little guidance in this work. You 

can retrieve hundreds of articles that 

seem relevant, spend hours reading 

material about some peripheral bit 

of research, and never find that key 

document describing the basic concepts 

of the field. 


A 


aG 

F

A 


aG 

F 

After years of doing this kind of 

research, I’ve learned that all new 

fields of scientific endeavor will go 

through periods of utter chaos before 

they settle into a regular discipline 

with well-defined terms, stable categories, 

and general methods. The task 

of any researcher is to work through 

the early literature and find that key 

moment when ideas begin to solidify. 

THE CHAOS OF 

RÖNTGEN’S XfiRAYS 

The early days of physics produced 

many fields that began in chaos and 

moved quickly into order. Few present 

a more dramatic story than the discovery 

of X-rays by Wilhelm Röntgen 

in late 1895. To both scientists and the 

public, X-rays seemed to be an entirely 

new phenomenon that didn’t exactly 

fit into the theories of electricity and 

magnetism that had been maturing 

over the prior 50 years. 

Röntgen discovered X-rays in 

early November and was mystified 

by them. Being a careful scientist, he 

attempted to identify the basic nature 

of this radiation. He assembled his 

notes from this work into an article 

that is little more than a list of observations. 

“A Discharge from a large 

induction coil is passed through a 

Hittorf’s vacuum tube,” he begins. 

This apparatus will cause a piece of 

paper covered with barium platinocyanide 

to glow. “The fluorescence is 

still visible at two meters distance,” 

he wrote. “It is easy to show that the 

origin of the fluorescence lies within 

the vacuum tube.” 

In the last paragraphs of the article, 

Röntgen admitted that he didn’t have 

a theory that explained the properties 

of X-rays. He rejected as “unlikely” the 

idea that they were a form of electromagnetic 

radiation and hence related 

to visible light. Instead, he speculated 

that his discovery was an entirely new 

form of energy. “Should not the new 

rays,” he asked, “be ascribed to longitudinal 

waves in the ether?” 

Had X-rays been less novel, had 

they not been able to a pass through 

soft human tissue and reveal the 

hidden bones, then Röntgen’s speculations 

about the luminiferous ether 

and longitudinal waves would have 

percolated through the community 

of physicists and would have been 

tested in a relatively systematic way. 

However, X-rays were indeed novel, 

and they could reveal the inner parts 

of the human body, so Röntgen’s 

ideas quickly moved into the laboratory 

and living room. 

All new fields of 

scientific endeavor 

will go through 

periods of utter 

chaos before they 

settle into a regular 

discipline with welldefined 

terms, stable 

categories, and 

general methods. 

THE EXCITEMENT 

OF THE POPULAR PRESS 

Barely four weeks after the original 

article appeared, news of this 

strange new ray reached all the 

capitals of Europe. “The invention of 

neither the telephone nor the phonograph 

has stirred up such scientific 

excitement as this Röntgen discovery 

in photography,” reported the London 

correspondent for The New York 

Times. “The papers everywhere are 

full of reports of experiments, and of 

reproductions of more or less ghastly 

anatomical pictures.” 

In general, the Times and other 

newspapers offered accurate descriptions 

of the experiments and the 

phenomena that had been observed. 

Yet, the papers weren’t scholarly 

journals and couldn’t distinguish 

well-conceived theory from rampant 

assumption. They speculated about 

what Röntgen meant when he mentioned 

“longitudinal waves in the 

ether” and listened to everyone who 

claimed credit for the discovery. “A 

French savant is said to have been 

securing similar results for several 

years by the use of an ordinary kerosene 

lamp,” offered the Times, which 

also claimed that it had found “that 

this discovery was not only made by 

a Prague professor in 1885” and that 

“a full report of the achievement was 

made to the Austrian Academy of Sciences 

in 1885.” 

ORGANIZING IGNORANCE 

Surprisingly, the popular press 

wasn’t alone in publishing speculation. 

The discovery of X-rays 

created a massive wave of papers 

that crashed on the scholarly community 

that winter. More than 125 

appeared before the first anniversary 

of the discovery. Some were good. 

Some were poorly conceived and ill 

written. Some identified the major 

applications for the waves. Some 

wandered into strange and unsupported 

speculations. 

The articles appeared so quickly 

that few could have been reviewed 

only by an editor and not an outside 

referee. Researchers would read a 

description of X-rays in a newspaper, 

conduct an experiment, and see their 

results published in fewer than six 

weeks. Many of these articles were 

part of a growing chain of work, in 

which one article spawns new experiments, 

which in turn spawn new 

work. Unfortunately, most of these 

researchers were engaged in speculation 

that was no better informed 

than the newspapers. On 14 February 

1896, Columbia physicist Michael 

Pupin published a simple framework 

for studying these waves, but his 

ideas had limited influence in those 

first months. 

Less than two weeks after Pupin’s 

paper appeared, a pair of researchers 

built a simple X-ray apparatus and 

started applying it to everything that 

they could find. A colleague asked 

them to “undertake the location of a 

bullet in the head of a child that had 

been accidentally shot.” They agreed 

APRIL 2010 

A 


aG 

F 

7

A 


aG 

F 

8 

THE KNOWN WORLD 

to do it but had enough doubts about 

the process to test the safety of their 

machine. “Accordingly, Dr. Dudley, 

with his characteristic devotion to 

the cause of science,” explained the 

report, “lent himself to the experiment.” 

They taped a coin to one side 

of his head, placed a photographic 

plate on the other, and exposed him 

to the X-ray machine for an hour. 

The experiment produced no 

useful outcome, but it may have 

spared the child. “The plate developed 

nothing,” they reported. But 

three weeks after the experiment, 

Dudley lost all of his hair in the area 

where the X-rays had been directed. 

“The spot is perfectly bald, being two 

inches in diameter,” they noted. “We, 

and especially Dr. Dudley, shall watch 

with interest the ultimate effect.” 

THE ELECTRICAL ENGINEERS 

START ORGANIZING 

Twelve full months passed before 

the scientific community began to 

develop a coherent approach to the 

study of X-rays. The key article is a 

report from the American Institute 

of Electrical Engineers (AIEE) that 

appeared in December 1896. “We 

were very much surprised, something 

like a year ago, by this very great discovery,” 

the report begins, “but we 

cannot say that we know very much 

more about it now than we did then. 

The whole world seems to have been 

working on it for all this time without 

having discovered a great deal with 

respect to it.” 

This report also reveals much 

about the nature of scientific practice 

at the time. It was a record of a discussion 

among a half-dozen senior 

scientists, all of whom lived within 

easy travel distance to New York and 

could easily congregate there for a 

meeting. Most important, they represented 

most of the major research 

universities and, among them, knew 

most of the scientists who could 

contribute to the field. When they discussed 

and debated ideas, they were 

speaking for an entire community. 

COMPUTER 

The report is strong and detailed 

but a little cautious as none of the participants 

wished to offend the others 

or give a conclusion that might be 

disproved later. The majority of the 

group concluded that X-rays were a 

form of electro-magnetic radiation 

and that they “most likely consist of 

vibrations of such small wavelengths 

as to be comparable with the distances 

between molecules of the most 

dense substances.” That hypothesis 

would be the basis for the serious 

study of X-rays. 

The flood of papers 

has continued to rise, 

as scientists from 

laboratories near and 

far try to make a name 

for themselves. 

THE POWER OF PUBLICATION 

Scientific periodicals have always 

had a central role in the process that 

turns chaos into order, although that 

role has often changed. At the time 

of Röntgen’s discovery, the editors 

of the oldest and most prestigious 

scientific journal in the English language, 

the Philosophical Transactions, 

stated that they selected articles for 

publication because of “the importance 

and singularity of the subjects, 

or the advantageous manner of 

treating them.” At the same time, 

they rejected the idea that they 

were validating or proving the science 

of any article that appeared in 

the Transactions. They didn’t pretend 

“to answer for the certainty of the 

facts, or propriety of the reasonings 

contained in the several papers so 

published,” they explained, “which 

must still rest on the credit or judgment 

of their respective authors.” 

The editors could take such a stand 

because the scientific community 

was still relatively small. Readers had 

other information to help them assess 

the articles. Moreover, they actually 

knew most of the people and laboratories 

engaged in active research and 

contributing to the field. 

The role of periodicals changed 

substantially over the next 50 years. 

By the middle of the 20th century, 

scientific periodicals were acting as 

validators—institutions that verified 

scientific research. “Science is 

moving rapidly,” noted the editor of 

the journal Science. He argued that scientific 

periodicals “in every discipline 

should take their responsibilities with 

respect to critical reviews” of every 

paper submitted to them. As an editor 

of a leading journal, he felt that Science 

needed to set the standard for 

scientific publication and regularly 

published articles that described 

the process of peer review, the rules 

that should guide reviewers, and the 

kinds of judgments that editors should 

make. 

At the same time it was promoting 

peer review, Science’s editor acknowledged 

that the expansion of scientific 

research was making it difficult for 

editors to validate every article that 

reached their desks. “With the great 

flood of manuscripts that today’s 

editors receive, in every scientific 

discipline, it is not possible to spend 

so much time and effort debating 

details,” he wrote. And as the century 

progressed, the flood of manuscripts 

only increased. 

THE ORIGINAL COLD FUSION 

As the waters rose, so rose the calls 

for scientific periodicals to test and 

validate the papers that they published. 

These calls reached a high 

point in 1989, when a pair of researchers 

announced to the public that they 

had made a discovery every bit as dramatic 

as the story of X-rays a century 

before. Two chemists “started a scientific 

uproar last month by asserting 

that they had achieved nuclear fusion 

at room temperature,” reported The 

New York Times. The writer added, 

“Many scientists remain skeptical.” 

The story of cold fusion lived 

for two years as laboratories 

A 


aG 

F

A 


aG 

F 

tried to make sense of the report. As the 

narrative unfolded, it quickly turned into a 

debate over the process of validating scientific 

ideas. Many claimed that the scientists who had 

performed the original experiments had taken 

their ideas too quickly to the public press. Their 

supporters argued that traditional laboratories 

were too skeptical of new ideas and that the 

scientific journals wouldn’t protect the patent 

rights of the discoverers. As the excitement of 

the moment faded and the realities of the story 

became clear, most commentators concluded 

that the public had not been well served. “Some 

scholars lunge too hastily for glory,” editorialized 

Times science writer Nicholas Wade. He 

told scientists to never announce their claims 

“until your manuscript has at least been 

accepted for publication in a reputable journal.” 

Nothing has become easier since the 

cold fusion controversy or those 

first exciting days of X-rays. Science 

has expanded. The number of 

subdisciplines has increased. The 

flood of papers has continued to rise, as scientists 

from laboratories near and far try to 

make a name for themselves. In that flood are 

ideas good and bad, breathtakingly original 

and utterly derivative. They’re the gift of painstaking 

research, the theft by 15 minutes of edits 

with a word processor, an instantaneous forgery 

by a clever bit of code that mocks the way 

that scientists write. 

By themselves, scientific journals aren’t the 

only institutions that attempt to organize the 

utter chaos of scientific research. They’re part 

of an infrastructure that includes laboratories, 

funding agencies, professional societies, review 

boards, and other institutions. These organizations 

will have to continue to grow and evolve as 

editors and publishers grapple with the forces 

acting on the scientific community, just as my 

young physician friends will have to mature 

into wise doctors who know how to draw disciplined 

judgment from the anarchy of energetic 

speculation. 

David Alan Grier is an associate professor of 

international science and technology policy at 

George Washington University and is the author 

of When Computers Were Human (Princeton, 

2005) and Too Soon to Tell (Wiley, 2009). Contact 

him at grier@computer.org. 

_____________ 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

UMassAmherst 

ISENBERG 

SCHOOL OF MANAGEMENT 

 

_____________________ 

APRIL 2010 

A 


aG 

F 

9

A 


aG 

F 

32 & 16 YEARS AGO 

APRIL 1978 

INTERNATIONAL DATA (p. 6) “AFIPS has named a panel 

of US experts to study foreign countries’ activities which 

restrict the international exchange of information. The 

panel is also expected to respond to the State Department’s 

position paper, Protecting Privacy in International 

Data Processing.” 

SOFTWARE QUALITY (p. 10) “For computer software 

systems, quality appears to be a characteristic that can 

be neither built in with assurance at system creation 

time nor retrofitted with certainty after a product is 

in use. Apart from the issues of logistics (copying and 

distributing software) there appears to be a continuing 

need to demonstrate some minimum level of quality in 

typical wide-use software systems. At the same time 

this appears to be impossible to accomplish: the best 

designed systems have had errors revealed many years 

after introduction, and several programs that had been 

publicly ‘proved correct’ contained outright mistakes!” 

SOFTWARE REVALIDATION (p. 14) “The problems of revalidating 

software following maintenance have received 

little attention, yet typically 70 percent of the effort 

expended on software occurs as program maintenance. 

There are two reasons to perform maintenance on a 

software system: to correct a problem or to modify the 

capabilities of the system. Revalidation must verify the 

corrected or new capability, and also verify that no other 

capability of the system has been adversely affected 

by the modification. Revalidation is facilitated by good 

documentation and a system structure in which functions 

are localized to well-defined modules.” 

ERROR DETECTION (p. 25) “It is well known that one 

cannot find all the errors in a program simply by testing 

it for a set of input data. Nevertheless, program testing is 

the most commonly used technique for error detection 

in today’s software industry. Consequently, the problem 

of finding a program test method with increased errordetection 

capability has received considerable attention 

in the field of software research. 

“There appear to be two major approaches to this problem. 

One is to find better criteria for test-case selection. The 

other is to find a way to obtain additional information (i.e., 

information other than that provided by the output of the 

program) that can be used to detect errors.” 

PROGRAM TESTING (p. 41) “So, there is certainly no need 

to apologize for applying ad hoc strategies in program 

testing. A programmer who considers his problems well 

and skillfully applies appropriate techniques—regardless 

of where the techniques arise—will succeed.” 

SYMBOLIC 

PROGRAM 

EXECUTION 

(p. 51) “The 

notion of 

symbolically 

executing a program 

follows quite 

naturally from normal 

program execution. First assume 

that there is a given programming language (say, Algol- 

60) and the normal definition of program execution for 

that language. One can extend that definition to provide 

a symbolic execution in the same manner as one extends 

arithmetic over numbers to symbolic algebraic operations 

over symbols and numbers. The definition of the symbolic 

execution is such that trivial cases involving no symbols 

are equivalent to normal executions, and any information 

learned in a symbolic execution applies to the corresponding 

normal executions as well.” 

COMPONENT PROGRESS (p. 64) “It is clear, looking back, 

that progress in component performance, size, cost, and 

reliability has been reflected in corresponding improvements 

in systems. Looking forward then, an analysis of 

coming developments in logic and memory components 

appears to be one approach to forecasting future system 

characteristics. Prediction at the system level, however, 

is complicated by various options in machine organization, 

such as parallel processing and pipelining, which 

interact with component features to establish the performance 

limits of the system.” 

HEMISPHERICAL KEYBOARD (p. 99) “A new hemisphericalshaped 

typing keyboard, called the Writehander, permits 

typing all 128 characters of the ASCII code with 

one hand. 

“To use the keyboard, the typist places his four fingers 

on four press-switches and his thumb on one of eight 

press-switches. The four finger-switches operate as the 

lower four bits of the 7-bit ASCII code, selecting one of 

16 character groups which contain the desired character. 

Each group contains eight letters, numerals, or symbols. 

The thumb then presses one switch to select the desired 

character from the group of eight. The shape and switch 

locations have been designed so that the fingers naturally 

locate themselves on the switches, according to the 

designer, Sid Owen.” 

PDFs of the articles and departments of the April issues of 

Computer for 1978 and 1994 are available through the IEEE 

Computer Society’s website: www.computer.org/computer. 

10 COMPUTER 


A 


aG 

F

A 


aG 

F 

APRIL 1994 

ATM (p. 8) “This year is an important one for asynchronous 

transfer mode. The technology now has the backing 

of vendors. 

“Membership of the ATM Forum, a consortium formed 

in October 1991 to accelerate the development and implementation 

of ATM products and services, has risen to 465 

with new faces from the likes of Microsoft and Novell. 

According to Fred Sammartino, forum president, 1994 

will be the year many companies set up serious trial 

implementations with intent to later incorporate the 

technology.” 

MBONE (p. 30) “Short for Multicast Backbone, MBone is 

a virtual network that has been in existence since early 

1992. It was named by Steve Casner of the University of 

Southern California Information Sciences Institute and 

originated from an effort to multicast audio and video 

from meetings of the Internet Engineering Task Force. 

Today, hundreds of researchers use MBone to develop 

protocols and applications for group communication. 

Multicast provides one-to-many and many-to-many 

network delivery services for applications such as 

videoconferencing and audio where several hosts need 

to communicate simultaneously.” 

MOBILE COMPUTING (p. 38) “Recent advances in technology 

have provided portable computers with wireless 

interfaces that allow networked communication even 

while a user is mobile. Whereas today’s first-generation 

notebook computers and personal digital assistants 

(PDAs) are self-contained, networked mobile computers 

are part of a greater computing infrastructure. Mobile 

computing—the use of a portable computer capable of 

wireless networking—will very likely revolutionize the 

way we use computers.” 

TELEROBOTICS (p. 49) “The significant communication 

latency between an earth-based local site and an 

on-orbit remote site drove much of the Steler [Supervisory 

Telerobotics Laboratory (at the Jet Propulsion 

Laboratory)] design. Such a large latency precludes 

direct real-time control of the robot in what nonetheless 

remains a real-time operation. To meet these 

real-time control requirements, our group devised 

a control scheme similar to the one used to control 

spacecraft.… Instead of transmitting programs to the 

remote site, we transmit command blocks (a set of data 

parameters) that control the execution of the remote 

site software that provides task-level control. This 

allows the remote site to handle a variety of control 

modes without requiring changes in the remote site 

software.” 

DIGITAL SOUPS (p. 65) “Last August, Apple Computer 

released a much-talked-about new computer called the 

Newton MessagePad. It’s the first platform in a series of 

Newton products that fall under the term personal digital 

assistant (Sharp’s ExpertPad is another). Although Apple 

doesn’t describe ‘the Newton’ as a computer, it’s actually 

an advanced system that includes handwriting recognition, 

full memory management and protection, and 

preemptive multitasking. But one feature that the MessagePad 

doesn’t have is files. The Newton stores similar 

data objects in formats called ‘soups.’ These soups are 

available to all applications.” 

SOFTWARE STANDARDS (p. 68) “The International 

Organization for Standardization is developing a suite 

of standards on software process under the rubric of 

Spice, an abbreviation for Software Process Improvement 

and Capability Determination. Spice was inspired 

by numerous efforts on software process around the 

world, including the Software Engineering Institute’s 

work with the Capability Maturity Model (CMM), Bell 

Canada’s Trillium, and ESPRIT’s Bootstrap.” 

VIDEOCONFERENCING (p. 96) “For the past two years, 

United Technologies Corp. (UTC) has been experimenting 

with desktop video communications in its Topdesc (Total 

Personal Desktop Communications) program, which has 

the goals of completeness and convenience: allowing 

separated personnel to communicate as completely and 

easily as if they were at the same location, and doing 

so whenever and from wherever it is convenient. This 

means that parties should be able to see objects and 

documents and collaborate easily using computer data. 

In addition, the communications unit must be sized and 

priced to fit into an office or lab.” 

THE SOFTWARE CRISIS (p. 104) “There’s been so much 

talk about a ‘software crisis’ over the past decade or two 

that you’d think software practitioners were the original 

Mr. Bumble, barely able to program their way out of a 

simple application problem. But when I look around, I see 

a world in which computers and the software that drives 

them are dependable and indispensable. They make my 

plane reservations, control my banking transactions, and 

send people into space. They even wage war—in entirely 

new and apparently successful ways.” 

Editor: Neville Holmes; neville.holmes@utas.edu.au 

_________________ 

APRIL 2010 11 

A 


aG 

F

A 


aG 

F 

______________________ 

_______________________ 

_______________________ 

A 


aG 

F

A 


aG 

F 

TECHNOLOGY NEWS 

Will HTML 5 

Restandardize 

the Web? 

Steven J. Vaughan-Nichols 

The World Wide Web Consortium is developing HTML 5 as a standard 

that provides Web users and developers with enhanced functionality 

without using the proprietary technologies that have 

become popular in recent years. 

In theory, the Web is a resource 

that is widely and uniformly 

usable across platforms. As 

such, many of the Web’s 

key technologies and architectural 

elements are open and 

platform-independent. 

However, some vendors have 

developed their own technologies 

that provide more functionality than 

Web standards—such as the ability to 

build rich Internet applications. 

Adobe System’s Flash, Apple’s 

QuickTime, and Microsoft’s Silverlight 

are examples of such proprietary 

formats. 

In addition, Google’s Gears and 

Oracle’s JavaFX—which the company 

acquired along with Sun Microsystems—have 

technologies that enable 

creation of offline and client-side Web 

applications. 

Although these approaches provide 

additional capabilities, they have 

also reduced the Web’s openness and 

platform independence, and tend to 

lock in users to specific technologies 

and vendors. 

In response, the World Wide Web 

Consortium (W3C) is developing 

HTML 5 as a single standard that provides 

Web users and developers with 

0018-9162/10/$26.00 © 2010 IEEE 

enhanced functionality without using 

proprietary technologies. 

Indeed, pointed out Google 

researcher Ian Hickson, one of the 

W3C’s HTML 5 editors, “One of our 

goals is to move the Web away from 

proprietary technologies.” 

The as-yet-unapproved standard 

takes HTML from simply describing 

the basics of a text-based Web to creating 

and presenting animations, audio, 

mathematical equations, typefaces, 

and video, as well as providing offline 

functionality. It also enables geolocation, 

a rich text-editing model, and 

local storage in client-side databases. 

The Web isn’t just about reading 

the text on the page and clicking 

on the links anymore, noted Bruce 

Lawson, standards evangelist at 

browser developer Opera Software. 

Added W3C director Tim Berners- 

Lee, “HTML 5 is still a markup language 

for webpages, but the really big 

shift that’s happening here—and, you 

could argue, what’s actually driving 

the fancy features—is the shift to the 

Web [supporting applications].” 

“HTML 5 tries to bring HTML into 

the world of application development,” 

explained Microsoft senior 

principal architect Vlad Vinogradsky. 

Published by the IEEE Computer Society 

“Microsoft is investing heavily in 

the W3C HTML 5 effort, working with 

our competitors and the Web community 

at large. We want to implement 

ratified, thoroughly tested, and stable 

standards that can help Web interoperability,” 

said Paul Cotton, cochair of 

the W3C HTML Working Group and 

Microsoft’s group manager for Web 

services standards and partners in 

the company’s Interoperability Strategy 

Team. 

At the same time though, Web 

companies say their proprietary technologies 

are already up and running, 

unlike HTML 5. 

Adobe vice president of developer 

tools Dave Story said, “The HTML 5 

timeline states that it will be at least a 

decade before the evolving efforts are 

finalized, and it remains to be seen 

what parts will be implemented consistently 

across all browsers.” 

In fact, while HTML 5 recently 

became a working draft, it’s not 

expected to become even a W3C candidate 

recommendation until 2012 or 

a final W3C standard until 2022. 

Nonetheless, some browser 

designers, Web authors, and websites— 

such as YouTube—are already adopting 

HTML 5 elements. For more 

APRIL 2010 

A 


aG 

F 

13

A 


aG 

F 

14 

TECHNOLOGY NEWS 

PLAYING WITH HTML 5 

Several websites očer a taste of what HTML 5 will bring. Some of the applications work 

with only certain browsers. 

YouTube’s beta HTML 5 video project (www.youtube.com/html5) works with the standard’s 

video tag. Browsers must support the video tag and have a player that uses the 

H.264 codec. Otherwise, YouTube will use Flash to play video. 

Mozilla Labs’ BeSpin (https://bespin.mozillalabs.com) is an experimental programmer’s 

editor that uses a variety of HTML 5 elements. 

FreeCiv.net (www.freeciv.net) is an online game by the FreeCiv.net open source project 

that supports HTML 5’s Canvas element. HTML 5-compatible browsers display map 

changes faster than those that aren’t compatible. 

Google Wave (https://wave.google.com/wave), a cross between social networking 

and groupware, uses several HTML 5 elements. 

Merge Web Design’s HTML 5 Geolocation (http://merged.ca/iphone/html5geolocation) 

is, as the name indicates, a demo of HTML 5-based geolocation. 

Sticky Notes (http://webkit.org/demos/sticky-notes/index.html) is the WebKit Open 

Source Project’s demo of HTML 5’s client-side database storage API. WebKit is an open 

source Web browser engine now used by, for example, Apple’s Safari browser. 

information, see the “Playing with 

HTML 5” sidebar. 

BACKGROUNDER 

HTML is the predominant markup 

language for webpages. It uses tags 

to create structured documents via 

semantics for text—such as headings, 

paragraphs, and lists—as well 

as for links and other elements. HTML 

also lets authors embed images and 

objects in pages and can create interactive 

forms. 

HTML, which stemmed from the 

mid-1980s Standardized General 

Markup Language, first appeared as 

about a dozen tags in 1991. 

The Internet Engineering Task 

Force began the first organized effort 

to standardize HTML in 1995 with 

HTML 2.0. 

The IETF’s efforts to maintain 

HTML stalled, so the W3C took over 

standardization. 

In 1996, the W3C released HTML 

3.2, which removed the various proprietary 

elements introduced over 

time by Microsoft and Netscape. 

HTML 4.0, still not a final standard, 

followed in 1998. The approach 

provides mechanisms for style 

sheets, scripting, embedded objects, 

richer tables, enhanced forms, and 

improved accessibility for people 

with disabilities. 

COMPUTER 

However, HTML was still primarily 

focused on delivering text, not multimedia 

or client-based applications. 

Because of this, proprietary technologies 

such as Apple’s QuickTime 

and Microsoft’s multimedia players, 

both first released in 1991; and Adobe 

Flash, which debuted in 1996, have 

been used for video. 

Technologies such as Google 

Gears and Oracle’s JavaFX, both 

first released in 2007, make creating 

Web-based desktop-style applications 

easier for developers. 

HTML 5 

W3C is designing HTML 5 to create 

a standard with a feature set that 

handles all the jobs that the proprietary 

technologies currently perform, 

said specification editor Michael 

Smith, the consortium’s special-missions-subsection 

junior interim floor 

manager. 

In addition, HTML 5 will support 

newer mobile technologies such as 

geolocation and location-based services 

(LBS), as well as newer open 

formats such as scalable vector 

graphics. SVG, an open XML-based 

file format, produces compact and 

high-quality graphics. 

Developers would thus be able 

to develop rich webpages and Webbased 

applications without needing to 

master or license multiple proprietary 

technologies. 

And browsers would be able to do 

more without plug-ins. 

Canvas. One of HTML 5’s key 

new features is Canvas, which lets 

developers create and incorporate 

graphics, video, and animations, usually 

via JavaScript, on webpages. 

HTML Canvas 2D Context is an 

Apple-originated technology for rendering 

2D graphics and animations on 

the client rather than on Web servers. 

By rendering graphics locally, the 

bottlenecks of server and bandwidth 

restrictions are avoided. This makes 

graphics-heavy pages render faster. 

Video tags. HTML 5’s codecneutral 

video tags provide a way to 

include nonproprietary video formats, 

such as Ogg Theora and H.264, 

in a page. 

The tag and underlying code 

tell the browser that the associated 

information is to be handled as an 

HTML 5-compatible video stream. 

They would also let users view 

video embedded on a webpage without 

a specific video player. 

Location-based services. A location 

API offers support for mobile 

browsers and LBS applications by 

enabling interaction with, for example, 

GPS technology and data. 

Working offline. AppCache lets 

online applications store data and 

programming code locally so that 

Web-based programs can work as 

desktop applications, even without 

an Internet connection. 

HTML 5 has several other features 

that address building Web applications 

that work offline. These include 

support for a client-side SQL database 

and for offline application and data 

caching. 

Web applications thus can have 

their code, graphics, and data stored 

locally. 

Web Workers. The Web Workers 

element runs scripts in the background 

that can’t be interrupted by 

other scripts or user interactions. 

This speeds up background tasks. 

A 


aG 

F

A 


aG 

F 

Syntax and semantics. HTML 5 

makes some changes to the syntax 

and the semantics of the language’s 

elements and attributes. 

For example, as Figure 1 shows, 

HTML can be written in two syntaxes: 

HTML and XML. 

Using XML will enable more complex 

webpages that will run faster on 

Web browsers. 

XML requires a stricter, more accurate 

grammar than HTML and thus 

necessitates less work by the local 

computer to run quickly and correctly. 

However, XML pages require more 

work by the developer to achieve the 

higher accuracy level. 

HTML+RDFa. The W3C recently 

began dividing HTML 5 into subsections 

for easier development. 

For example, HTML+RDFa 

(Resource Description Framework in 

attributes) provides a way to embed 

resource description frames in 

webpages. 

RDF is a method for conceptually 

describing or modeling information 

implemented in Web resources. 

This would let developers incorporate 

machine-readable data into 

webpages, which would enable faster 

page rendering. 

TRIALS AND TRIBULATIONS 

Both nailing down details and the 

standardization process itself have 

been problematic for HTML 5. 

For example, the open source 

community supports making the 

open Ogg Theora format the default 

video codec for HTML 5’s video tag. 

Adobe, Google, and Microsoft 

opposed this, expressing videoquality, 

patent-related, and other 

concerns. 

According to Google’s Hickson, 

“There is no suitable codec that all 

vendors are willing to implement 

and ship.” 

The W3C thus decided not to select 

one as a default. 

What’s really slowing HTML 5 

is that the standardization effort 

“combines the worst elements of the 

Figure 1. HTML 5 lets HTML be written in two syntaxes: HTML and XML. XML’s stricter, 

more accurate grammar requires more work by the developer but less work by the 

local computer to run quickly and correctly. The browser ćrst parses out the various 

terms in HTML and XML and then sends them serially in an e cient and logical manner 

for rendering. 

IETF process and the W3C process,” 

said Adobe principal scientist Larry 

Masinter, a member of the W3C’s 

HTML Working Group. 

“[As is the case with] the IETF [process], 

there is the chaos of an open 

mailing list, wide-ranging comments, 

and free participation but without 

the ‘adult supervision’ that the IETF 

supplies in the form of the Internet 

Engineering Steering Group and area 

directors,” he explained. 

From the W3C, he added, there’s 

too much bureaucracy and the voting 

members don’t have enough responsibility 

and accountability for the 

final results. 

Another potential implementation 

concern is that some companies are 

already extending HTML 5 in ways 

that the final standard is unlikely to 

support. 

HTML 5 will blur the line between 

desktop and online applications and 

thus create an opportunity for malware 

writers, according to Dmitri 

Alperovitch, vice president of threat 

research with security vendor McAfee. 

By letting Web applications run on 

local systems, HTML 5 would allow 

Web-based malware to do the same. 

Timely user acceptance may 

be an issue for HTML 5. 

Trevor Lohrbeer, CEO 

of Lab Escape, a data-visualization-tool 

vendor, said 

sweeping Internet technologies must 

be supported by at least 80 percent 

of browsers before being considered 

reliable enough for deployment. 

He estimated HTML 5 won’t 

achieve this level of support until 

2013 at the earliest. 

However, Jay Baker, director of 

architecture at Viewlocity, a supplychain-management 

software vendor, 

said his company plans to adopt 

HTML 5 fairly aggressively. 

“Although there is a learning 

curve associated with HTML 5,” he 

explained, “many of the new features 

make today’s common hacks and 

workarounds obsolete.”fi 

Steven J. Vaughan-Nichols is a freelance 

technology writer based in Mills 

River, North Carolina. Contact him at 

sjvn@vna1.com. 

__________ 

Editor: Lee Garber, Computer; 

l.garber@computer.org 

_____________ 

APRIL 2010 

A 


aG 

F 

15

A 


aG 

F 

16 

NEWS BRIEFS 

New Technique Provides 

Energy Wirelessly 

Amethod for transferring 

energy wirelessly promises 

to streamline the 

device-charging process 

as well as provide power 

to an array of electronic devices such 

as cellular phones and MP3 players. 

WiTricity Corp.’s wireless energy 

system comes as a flat pad on which 

devices sit or as a domed pod that 

powers or charges devices placed 

nearby. Manufacturers could also 

place the WiTricity components into 

a computer or other electronic device 

that requires charging. 

The charging system is powered 

via an electrical outlet or battery, 

WiTricity Corp.’s energy system can provide power to or charge nearby devices 

wirelessly. The charging system turns incoming electric current into a magnetic field, 

which then induces current in and a magnetic field around a resonating coil. This 

induces a magnetic field in a coil in a nearby device, thereby transferring energy. 

Source: WiTricity Corp. 

COMPUTER 

noted David Schatz, the company’s 

director of business development and 

marketing. 

First, the charger converts incoming 

AC power to DC, then the radio 

amplifier creates and amplifies 

high-frequency signals, said Schatz. 

The high-frequency current flowing 

through the charger’s power source 

excites a tuned resonator coil that 

starts oscillating, which creates an 

oscillating magnetic field. 

The device being powered has a 

coil tuned to the same frequency as 

the charger. This induces the magnetic 

field to flow from the charger 

to the other device, thereby trans- 

ferring energy. 

So far, Schatz noted, the researchers 

have transferred electricity 

wirelessly over a distance of two 

meters. However, he said, the larger 

the resonator, the stronger the magnetic 

field and the farther the system 

could transfer power to devices. 

Because WiTricity’s system requires 

access to an energy source, it will not 

be a go-anywhere technology, Schatz 

noted. However, he added, it eventually 

may be able to work at outdoor 

locations with their own power 

sources, such as perhaps bus stops. 

Menno Treffers, chair of the 

Wireless Power Consortium and 

Philips Electronics’ senior director 

of standardization, said WiTricity’s 

approach appears to have the same 

limited-transmission-range issues 

as the technology developed by the 

consortium, to which WiTricity does 

not belong. 

When WiTricity demonstrated its 

product at January’s Computer Electronics 

Show, it powered a 32-inch 

TV from about half a meter away. 

WiTricity is also working on some 

specialized, classified military systems, 

according to Schatz. 

The company is already selling 

its components and systems 

directly to vendors and also plans to 

license the intellectual property to 

manufacturers. 

Now, Schatz said, research must 

look into how to integrate the technology 

within electronics devices. 

He predicted computer electronics 

companies will soon start working 

on incorporating WiTricity technology 

into products that will ship in 

2011. 


A 


aG 

F

A 


aG 

F 

Researchers Pioneer Innovative 

Antiphishing Approach 

Researchers have developed 

an innovative approach 

to identifying webpages 

that phishers use to trick 

victims into entering confidential 

information such as user 

names, passwords, and bank-account 

and credit-card numbers. 

Scientists at Taiwan’s Academia 

Sinica focused their efforts on 

recognizing the pages based on 

their appearance rather than their 

content, as current antiphishing 

approaches do. 

Phishers typically create webpages 

that look like those belonging 

to banks, e-commerce operations, 

or other businesses on which users 

might enter financial or accountaccess 

information. When a user 

enters such data on a fake page, the 

phisher captures the information and 

utilizes it to defraud the victim. 

Surveys by Gartner Inc., a market 

research firm, found that due to 

phishing attacks in the US alone, 3.6 

million people lost about $3.2 billion 

in 2007 and 5 million lost about $1.8 

billion in 2008. 

Traditional antiphishing approaches 

focus on user complaints and content 

analysis, explained Dave Cowings, 

Symantec Security Response’s senior 

manager for operations. 

Verified user complaints will 

put phishers’ pages on blacklists of 

websites. 

Some traditional approaches 

analyze the content of websites that 

appear similar to well-known sites. If 

the content on a suspect page is the 

same as on a genuine site but the URLs 

are different, the software assumes the 

suspect page originated with phishers. 

However, this approach can miss 

small differences between pages and 

thereby yield false positive findings. 

Also, said Academica Sinica 

researcher Kuan-Ta Chen, phishers 

commonly attempt to circumvent 

traditional approaches by designing 

webpages without actual text—such 

as by using images of text instead— 

thereby making content analysis 

useless. 

The Academia Sinica system analyzes 

the appearance of genuine and 

phishers’ versions of pages and looks 

for subtle differences. 

The system examines the appearance 

of websites that are common 

phishing targets, focusing on keypoints. 

According to Chen, keypoints 

are small, easily identifiable parts 

of a page, such as edges, corners, or 

areas with large color or contrast differences. 

Analyzing keypoints can 

identify tiny differences and also 

eliminates the need to examine entire 

pages, saving both time and computational 

overhead. 

In addition to copying pages 

imprecisely, Chen explained, “phishers 

[sometimes] add messages to their 

pages designed to convince victims 

that they must enter private information 

immediately and also include 

advertisements that, when clicked on, 

direct users to yet another malicious 

website.” 

The Academia Sinica system looks 

for such differences. 

The researchers have already 

examined 2,058 pages on 74 websites 

that phishers commonly replicate, 

including the top five targets: eBay, 

PayPal, Marshall & Ilsley Bank, Charter 

One Bank, and Bank of America. 

In tests, Chen said, his system has 

accurately recognized phishing sites 

between 95 and 98 percent of the 

time. 

The scientists have implemented 

their technology in their Phishgig 

website (http://mmnet.iis.sinica.edu. 

___________ 

tw/proj/phishgig), from which users 

can download a Firefox browser 

plugin capable of detecting phishing 

pages. Once Phishgig detects a fraudulent 

page, it displays a warning that 

users can heed or ignore. 

The team is developing plugins for 

other browsers, including Internet 

Explorer and Google Chrome. 

Marty Lindner—principal engineer 

at CERT, a security research center 

at Carnegie Mellon University’s Software 

Engineering Institute—said the 

approach will probably be effective 

only long enough for phishers to 

figure out how to circumvent it. 

The most important aspect of 

phishing security is careful behavior 

by potential victims, he explained, 

and many users are insufficiently 

cautious. 

Chen said he realizes that phishers 

commonly change their methods, 

so his team is monitoring activity 

to see if and how they alter their 

approaches in response to the new 

technique. 

Meanwhile, he noted, the technology 

works only with copies of 

the pages the system has already 

examined. 

Chen said the researchers are 

open to commercialization opportunities. 

News Briefs written by Linda Dailey 

Paulson, a freelance technology 

writer based in Portland, Oregon. 

Contact her at ldpaulson@yahoo.com. 

Editor: Lee Garber, Computer; 

l.garber@computer.org 

_____________ 

______________ 

APRIL 2010 

A 


aG 

F 

17

A 


aG 

F 

18 

NEWS BRIEFS 

MIT Media Lab researchers have built an 

interface project called the Living 

Wall. The project would let users touch the 

wall to activate lamps, music systems, and 

other items in a room. 

The Living Wall uses electronically 

enabled wallpaper covered with electronically 

conductive paints that create not only 

decorative patterns but also circuitry that 

helps enable device control. 

Earlier wall-based interfaces required 

energy-hungry projectors to throw images 

on the walls for decoration. 

The Living Wall project, led by MIT assistant 

professor Leah Buechley, takes a 

dičerent, inexpensive, and energy-ećcient 

approach. 

The researchers used a copper-based 

conductive paint to draw lines that function 

as circuits. This, in essence, applies a large, 

exible printed circuit board to a wall. 

They then magnetically attached electronic 

components—such as accelerometers, 

light sensors, touch sensors, temperature 

sensors, and Bluetooth wireless-communications 

modules for controlling smaller devices 

such as computers and cellular phones—to 

the wall’s circuit board. 

The wireless modules could be used to 

control larger appliances in the future, 

according to Buechley. 

Users press the touch sensors to control 

devices. The other sensors monitor envi- 

Anew technology promises 

to enable fast, wireless 

personal-area networks 

that could let users download 

videos to and access 

the Internet via mobile devices at 

high speeds. 

IEEE has approved standard 

802.15.3c-2009, which defines how 

to design interoperable WPAN radio 

equipment capable of providing ad 

hoc connectivity, quality of service, 

and reliability. 

The technology would offer theo- 

COMPUTER 

RESEARCHERS BUILD A WALL THAT ACTS LIKE A REMOTE CONTROL 

ronmental variables near the wall to make 

lighting, temperature, and other adjustments 

as necessary. 

The system runs at about 20 volts, she 

noted, and draws about 2 amps and 40 

watts when fully loaded with devices to 

retical maximum data rates of 5.8 

gigabits per second at distances 

up to 10 meters under ideal conditions, 

4 to 5 Gbps under typical 

conditions, and 1 to 2 Gbps under 

conditions degraded by, for example, 

obstacles in transmission paths 

or interference from other signals, 

noted Bob Heile, chair and CEO of 

the ZigBee Alliance and also chair 

of the IEEE’s 802.15 WPAN Working 

Group. 

Today’s fastest wireless systems 

are generally much slower, he said. 

control. By comparison, a laptop consumes 

about 50 watts and a microwave, about 

1,000 watts. 

Buechley said her team is discussing 

commercializing the technology with several 

companies, which she declined to name. 

MIT’s Living Wall project uses wallpaper and paint, both electronically enabled, 

that create not only decorative patterns but also electronic circuitry. This lets users 

touch the wall and control electronic items in the same room. 

Wireless Technology Promises 

Ultrafast Personal Area Networks 

For example, IEEE 802.11n, Wi-Fi’s 

speediest version, offers up to 600 

Mbps. 

IEEE 802.15.3c-2009 is fast largely 

because it works in the unlicensed 

57-to-64-GHz frequency range, which 

offers large amounts of available 

bandwidth. 

The standard’s completion is a 

key data-communications milestone 

because it is the first IEEE 802 series 

wireless standard to specify operation 

in the 60-GHz spectrum and 

offer data rates of at least 1 Gbps, 

A 


aG 

F

A 


aG 

F 

explained Paul Nikolich, IEEE 802 

LAN/MAN standards committee 

chair and chief strategy officer at YAS 

Broadband Ventures. 

This technology is fast enough to 

provide the consumer-electronics 

market with effective options for wirelessly 

connecting high-performance 

elements of home-entertainment 

systems, such as a high-definition TV 

and a set-top box, Heile noted. 

He said IEEE 802.15.3c-2009 could 

also provide connections in other set- 

tings, such as body-area networks for 

medical applications. 

In addition, the standard’s high 

data rates and ability to provide 

guaranteed time slots for designated 

transmissions support applications, 

such as high-end video, that require 

quality of service. 

Like other radio-based technologies, 

IEEE 802.15.3c-2009 is 

implemented via a chip that includes 

a transmitter, receiver, and antenna. 

Heile said the system can identify 

lost packets and get replacements to 

the destination fast enough to preserve 

packet order with minimal 

delay. Thus, he noted, it can coexist 

with interference from other technologies 

operating in the 60-GHz band. 

If successful, the new technology 

could provide an improved way to 

play video and other types of multimedia 

via wireless technologies, 

said Filomena Berardi, Connectivity 

Group research analyst with IMS 

Research. 

IEEE SOFTWARE 

offers pioneering ideas, 

expert analyses, and 

thoughtful insights for 

software professionals 

who need to keep up 

with rapid technology 

change. It’s the authority 

on translating software 

theory into practice. 

www.computer.org/ 

software/SUBSCRIBE 

___________________ 

 

APRIL 2010 

A 


aG 

F 

19

A 


aG 

F 

20 


Puzzle-Based Learning 

for Engineering and 

Computer Science 

Nickolas Falkner, University of Adelaide 

Raja Sooriamurthi, Carnegie Mellon University 

Zbigniew Michalewicz, University of Adelaide 

To attract, motivate, and retain students and increase their 

mathematical awareness and problem-solving skills, universities are 

introducing courses or seminars that explore puzzle-based learning. 

We introduce and define this learning approach with a sample syllabus 

and course material, describe course variations, and highlight early 

student feedback. 

Arecent article describes a puzzle-based freshman 

seminar introduced at the University of 

California, Santa Barbara, to motivate and 

retain computer engineering students. 1 The 

author argues that attracting students to computer 

science and engineering programs represents only 

one aspect of a broader problem, the shortage of a skilled 

information technology workforce, and that recruitment 

efforts must be augmented with additional strategies for 

retaining and motivating students—strategies that are 

missing in curricula recommendations of the IEEE Computer 

Society and the ACM. 

The problem may be even broader. Today’s marketplace 

needs more skilled graduates capable of solving real 

problems of innovation in a changing environment. Missing 

in the majority of engineering and computer science 

curricula is a focus on developing problem-solving skills. 

Further, many courses that introduce elements of problemsolving 

skills do so at the programs’ third or fourth level, 

after students have already faced the majority of their inacademy 

intellectual challenges. 

While some courses with a design content emphasis 

might meet this requirement, most engineering students 

never learn how to think about solving problems in general. 

Throughout their education, they are constrained 

to concentrate on textbook questions at the end of each 

COMPUTER 

chapter, solved using material discussed in the chapter. 

This constrained form of “problem solving” is not sufficient 

preparation for addressing real-world problems. On 

entering the real world, students find that problems do not 

come with instructions or a guidebook. One of our favorite 

examples for illustrating this point is a puzzle on breaking 

a chocolate bar: 

A rectangular chocolate bar consists of m × n small rectangles, 

and you wish to break it into its constituent parts. At 

each step, you can only pick up one piece and break it along 

any of its vertical or horizontal lines. How should you break 

the chocolate bar using the minimum number of steps? 

If you do not know the answer, which textbook would 

you search to discover the solution? The same applies to 

solving many real-world problems: Which textbook should 

you search to find a solution, if that is the solution strategy 

you’ve learned? 

Students often have difficulty applying independent 

thinking or problem-solving skills regardless of the nature 

of a problem. At the same time, educators are interested 

in teaching “thinking skills” rather than “teaching information 

and content.” The latter approach has dominated 

in the past. As Alex Fisher 2 wrote “though many teachers 

would claim to teach their students ‘how to think’, most 


A 


aG 

F

A 


aG 

F 

would say that they do this indirectly or implicitly in the 

course of teaching the content which belongs to their special 

subject. Increasingly, educators have come to doubt 

the effectiveness of teaching ‘thinking skills’ in this way, 

because most students simply do not pick up the thinking 

skills in question.” 

Further, many analysts lament students’ decreasing 

mathematical skills. A recent Mathematics Working 

Party Final Report, issued by the University of Adelaide, 3 

includes statements such as, “There is an urgent need to 

raise the profile and importance of mathematics among 

young people” and “The declining participation in mathematics 

and related subjects is not limited to Australia.” 

Our universities, the University of Adelaide and Carnegie 

Mellon University, have introduced a new puzzle-based 

learning course to address all the issues raised here. 

PUZZLEfiBASED LEARNING APPROACH 

The puzzle-based learning approach aims to encourage 

engineering and computer science students to think about 

how they frame and solve problems not encountered at 

the end of some textbook chapter. Our goal is to motivate 

students while increasing their mathematical awareness 

and problem-solving skills by discussing a variety of puzzles 

and their solution strategies. The course is based on 

the best traditions introduced by Gyorgy Polya and Martin 

Gardner over the past 60 years. 4,5 

Many teachers have used puzzles for teaching purposes, 

and the puzzle-based learning approach has a tradition that 

exceeds 60 years. 6 Historians found the first mathematical 

puzzles in Sumerian texts from circa 2500 BC. However, 

some of the best evidence for the puzzle-based learning 

approach can be found in the works of Alcuin, an English 

scholar born around 732 AD, whose main work, Problems 

to Sharpen the Young, included more than 50 puzzles. 

Some 1,200 years later, one of Alcuin’s puzzles—the “river 

crossing problem”—is still used in artificial intelligence 

textbooks to educate computer science students. 

In our course, we concentrate on educational puzzles 

that support problem-solving skills and creative thinking. 

These educational puzzles satisfy most of the following 

criteria. 

Independence 

The puzzles are not specifically tied to a single problem-solving 

domain. 

Generality 

Educational puzzles should explain some universal mathematical 

problem-solving principles. This is key. Most people 

agree that problem solving can only be learned by actually 

solving problems. This activity, however, must be supported 

by instructor-provided strategies. These general strategies 

allow for solving yet unknown problems in the future. 

Simplicity 

Educational puzzles should be easy to state and remember. 

This is important because easy-to-remember puzzles 

increase the chance students will remember the solution 

method, including the universal mathematical problemsolving 

principles. 

Eureka factor 

Educational puzzles should initially frustrate the problem 

solver, but hold out the promise of resolution. A puzzle 

should be interesting because its result is not immediately 

intuitive. Problem solvers often use intuition to start their 

quest for the solution, and this approach can lead them 

astray. Eventually they reach a “Eureka moment”—Martin 

Gardner’s Aha!—when students recognize the correct path 

to solving the puzzle. A sense of relief accompanies this 

moment, and the frustration felt during the process dissipates, 

giving the problem solvers a sense of reward at their 

cleverness for solving the puzzle. The Eureka factor also 

implies that educational puzzles should have elementary 

solutions that are not obvious. 

There is a strong connection 

between the ability to solve puzzles 

and the ability to solve industry and 

business problems. 

Entertainment factor 

Educational puzzles should be entertaining and engaging. 

Entertainment is often a side effect of simplicity, 

frustration, the Eureka factor, and an interesting setting 

such as playing in a casino environment, fighting against 

dragons, or dropping eggs from a tower. 

THE LURE OF EDUCATION 

Educational puzzles can play a major role in attracting 

students to computer science and engineering programs, 

and can be used in talks to high school students and 

during open-day events. Puzzles can also be a factor that 

helps retain and motivate students. Above all, they are 

responsible for developing critical thinking and problemsolving 

skills as well as raising the profile and importance 

of mathematics. Further, there is a strong connection 

between the ability to solve puzzles and the ability to 

solve industry and business problems. Many real-world 

problems can be perceived as large-scale puzzles. William 

Poundstone, when investigating the purpose of famous 

Microsoft/Silicon Valley interview puzzles, 7 wrote, 

“At Microsoft, and now at many other companies, it is believed 

that there are parallels between the reasoning used to solve 

puzzles and the thought processes involved in solving the real 

APRIL 2010 

A 


aG 

F 

21

A 


aG 

F 

22 


Real world 

Abstract/model 

world 

problems of innovation and a changing marketplace. […] When 

technology is changing beneath your feet daily, there is not much 

point in hiring for a specific, soon-to-be-obsolete set of skills. You 

have to try to hire for general problem-solving capacity, however 

difficult that may be. […] Both the solver of a puzzle and a 

technical innovator must be able to identify essential elements 

in a situation that is initially ill-defined. It is rarely clear what 

type of reasoning is required or what the precise limits of the 

problem are.” 

Puzzle-based versus problem-based versus 

project-based learning 

The ultimate goal of puzzle-based learning is to lay a 

foundation for students to be effective problem solvers in 

the real world. At the highest level, problem solving in the 

real world calls into play three categories of skills: dealing 

with the vagaries of uncertain and changing conditions; 

harnessing domain-specific knowledge and methods; and 

critical thinking and applying general problem-solving 

strategies. 

These three skill categories are captured in the three 

forms of learning Figure 1 depicts. In this continuum, each 

layer of skills builds upon the layers below it. Puzzle-based 

learning focuses on domain-independent, transferable 

skills. In addition, we aim to foster introspection and 

reflection on the personal problem-solving process: What 

was I thinking? What is the solution? Why did I not see it? 

Both problem-based and project-based learning are well 

established methodologies. 8,9 By our description, problembased 

learning requires significant domain knowledge. 

This is the form of learning typically emphasized in a 

domain-specific undergraduate course such as electromagnetism, 

data structures, or circuit theory. Project-based 

learning, on the other hand, deals with complex situations 

in which usually no clearly unique or correct way 

of proceeding exists. For example, “How can we increase 

the adherence of cystic fibrosis patients to following their 

treatment protocol?” Determining the best solution in such 

a situation can be difficult. 

The pedagogical objectives of project-based learning 

include dealing with ambiguity and complexity, integrating 

COMPUTER 

Working in teams 

identifying the question 

Acquisition of 

domain knowledge 

Critical thinking 

logical reasoning 

Dealing with uncertainty 

and changing conditions 

Reasoning with 

domain-specific methods 

Abstract reasoning 

domain independent 

Figure 1. Problem solving in the real world requires a continuum of 

learning and skills in which each layer of skills builds upon the layers 

below it. 

a variety of approaches, user testing of the proposed 

solutions’ value, and working with a team 

composed of diverse backgrounds and skills. In 

both problem-based and project-based learning, 

the problem drives the learning: Students must 

assess what they already know, the knowledge 

they need to address the problem, and how to 

bridge the knowledge and skill gap. 

Puzzle-based learning focuses on domainindependent 

critical thinking and abstract 

reasoning. This leads us to ask, “What is the difference 

between a puzzle and a problem?” One 

way to characterize the difference measures the 

extent to which domain-specific knowledge is needed to 

solve it. The general flavor of puzzles asserts that their 

solution should require only domain-neutral general reasoning 

skills—biologists, musicians, and artists should all 

be able to solve the same puzzle. The different styles of 

reasoning required for problem-based and puzzle-based 

learning could be compared to the difference between a 

field investigator and an armchair detective: one emphasizes 

pure reasoning more. 

Dropping eggs 

The well-known egg-drop experiment provides an 

example that compares and contrasts problem-based and 

puzzle-based learning. The traditional problem-based 

learning version of this experiment involves finding a way 

to drop an egg from a maximal height without breaking 

it. A puzzle-based learning version of this experiment also 

involves dropping an egg from a building, but the question 

under investigation, although related, is quite different. 

In the problem-based learning approach, students conduct 

a series of physical experiments to determine how to 

maximize the height from which an egg can be dropped 

without breaking. There are two broad approaches: 

Dampen the impact’s effect (leading to padding-based 

solutions) or lessening the impact (leading to delivery 

mechanisms such as a parachute). The team-based learning 

outcomes of such an experiment determine different 

ways to dampen or lessen an impact. 

A puzzle-based learning approach to a similar problem 

does not involve a physical experiment, but rather 

a thought experiment. One approach would be to ask a 

question along the lines of the following: Using multiple 

eggs, what would be an effective strategy for determining 

the highest floor from which I could drop an egg without 

breaking it? This question has interesting variations. This 

thought experiment has three entities: the number of eggs, 

number of drops, and number of floors. 

One puzzle-based learning question could be, “Given a 

fixed number of eggs and a number of allowed drops, what 

is the maximum height of a building whose breaking floor 

we can determine? This could be denoted as F e,d . 

A 


aG 

F

A 


aG 

F 

An alternate question could be, “Given a fixed number of 

floors—say, 100—and a number of eggs—say, three—what 

is the maximum number of drops needed to determine 

the breaking floor (D e,f )?” Yet another version might be to 

ask how many eggs would be needed to determine the 

breaking floor given a fixed number of floors and allowed 

number of drops (E f,d ). 

Note that all three puzzle-based learning versions of 

the problem require only basic math skills and analytical 

reasoning. One goal of puzzle-based learning is to foster 

the skill of analyzing and understanding problems clearly. 

Part of this requires the clarification of any assumptions 

needed to solve the problem. For example, for the egg-drop 

thought experiment, some reasonable assumptions include 

“all eggs behave the same way” and “an egg that survives 

a drop is not damaged and may be dropped again.” To 

constrain the problem, we would consider assumptions 

such as “an egg surviving a drop from floor x would have 

survived a drop from any floor less than x.” 

Suppose, given a fixed number of eggs, e, and a specified 

number of drops, d, we want to determine the maximal 

height of a building whose breaking floor we can determine 

(F e,d ). Applying the heuristic of “try to solve a similar 

but simpler problem,” let us consider the situation where 

we have only one egg (e = 1). In this case, we are required 

to search sequentially. 

If we are allowed 10 drops (d = 10), then we can determine 

the breaking floor of a ten-floor building by starting 

at floor one and working our way up. Now suppose we had 

two eggs (e = 2). What strategy could we follow? Students 

who have had some prior programming experience often 

give binary search as a possible strategy, although this is 

not the best solution. 

Students are led through the reasoning process in a 

lecture environment and encouraged to contribute and 

refine their suggestions, with controlled prompting. By 

considering examples and reasoning about what happens 

if the first egg does or doesn’t break, students are guided 

through the general version of this puzzle, culminating in 

the derivation of the general solution. 

Puzzle-based learning shares many pedagogical goals 

with the emerging paradigm of computational thinking. 10 

Puzzle-based learning resonates with the computational 

thinking emphasis on abstraction and analytical thinking. 

With reference to Figure 1, computational thinking straddles 

the problem skill spectrum but places more emphasis 

on problem-based and project-based learning. With its 

emphasis on domain-independent, rigorous, and transferable 

reasoning, we believe that puzzle-based learning 

lays a basis for computational thinking in the curriculum. 

PUZZLEfiBASED LEARNING COURSES 

A few different versions of the puzzle-based learning 

course are currently being taught. The course can be 

offered as a three-unit full semester elective course, typically 

of three contact hours per week, split into lectures 

and tutorials; a three-unit full-semester freshman seminar 

of three contact hours per week; or a one-unit freshman 

seminar and a one-unit core module as part of some other 

course. 

One important point about puzzle-based learning 

courses is that they are not about presenting and discussing 

a variety of puzzles, but rather about presenting, 

discussing, and understanding problem-solving principles 

and some mathematical principles in the context 

of puzzles that serve as entertaining illustrations of the 

concepts presented. Also, the process of understanding 

problem-solving principles leads students through a variety 

of topics, exposing them to many important concepts 

at early stages of their college education. 

The process of understanding 

problem-solving principles leads 

students through a variety of topics, 

exposing them to many important 

concepts. 

Despite a variety of possible puzzle-based learning 

offerings, the course’s structure is very much the same. 

The topics listed below correspond to a 12-week semester 

regardless of whether each topic is allocated one hour 

or three. Although the topics have some dependency, as 

we build and develop our model for problem solving, the 

later topics’ order can be rearranged. The topic structure 

also supports a high-level first pass and a secondary, 

detailed pass model for younger students or for development 

over a multiyear curriculum. Similar topic structures 

are employed across both secondary and tertiary student 

environments, as well as our presentations of puzzle-based 

learning concepts to industry, as the following list shows: 

1. Introduction. What it is it all about? 

2. The problem. Rule #1: Understand the problem. 

3. Intuition. Rule #2: Be cautious of your intuitions— 

guess, but verify. 

4. Modeling. Rule #3: Reason, model, calculate. 

5. Some mathematical principles. Do you know how much 

you already know? 

6. Constraints. Am I aware of all of the constraints? 

7. Optimization. What is the best arrangement, and is it 

one that I can actually use? 

8. Probability. Counting skills and counterintuitive 

results. 

9. Statistically speaking. What does an apparently convincing 

statement actually mean? 

APRIL 2010 

A 


aG 

F 

23

A 


aG 

F 

24 


10. Let’s simulate. Can we generate the answer? 

11. Pattern recognition. What is next? 

12. Strategy. Shall we play? 

We illustrate each topic with a variety of puzzles presented 

interactively. The course introduces a few simple 

problem-solving rules that we refer to in every class. 

Each week, students receive homework assignments 

that cover one or more puzzles addressed in class. The 

following week, at the beginning of class, the instructor 

presents and discusses solutions. In one instance of the 

course, homework contributes 30 percent toward the 

final grade, and the final exam contributes the remaining 

70 percent. Students can access all lecture slides, audio 

Without understanding the problem, 

all efforts to find a solution usually 

simply waste time: the dictum of 

solve the right problem and solve the 

problem right. 

lecture recordings, and additional material, including 

course software. Sample course work might include a 

sample lecture that poses the following question: “The 

problem: What are you after?” The lecture introduces the 

most important problem-solving rule. 

First rule of problem solving 

Be sure you understand the problem and all the basic 

terms and expressions used to define it. 

Indeed, without understanding the problem, all 

efforts to find a solution usually simply waste time: the 

dictum of solve the right problem and solve the problem 

right. Underspecificationfican be used as a tool that 

encourages students to determine what they know, 

what they don’t know and must find out, and what they 

cannot find out. 

The approach places the emphasis for knowledge discovery 

on the students and forces them to accept that, 

on occasion, they must provide their best guess. The first 

puzzle we use to illustrate this simple observation is one 

of Martin Gardner’s favorites. 

Puzzle 1 

A farmer has the following: 20 pigs, 40 cows, and 60 

horses. How many horses does he have, if he calls the 

cows horses? 

It takes students a short time to understand the problem, 

“calling,” which has little to do with “having.” The 

farmer still has 60 horses. 

This example can be followed by another classic. 

COMPUTER 

Puzzle 2 

You drive a car at a constant speed of 40 kph from A to 

B, and on arrival at B, you return immediately to A but at 

a higher speed of 60 kph. What was your average speed 

for the whole trip? 

Again, many students would jump immediately into 

the obvious answer of 50 kph without much understanding 

of what the average speed is—or rather, how 

average speed is defined. Most students are surprised to 

discover the correct answer of 48 kph. The next time, in 

any course they take in their programs, they will think 

twice before they answer a question on some average. 

We seek to foster this clear and thoughtful analysis to 

hone and guide intuition. 

Homework 

Clearly, there are strong connections between the process 

of understanding the problem and critical thinking. A lecture 

might include a slide with a statement containing loose terminology, 

strawman arguments, and logical fallacies, and 

students are asked to discuss it. We seek to encourage a focus 

on critical thinking toward finding a solution. 

One weekly assignment given to the students at the end 

of this lecture proceeded as follows. 

With a 7-minute hourglass and an 11-minute hourglass, 

find the simplest way to time the boiling of an egg for 15 

minutes. 

Note that we are not after any solution, but the simplest 

solution. One week later, after all homework has been 

handed in, the lecturer has interesting material for discussion, 

as some students found the solution A: 

Start with the 7- and 11-minute hourglasses, when the 

egg is dropped into the boiling water. 

After 7 minutes, invert the 7-minute hourglass. 

After 4 additional minutes (when sand in the 

11-minute hourglass stops), invert the 7-minute hourglass 

again. 

When the sand stops in the 7-minute hourglass, 15 

minutes will have elapsed. 

Whereas other students found solution B: 

Start the 7- and 11-minute hourglasses. 

After 7 minutes, drop the egg into the boiling water. 

After 4 additional minutes (when the sand in an 

11-minute hourglass stops), invert the 11-minute 

hourglass. 

When the sand stops in the 11-minute hourglass 

again, 15 minutes will have elapsed from the time 

the egg dropped into the water. 

Which of these solutions is simpler? Solution A takes 

15 minutes to complete and requires two inversions. 

A 


aG 

F

A 


aG 

F 

Solution B requires 22 minutes, but only one inversion. 

Most students believed that solution A was simpler, as 

it required less time. They were, however, less certain 

about that when they were told that the hourglasses 

involved were quite heavy, weighing 100 kg each. Such 

a puzzle provides excellent material not only for a discussion 

on understanding the problem, but also on what 

the simplest solution means. This also introduces the 

concept of multiobjective optimization, a concept students 

are usually exposed to during their third year of 

studies. 

OTHER EXAMPLES 

We can use many different puzzles to illustrate fundamental 

concepts of probability, statistics, pattern 

recognition, games, constraint processing, and optimization. 

Here we present a few examples. 

Puzzle 3 

A farmer sells 100 kg of mushrooms for $1 per kg. The 

mushrooms contain 99 percent moisture. A buyer makes 

an offer for these mushrooms at the same price a week 

later. However, another week later the mushrooms would 

have lost 98 percent of their moisture content. How much 

will the farmer lose if he accepts the offer? 

This is a good example for resisting immediately intuitive 

answers—it might not be obvious that the farmer will 

lose $50. 

Puzzle 4 

A bag contains a single ball, which is known to be either 

white or black, with equal probability. A white ball is put in, 

the bag shaken, and a ball is then randomly removed. This 

ball happens to be white. What is the probability now that 

the bag currently contains the white ball? Puzzle 4 thus 

introduces basic concepts in probability. 

Puzzle 5 

There are n dots on the flat surface of a plane. Two 

players, A and B, move alternatively, with A moving first. 

The game’s rules are the same for both players: At each 

move, they can connect two points, but they cannot 

connect points that were already directly connected to 

each other or connect a point with itself. They build a 

graph with predefined n vertices by connecting some of 

the dots. 

The winner is the one who connects the graph (a graph 

is connected if there is a path between any two nodes of 

the graph; however, not every two nodes must be connected 

directly). Which player, A or B, has a winning 

strategy? 

This puzzle introduces graphs and investigating the 

concept of strategies—discovering the winning strategies 

of the first or second player is not trivial. 

More puzzles 

We collected and organized a few hundred educational 

puzzles into meaningful subsets. All teaching materials 

and the new textbook (Puzzle-Based Learning: Introduction 

to Critical Thinking, Mathematics, and Problem Solving) are 

now in active use—the text follows the structure of the 

course given earlier. Chapter 13 of the text includes a collection 

of problems without a solution. These can be used 

for homework, assignments, and exams. 

There are many ways to evaluate students’ progress in 

the puzzle-based learning course. These vary from evaluations 

based on participation through evaluations based 

From this point, students could still 

discuss different interpretations of 

the result, encouraging thought. 

on homework to final exams. Many students might be 

a bit nervous on encountering a final exam loaded with 

puzzles. However, the exam can be organized in many 

ways that will make it meaningful. For example, last 

semester the final exam questions included the following 

two examples. 

Question 1 

Five minutes after midnight of April 13th, 2004, a heavy 

rain fell in Melbourne. What is the probability that, 72 hours 

later, it would be sunny there? Justify your answer. 

This question checked students’ skills for understanding 

the problem before providing an answer. (For the curious, 

the probability is zero percent.) 

Question 2 

The hour and minute indicators of my watch cover each 

other every 65 minutes. Is my watch running too quickly or 

too slowly? Justify your answer. 

This question tested students’ modeling skills and 

rewarded them for identifying the implicit question “When 

should the hands of a watch cover each other,” and for 

modeling the problem, even if they reached an incorrect 

conclusion. 

Wide range 

As a glimpse into the range of approaches to a 

problem consider the following: In class we examined 

different reasoning styles—quantitative versus qualitative 

versus intuitive versus wild guess. Both quantitative 

and qualitative methods are rigorous, using numbers 

and algebra while the other uses language and logic. 

One student group went through the precise calculations 

to determine when an overlap should occur on 

a correct watch, while another group qualitatively 

APRIL 2010 

A 


aG 

F 

25

A 


aG 

F 

26 


reasoned that if a clock were correctly running, then at 

60 minutes past noon, the minute hand would be over 

the 12 and the hour hand over the 1. Five minutes later, 

the minute hand would be over the 1 and the hour hand 

would have moved a little forward. From this point, 

students could still discuss different interpretations of 

the result, encouraging thought. 

For more information on the nature of puzzles and 

the approaches used in puzzle-based learning, we direct 

readers to the website associated with the text, ____ 

www. 

PuzzleBasedLearning.edu.au. 

___________________ 

UNIVERSITY OF ADELAIDE 

We now explore two of this course’s implementations: 

the primary development site at the University of Adelaide 

and another at Carnegie Mellon University, Pittsburgh. The 

initial implementation of puzzle-based learning offered 

Students are interested in the 

material, but their interest can 

easily be capped when they feel 

constrained by the assessment 

mechanisms. 

a one-unit course set as a component of a thee-unit firstyear 

course, Introduction to Electronic Engineering. After 

gaining support from the Dean of the Faculty of Engineering, 

Computer and Mathematical Sciences, from 2009, the 

university placed this one-unit offering into introductory 

courses across the engineering programs available within 

the university. 

A three-unit, first-year course for students planning 

to major in computer science launched simultaneously 

in 2009, and was made available to all nonengineering 

students at the university. We refer to the one-unit offering 

as PBL-E (PBL for Engineers), and the three-unit offering 

as PBL Main. The courses cover the same material, but at 

different levels of depth. 

The intake for the two courses is quite different, as 

PBL-E students have a higher Tertiary Entrance Rank on 

average and have also taken two mathematics courses 

from secondary school. The students in PBL Main might 

have a single course of mathematics if enrolled in the 

Bachelor of Computer Science program, or might have no 

mathematics beyond the middle of Secondary School, if 

from other programs. 

In 2008, 325 students undertook the first offering 

of one-unit PBL, with a weekly one-hour lecture, 

supported with online forums and assessed through 

weekly assignments and a final, hour-long examination. 

In 2009, 428 students undertook the one-unit PBL 

COMPUTER 

course, with another 102 students undertaking the 

three-unit PBL Main course. The PBL-E course remained 

essentially the same in structure, but the three-unit 

course added an additional lecture per week, along 

with weekly tutorials. This let instructors explore the 

material’s development in further depth. The majority 

of PBL Main homework consisted of two questions to 

be completed during the week, rather than the single 

question posed by PBL-E. 

Patterns of learning 

Lectures in PBL follow a set pattern. Each week, the 

first lecture presents the solution to the previous homework, 

identifies key points for that week’s lectures, then 

builds on the topic area. The lecture concludes with the 

next assignment. Lectures are recorded and the lecture 

slides, recordings, and all assignment work made available 

on the course’s electronic forum. These forums also 

provide message boards for student interaction. 

PBL Main has a second lecture that develops the 

themes of the week’s topic. Lecture materials are developed 

in parallel, with the single PBL-E lecture derived 

from a revision and abridgement of the two PBL Main 

lectures for that topic to maintain currency between 

the two courses. 

The university offers tutorials for PBL Main that let students 

take part in collaborative problem-solving exercises, 

while a tutor provides assistance and guidance. Tutorial 

groups can hold up to 25 students, divided into subgroups 

of five to eight for problem solving. During these sessions, 

we introduce fundamental mathematical concepts useful 

in the later course, including counting and the bases of 

probability, such as factorials, combinations, and permutations. 

This addresses the differences in mathematical 

preparation identified in the intake. 

While a good grasp of mathematics can be useful for 

PBL, it is not essential. Problem specification has been 

a key concern, as the larger classes contain students 

accustomed to a completely specified problem, and thus 

feel uncomfortable when confronted with problems not 

completely specified or, in the student’s opinion, not sufficiently 

and exactly specified. While some students regard 

this as a challenge, and also as an intellectual freedom, 

others have found it to be a stumbling block. 

Assessment of the course has proven to be one of the 

largest implementation issues. Students are interested in 

the material, but their interest can easily be capped when 

they feel constrained by the assessment mechanisms, or 

feel they haven’t received sufficient, personalized feedback. 

Early assessment for PBL-E revolved around a mark 

for each assignment out of five, followed by feedback to 

the group that demonstrated the marking scheme and 

solution. Instructors also presented the previous lesson’s 

solution at the lecture’s start, which corresponded with the 

A 


aG 

F

A 


aG 

F 

hand-in time, to let students immediately gain feedback on 

the quality of their solution. 

PBL-E’s student numbers posed a significant resource 

issue: Without detailed feedback, it takes approximately 

two to three minutes to mark each assignment. Thus, the 

marking load starts at approximately eight hours for each 

assignment, with a team of markers employed and trained 

to provide consistency of response. 

PBL Main has a much smaller enrollment but employs 

detailed, personalized feedback that also takes approximately 

eight hours to complete a week’s assignments. The 

requirement for a consistent and reproducible marking 

scheme that can be assigned to multiple markers constrains 

the range of problems that can be offered. Problems 

with too many possible solutions become effectively 

impossible to mark across 450 students. 

In response to this, we have considered many alternatives 

and are currently developing problems that might 

have multiple possible solutions, all of which may appear 

to be correct when, in fact, only one is. Again, this is an 

issue of problem and solution specification. Controlled use 

of multiple-choice questions, with between eight and ten 

options, lets markers quickly identify the flaw in reasoning 

and correctly mark the student’s work. We also investigate 

the possibility of reducing the dependency on a markbased 

assessment for this course. 

Early student response shows that they enjoy the course 

material and it does develop their thinking skills. However, 

several students, especially in PBL-E, encounter issues 

with the assessment model and its perceived lack of feedback. 

Others worry that the assessment mechanism can 

develop a negative and unproductive approach to the 

course. We are actively seeking to address these concerns 

by allocating more resources to marking and feedback, 

and through the use of automated marking mechanisms 

that allow more rapid response. Future implementations 

of PBL-E may include tutorials or alternative assessment 

mechanisms. 

CARNEGIE MELLON UNIVERSITY 

Carnegie Mellon University offered puzzle-based learning 

as a nine-unit, three-credit freshman seminar in spring 

2009. Given the spring course’s seminar nature, instructors 

capped enrollment at 15, but we found it encouraging to 

see that the wait list exceeded the class’s enrollment.fiThe 

class had an interdisciplinary mix of students majoring 

in Information Systems, Computer Science, Psychology, 

Statistics, Cognitive Science, Economics, and Physics. The 

class met twice a week for 80 minutes. 

In addition to the knowledge gained from the Adelaide 

experience, the smaller size of this class let us experiment 

with several alternative themes. For example, after the 

introductory classes, each session started with a puzzle 

of the day. 

First, a student would present a puzzle of his or her 

choice. The class as a whole tried to solve this puzzle, 

with hints and guidance provided by the puzzle poser. 

Puzzles chosen by the students ranged the gamut from 

logic puzzles to diagrammatic reasoning to physical-puzzle 

tangrams. Students submitted a one-page write-up of their 

puzzle, solution, and—most importantly—their reflection 

on the puzzle: What captured their interest in the puzzle 

and its variations, and how did the solution tie in with the 

overall course’s instruction? 

In addition to the daily puzzle, we also conducted a 

puzzlethon that again presented a puzzle of the students’ 

choice. But this time the class voted on the best puzzle (a 

combination of presentation and the puzzle’s nature) and 

instructors distributed prizes to the winners. 

During our discussion of scientific and mathematical 

induction, given the smaller size of the class, we played 

Robert Abbott’s inductive game Eleusis, 11 which models the 

To emphasize the link between 

the thought processes involved in 

solving puzzles and addressing open 

real-world problems, we examined a 

few case studies. 

scientific method. To introduce students to some problemsolving 

thoughts of leaders in the field, we watched a few 

videos. These included Polya’s “Let us teach guessing” in 

which he beautifully illustrates several problem-solving 

heuristics embraced by puzzle-based learning in the process 

of deriving a solution to the five-plane problem; an 

interview with Herb Simon on being a researcher with 

advice to undergraduates; Richard Feynman’s discussion 

of problem solving and induction; and Will Shortz’s documentary 

Wordplay on crossword puzzles, their creators, 

and solvers. 

We also visited the Pittsburgh Supercomputing Center 

open house to glimpse problem solving in the real world. 

To emphasize the link between the thought processes 

involved in solving puzzles and addressing open real-world 

problems, we examined a few case studies, including the 

recently cracked Netflix Prize (www.netflixprize.com) and 

the classic work of Mosteller 12 in resolving the authorship 

of some of the disputed Federalist papers. 

Student evaluation was done with components for class 

participation, puzzle presentations, homework assignments, 

three exams, and a few in-class quizzes. It has 

been gratifying to see that the response to the class has 

been favorable, with some students commenting it was 

the best class they had that semester. Some have also commented 

on the revisiting of PBL themes in their subsequent 

APRIL 2010 

A 


aG 

F 

27

A 


aG 

F 

28 


discipline-specific courses. Carnegie Mellon University’s 

Doha, Qatar, campus hosted a summer 2009 version of this 

course, while the freshman seminar is slated to be offered 

in Pittsburgh during spring 2010. 

Puzzle-based learning is an in-progress experiment 

that seeks to foster general domain-independent 

reasoning and critical thinking skills that can lay 

a foundation for problem-solving in future course 

work. As fun as puzzles inherently are, they provide 

only a means to this pedagogical end. Our preliminary 

experience in different instantiations of the course and 

educational contexts has been encouraging and well 

received as we explore this approach. We continue collecting 

relevant data to demonstrate the benefit of our 

approach. Early results 13 indicate that students who enroll 

in our course perceive an improvement in their thinking 

and general problem-solving skills. 

References 

1. B. Parhami, Puzzling Problems in Computer Engineering, 

Computer, Mar. 2009, pp. 26-29. 

2. A. Fisher, Critical Thinking: An Introduction, Cambridge 

University Press, 2001. 

3. University of Adelaide, Mathematics Working Party 

Final Report, internal document, June 2008. 

4. G. Polya, How to Solve It: A New Aspect of Mathematical 

Method, Princeton Univ. Press, 1945. 

5. M. Gardner, Entertaining Mathematical Puzzles, Dover 

Publications, 1961. 

6. M. Danesi, The Puzzle Instinct: The Meaning of Puzzles 

in Human Life, Indiana Univ. Press, 2002. 

7. W. Poundstone, “How Would You Move Mount Fuji?” 

Microsoft’s Cult of the Puzzle—How the World’s 

Smartest Companies Select the Most Creative Thinkers, 

Little Brown and Company, 2000. 

8. P.C. Blumenfeld et al., “Motivating Project-Based 

Learning: Sustaining the Doing, Supporting the 

Learning,” Educational Psychologist, vol. 26, nos. 3 

and 4, 1991, pp. 369-398. 

9. J.D. Bransford et al., “Teaching Thinking and Problem 

Solving: Research Foundations,” Am. Psychologist, 

vol. 41, 1986, pp. 1078-1089. 

10. J.M. Wing, “Computational Thinking,” Comm. ACM, 

Mar. 2006, pp. 33-35. 

11. M. Gardner, Penrose Tiles to Trapdoor Ciphers, Mathematical 

Assoc. of America, 1997. 

12. F. Mosteller and D. Wallace, “Deciding Authorship,” 

Mathematics: People, Problems, Results, D.M. Campbell 

and J.C. Higgins, eds., Wadsworth Publishing, 

1984. 

COMPUTER 

13. N.J.G. Falkner, R. Sooriamurthi, and Z. Michalewicz, 

“Puzzle-Based Learning: The First Experiences,” 

Proc. 20th Ann. Conf. Australasian Assoc. for Eng. 

Education (AaeE 09), 2009. 

Nickolas Falkner is a lecturer in the School of Computer 

Science at the University of Adelaide. He received a PhD in 

discovery and classification of information in large systems 

from the University of Adelaide. His research interests 

include wireless sensor networks, automated network configuration, 

data fusion, and data stream management. 

He is also active in educational research, with a focus on 

increasing student participation, retention, and enthusiasm. 

Contact him at jnick@cs.adelaide.edu.au. 

_________________ 

Raja Sooriamurthi is an associate teaching professor 

with Carnegie Mellon's Information Systems Program and 

also has a faculty appointment at Carnegie Mellon's Heinz 

College. He received his PhD in computer science from Indiana 

University, Bloomington. His research interests are in 

artificial intelligence, case-based reasoning, and CS/IS pedagogy. 

His pedagogical efforts have been recognized with 

several awards for distinguished teaching. Sooriamurthi 

is a member of the IEEE Computer Society, the ACM, and 

Decision Sciences Institute. Contact him at raja@cmu.edu. 

__________ 

Zbigniew Michalewicz is a professor in the School of 

Computer Science, University of Adelaide. He is also at 

the Institute of Computer Science, Polish Academy of Sciences 

and at the Polish-Japanese Institute of Information 

Technology, Poland. He received a PhD in computer science 

from the Polish Academy of Science. His research interests 

include problem-solving methodologies, evolutionary computation, 

modern heuristic methods, and adaptive business 

intelligence. Michalewicz is a Fellow of the Australian Computer 

Society. Contact him at zbyszek@cs.adelaide.edu.au. 

___________________ 

Selected CS articles and columns are available for free 

at http://ComputingNow.computer.org. 

For more information 

on any topic 

presented in Computer, 

visit the IEEE 


Digital Library at 

www.computer.org/csdl 

A 


aG 

F

A 


aG 

F 

PERSPECTIVES 

Using Codes of Conduct 

to Resolve Legal Disputes 

Peter Aiken, Virginia Commonwealth University 

Robert M. Stanley and Juanita Billings, Data Blueprint 

Luke Anderson, Duane Morris LLC 

In the absence of other published standards of care, it is reasonable 

for contractual parties to rely on an applicable, widely available code 

of conduct to guide expectations. 

When legal disputes arise, the primary 

focus of judges, juries, and arbitration 

panels is on interpreting facts. In 

cases of alleged underperformance, 

they must evaluate facts against contract 

language, which typically states that services will 

be provided in accordance with industry standards. Legal 

arbiters seek well-articulated “standards of care” against 

which to evaluate the behavior of contractual parties and, 

in the absence of other published standards, increasingly 

rely on codes of conduct (CoCs) to establish an objective 

context. In fact, they have successfully applied CoCs— 

including the ACM/IEEE-CS CoC—in instances where the 

parties were not even affiliated with the CoC-sponsoring 

organization. 

We illustrate the current application of CoCs with a 

fictional enterprise resource planning (ERP) system implementation 

failure that is a compilation of real-life cases. 

Subject to binding panel arbitration, the plaintiff and defendant 

in the case presented conflicting interpretations of the 

same facts: From the plaintiff’s perspective, the defendant 

0018-9162/10/$26.00 © 2010 IEEE 


failed to migrate the ERP system as promised; the defendant 

countered that defective and poor-quality data delayed 

the migration. Using the ACM/IEEE-CS CoC as a reference, 

expert testimony convinced the arbitration panel that the 

defendant’s position was untenable, and the panel accordingly 

awarded the plaintiff a multimillion-dollar judgment. 

CASE STUDY 

Acme Co. received a directive from its parent corporation 

mandating replacement of its legacy pay and 

personnel systems with a specific ERP software package 

designed to standardize payroll and personnel processing 

enterprise-wide. Upon the vendor’s “referred specialist” 

recommendation, Acme Co. contracted with ERP Systems 

Integrators to implement the new system and convert its 

legacy data for $1 million. 

The contracted timeline was six months, beginning in 

July and wrapping up with a “big bang” conversion at the 

end of December. The year-end conversion failed, allegedly 

due to ERP Systems Integrators’ poor data migration 

practices, and Acme Co. had to run the old and new sys- 

APRIL 2010 

A 


aG 

F 

29

A 


aG 

F 

30 

PERSPECTIVES 

ARBITRATION VERSUS CIVIL SUITS 

In contrast to civil suits tried in a court of law, arbitration vests the 

functions of judge and jury in a panel of arbitrators, typically 

lawyers or industry professionals, whose time is paid for by the 

involved parties. Arbitration is private and frequently subject to 

cončdentiality and nondisclosure agreements—because only the 

participants know the details, arbitration inćuences future litigation 

at a slower rate than do public court proceedings. 

An arbitration panel does not issue an opinion; instead, it hands 

down a one-page decision to award damages (or not)—typically 

monetary—to one party or the other. The reasoning supporting 

any judgment thus must be inferred from the parties’ arguments. 

Arbitration decisions are čnal and generally cannot be appealed. 

Remaining rules of court are much like those in a trial. A preliminary 

phase is dedicated to evidence gathering, motion exchanges, 

depositions, and other discovery forms. Lawyers for each party try 

to convince the panel of the validity of their client’s position. 

tems in parallel—a complex and expensive situation that 

it had carefully planned to avoid and that ERP Systems 

Integrators had assured them would not occur. When the 

conversion was pushed into April of the following year, 

Acme Co. slowed and then ceased paying ERP Systems 

Integrators’ invoices. In July, ERP Systems Integrators 

pulled its implementation team and Acme Co. initiated 

arbitration. 

Most IT projects are governed by contracts that assign 

responsibilities to each party and provide specific remedies 

for delayed implementation or project failure. Such 

contracts require the parties to submit to private, binding 

arbitration to resolve disputes. As the “Arbitration versus 

Civil Suits” sidebar indicates, this process slightly differs 

from civil litigation in a court of law. However, the use of 

CoCs applies equally to both settings. 

Almost a year passed before the arbitration hearing. 

Meanwhile, Acme Co. and ERP Systems Integrators 

deposed witnesses, and experts scrutinized sales materials, 

project artifacts (e-mails, status reports, project plans, 

and so on), contract documents, application software, 

migration tools, and contents of the shared-drive implementation 

environment. 

THE “STANDARD OF CARE” DILEMMA 

The arbitration panel had to resolve three key issues: 

Who was responsible for project management? Acme 

Co. produced paperwork indicating that responsibility 

rested with ERP Systems Integrators. The plaintiff 

claimed that it had no idea how to implement such a 

system and had hired the defendant to provide such 

expertise—including project management. 

What standards applied to the programming used for 

data conversion? Acme Co. attacked specific conversion 

software changes as harmful in that they 

COMPUTER 

increased the amount of incorrect data within the 

converted database by an unnecessarily complicated 

order of magnitude. ERP Systems Integrators 

responded that the referenced software changes did 

not constitute “software engineering” and thus were 

not subject to CoC guidance. 

How significant were project communication failures? 

During discovery, numerous intracompany e-mails 

from ERP Systems Integrators described the project 

in a markedly more pessimistic tone than the communications 

delivered to Acme Co. in compliance with 

contract provisions. 

These issues collectively fell under the “standard of 

care” portion of the contract. The dilemma Acme Co. 

faced—one common to companies in the same position—was 

detailing the standard of care it expected from 

ERP Systems Integrators. The contract language specified 

that ERP Systems Integrators “warrants that the services 

it provides hereunder will be performed in a professional 

and workmanlike manner in accordance with industry 

standards.” 

As the following exchange shows, ERP Systems Integrators 

could not provide more detail regarding the warranty 

statement: 

Question: What are the industry standards that you are 

referring to? 

Defense: There is nothing written or codified, but they 

are the standards recognized by the consulting firms 

in our industry. 

Question: I understand that the industry standards that 

you are referring to here are not written down anywhere; 

is that correct? 

Defense: That is my understanding. 

Question: Have you made an effort to locate these 

industry standards and have simply not been able to 

do so? 

Defense: I would not know where to begin to look. 

For its part, Acme Co. argued that suitable Internetbased 

CoCs were available to guide various behaviors. 

The “Online Codes of Conduct” sidebar describes one 

useful resource, the Online Ethics Center, that aggregates 

numerous CoCs. In particular, the plaintiff referenced the 

ACM/IEEE-CS Software Engineering Code of Ethics and 

Professional Practice. The “SECEPP” sidebar provides a 

brief history of this CoC. 

Acme Co. successfully argued to the arbitration panel 

that, when faced with obscure or publicly available standards, 

contracting parties should expect the accessible 

standards to apply. The plaintiff then cited objective, 

concrete portions of SECEPP that directly supported its 

positions on the three disputed issues. 

A 


aG 

F

A 


aG 

F 

PROJECT 

MANAGEMENT 

RESPONSIBILITY 

In pre-arbitration depositions, 

ERP Systems Integrators 

asserted that it was not the 

project manager and that the 

contract specified its performance 

solely at the direction 

of Acme Co. The plaintiff contended 

that, while contract 

language did exist, overall 

project management lay 

with the defendant because 

it assumed that role in spite 

of its denials. In its defense, 

Acme Co. cited Section 2.7 

of SECEPP, which states that 

“computing professionals 

have a responsibility to share 

technical knowledge with 

the public by encouraging 

understanding of computing, 

including the impacts of 

computer systems and their 

limitations.” 

The panel members 

understood SECEPP to be 

analogous to a building 

code and that, because of its 

broad wording, applied to the 

project in general and not 

specifically to its software 

engineering aspects. 

Expert testimony supported 

Acme Co.’s claims. 

Referencing widely published 

and accepted principles 1 that 

supplemented SECEPP, the 

plaintiff prepared a framework of project management 

behaviors as shown in Table 1. Specific evidence included 

a timesheet signed weekly by ERP Systems Integrators 

charging approximately 2,000 hours against the 

job category “Project Manager” and the task “Project 

Management.” 

It was obvious from the evidence that ERP Systems Integrators 

was hired in a specialist capacity and that Acme Co. 

had no ability to provide oversight. The arbitration panel 

determined that the defendant acted as, and clearly was, 

the project manager. 

RELEVANT PROGRAMMING STANDARDS 

ERP Systems Integrators blamed the conversion failure 

on “bad data.” However, Acme Co. provided evidence of 

ONLINE CODES OF CONDUCT 

The Online Ethics Center for Engineering and Research (http://onlineethics.org) is a joint project of 

the Center for Ethics, Engineering, and Society at the National Academy of Engineering and the 

Ethics Education Library at the Center for the Study of Ethics in the Professions at the Illinois Institute 

of Technology. Funded by a grant from the National Science Foundation, the website brings together 

more than 50 CoCs from organizations including: 

Institute of Electrical 

and Electronics Engineers 

(IEEE) 

 

- 

 

 

 

 

 

- 

 

 

 

 

Association for 

Computing Machinery 

(ACM) 

 

 

 

 

 

Institute for Certification 

of Computing 

Professionals (ICCP) 

 

 

 

 

While not uničed, the collection exhibits a striking cohesiveness. 

SECEPP 

Data Management 

Association (DAMA) 

 

- 

 

 

Originally adopted in 1972 by the ACM and the IEEE Computer Society, the Software Engineering 

Code of Ethics and Professional Practice principally served as a method of “self-regulation,” 

listing violations and accompanying sanctions. In 1993, SECEPP was revised to “clarify and formally 

state” the consensus of professional ethical requirements for which “the profession (is) accountable 

to the public.” The more comprehensive code was also designed to serve “as an aid to individual 

decision making.” 1 In the years since adoption, SECEPP (www.computer.org/computer/code-ofethics.pdf) 

has provided the foundation for numerous subsequent guidelines. Knowledgeable 

persons are aware of the code’s positive impact on professionalism within the IT industry. 2 

______________________ 

______ 

References 

1. R.E. Anderson et al., “Using the New ACM Code of Ethics in Decision Making,” Comm. ACM, Feb. 

1993, pp. 98-107. 

2. S. Rogerson, “An Ethical Review of Information Systems Development and the SSADM 

Approach,” Centre for Computing and Social Responsibility, De Montfort Univ., UK, 2 Jan. 

2008; www.ccsr.cse.dmu.ac.uk/staff/Srog/teaching/ssadm.htm. 

flawed programming practices, missing analysis data, and 

measurably lower-quality converted data. 

Failure to test for other values 

The plaintiff alleged that the defendant failed to follow 

generally accepted testing standards. Evidence consisted 

of instances involving poorly implemented conversion 

software. 

For example, the conversion software was supposed to 

check a specific field value for one of two possible values— 

say, “1” and “2” corresponding to the values “male” and 

“female,” respectively. The software executed by ERP Systems 

Integrators checked to see if the source field value 

was “1” and, if so, assigned the value “male” to the converted 

field; if the source field value was not “1,” it assigned 

APRIL 2010 

A 


aG 

F 

31

A 


aG 

F 

32 

PERSPECTIVES 

Table 1. Summary evidence of project management behaviors. 

Process area 

the value “female” to the converted field without determining 

and reporting possible nonconforming values. 

Section 1.2 of SECEPP states that “to minimize the 

possibility of indirectly harming others, computing 

professionals must minimize malfunctions by following 

generally accepted standards for system design and 

testing.” Accepted software engineering programming 

standards would call for testing for positive confirmation 

of “2” before setting the converted value to “female” and 

for reporting incoming values and numbers of values not 

“1” or “2.” The defendant’s failure to follow these standards 

permits “3” in the source data to be assigned the value 

“female” after conversion, resulting in demonstrably lowerquality 

converted data. 

Failure to prevent duplicate record insertion 

Acme Co. demonstrated that ERP Systems Integrators’ 

software produced other structure-related conversion 

errors. 1 The defendant’s e-mail traffic revealed an urgent 

need to get records in the system “even if they weren’t 

the correct ones.” Instead of attempting to determine 

why the conversion programs would not successfully 

complete, ERP Systems Integrators identified the lines of 

code prohibiting the insertion of duplicate records and 

“commented them out,” thereby inactivating the software 

functionality. 

Consequently, there were 63,131 customers instead 

of approximately 6,000 and 100,236 employee records 

instead of approximately 10,000 in the system after con- 

COMPUTER 

Defendant lead 

Methodology Demonstrated 

Plainti 

lead 

 

 

 

 

 

 

 

 

 

 

? 

 

 

 

? 

 

 

version. This in turn increased the data clean-up 

cost. Due to the inherent complexities of working 

in a multiflawed environment, 2 the cost to clean 

up 10 times more data is often much greater than 

10 times the cost of cleaning up the original data. 

Section 6.08 of SECEPP states that software 

engineers should “take responsibility for detecting, 

correcting and reporting errors in software 

and associated documents.” The arbitration panel 

agreed with the plaintiff that the code provided 

an objective measure to assess responsibility for 

minimizing software malfunctions and correcting 

errors, and as such could reasonably guide Acme 

Co.’s expectations. 

COMMUNICATION FAILURE 

RESPONSIBILITY 

Acme Co. alleged that ERP Systems Integrators 

withheld important information from the on-site 

consulting team. The plaintiff presented e-mails of 

defendant personnel exchanging dire predictions 

about the project’s fate. One message warned it 

could become “our biggest mess!” These starkly 

contrasted with the rosy reports presented by ERP 

Systems Integrators during status meetings. 

On the subject of a client’s obligation to communicate 

project failure indicators, SECEPP is unambiguous: According 

to Section 2.06, “any signs of danger from systems 

must be reported to those who have opportunity and/ 

or responsibility to resolve them.” The evidence clearly 

showed a pattern by the defendant of communicating one 

message internally (project failure) and a second message 

to plaintiff (everything okay). 

A second communication failure occurred at a more 

systematically significant level. All project management 

guidelines stress the importance of treating project 

planning diagrams as living documents, and most are 

managed via specialized software that permits determination 

of planned versus actual. Drawing on the Project 

Management Body of Knowledge (PMBOK), Acme Co. demonstrated 

that by never updating the project plan shown in 

Figure 1, developed using a simple spreadsheet, ERP Systems 

Integrators was unable to report fact-based measures 

of progress and thus failed to meet expected standards. 

Project statistic metadata lets stakeholders and implementers 

respond to challenges with all parties speaking 

the same vocabulary. Static project plans are out of date 

as soon as any task deviates from the plan and, as a result, 

management cannot determine the status of and impact 

on subsequent tasks. 

Additional evidence indicating a vastly overbooked 

resource pointed to a project that was out of control. Figure 

2 indicates a “plan” for one individual to accomplish the 

work of 18 others. This kind of error occurs in projects 

A 


aG 

F

A 


aG 

F 

Figure 1. 

where the existing environment has not 

been understood well enough to properly 

plan. 

Acme Co. proved that ERP Systems 

Integrators had not performed legacy 

system analysis 3 and that it failed to 

adequately manage project risk. These 

activities are subsumed under Section 

2.5 of SECEPP: “Give comprehensive 

and thorough evaluations of computer 

systems and their impacts, including 

analysis of possible risks.” 

The arbitration panel concluded 

that ERP Systems Integrators’ failure 

to update project plans, communicate 

responsibly, and manage risk appropriately 

constituted an inadequate 

standard of care. The defendant had 

obviously foreseen failure and hid 

from the plaintiff information indicating 

that the project had no possibility 

of succeeding. 

RESOLUTION AND DISCUSSION 

Days after the arbitration hearing concluded, the panel 

issued a one-page decision awarding $5 million to Acme 

Co.: five times the project’s worth. The decision was particularly 

hard-hitting because ERP Systems Integrators’ 

insurance carrier denied coverage for the incident based 

on the evidence of its “failure to perform in a workmanlike 

manner.” 

The ruling favorable to the plaintiff indicated overwhelming 

support for its CoC-based case. The arguments 

Acme Co. presented are deciding factors in a growing 

number of real-life judicial disputes. In technology 

contexts where key issues revolve around competing 

Figure 2. 

 

interpretations of behaviors, CoCs are influencing various 

contracting parties as well as the IT, business, and consulting 

communities. 

Because arbitration results are private, word-of-mouth 

has been the chief means of propagating the success of 

comparing litigant behavior against CoCs. Moving from 

arbitration into case law, CoCs will be increasingly applied. 

In spite of limited current awareness, SECEPP is well on its 

way to becoming a de facto standard as it enjoys growing 

awareness throughout the legal community and increasing 

compliance in the IT profession. 4 

More extensive application of publicly available standards—and 

growing awareness of them—will positively 

impact the IT industry. Five initial benefits accrue to organizations 

capitalizing on CoC knowledge: 

APRIL 2010 

A 


aG 

F 

33

A 


aG 

F 

34 

PERSPECTIVES 

Increasing use and broader applicability. As awareness 

of CoCs grows, so also will their popularity. Moreover, 

as the legal community becomes more familiar 

with CoC-based arguments, they will be applied 

more broadly. CoC expertise could become ubiquitous 

and perhaps as branded as the PMP designation 

from the Program Management Institute (www.pmi. 

org/CareerDevelopment/Pages/Obtaining-Credential. 

___ aspx). CoCs are not only easy to use, they are unambiguous 

about specific, holistic IT professional 

responsibilities—to the project, stakeholders, our 

profession, and society. 

Public codification of conduct standards by IT professionals. 

Evidence is mounting that public CoCs 

serve as standards for evaluating the performance 

and determining the responsibilities not only of IEEE 

members, but IT professionals in general. Organizations 

and professionals are using CoCs to determine 

specific attributes of compliance and noncompliance. 

Preventing and resolving disputes. Guidance to first 

prevent and subsequently settle disputes is generally 

welcome. CoCs provide objective particulars that litigants 

can use in a proactive manner. Following a CoC 

is one way to promote a successful project environment 

and insulate contracting parties from potential 

legal liability. In doing so, it is possible to identify critical 

prelitigation and other decision points that allow 

parties to better deal with or entirely avoid disputes. 

Better understanding of IT project implementation. The 

current dispute resolution process favors contractors. 

Understanding CoC utility enables organizations to 

rethink their relationships with clients. This could 

impact how organizations evaluate, select, and interact 

with IT professionals. 

Organization-wide CoC applicability. SECEPP is guided 

by the philosophy that CoCs generally apply to organizations—including 

those that do not have members 

belonging to the ACM or IEEE Computer Society—as 

well as to their leaders. The potential implications for 

organizations and leadership are staggering. 

Seeking a legal resolution to a dispute over contracted 

IT services is a growing trend. This is not surprising given 

the alarming statistic that up to 70 percent of IT projects 

fail (www.it-cortex.com/stat_failure_rate.htm). Litigation 

of software-intensive endeavors has been called a “major 

growth industry,” with forecasted legal costs rising “faster 

than any other aspect of software development.” 5 

As companies increasingly rely on IT systems to drive 

their business, failures and delayed implementations can 

cause costly ripples throughout their organization. Many 

are unwilling to absorb these costs and, consequently, 

expect IT professionals and especially their vendor partners 

to share responsibility. 

COMPUTER 

The evidence speaks for itself. Courts, juries, 

and arbitration panels are finding that failure 

to follow generally accepted public 

standards for design and testing of software 

are grounds for seeking damages. A 

wider understanding of the existence and usefulness of 

existing ethical and professional standards will represent 

added value to in-house IT managers and enhance the 

stature of IT professionals. An organization’s ability to 

evaluate conduct and, when appropriate, consider potential 

legal matters more knowledgeably is paramount to 

imposing accountability on all participants. 

Acknowledgments 

The authors express gratitude to the anonymous reviewers 

for their comments. They also thank the hundreds of 

unnamed data management and legal professionals with 

whom they have worked and who have contributed to a better 

understanding of this field as it is practiced and as it should 

be practiced. 

References 

1. V.Y. Yoon, P. Aiken, and T. Guimaraes, “Managing Organizational 

Data Resources: Quality Dimensions,” Information 

Resources Management J., July-Sept. 2000, pp. 5-13. 

2. P. Aiken and P. Piper, “Estimating Data Reverse Engineering 

Projects,” Proc. 5th Ann. Systems Reengineering Workshop, 

Naval Surface Warfare Center, 1995, pp. 133-145. 

3. P. H. Aiken, “Reverse Engineering of Data,” IBM Systems J., 

Apr. 1998, pp. 246-269. 

4. D. Gotterbarn, “How the New Software Engineering Code 

of Ethics Affects You,” IEEE Software, Nov./Dec. 1999, pp. 

58-64. 

5. T. DeMarco and T. Lister, “Both Sides Always Lose: Litigation 

of Software-Intensive Contracts,” CrossTalk, Feb. 2000; 

www.stsc.hill.af.mil/crossTalk/2000/02/demarco.html. 

Peter Aiken is an associate professor of information systems 

at Virginia Commonwealth University and founding 

director of Data Blueprint, a data management consulting 

firm based in Richmond, Virginia. Contact him at _____ 

peter@ 

datablueprint.com. 

Robert M. Stanley is a senior member of the technical 

staff and a project lead specializing in project and quality 

management for Data Blueprint. Contact him at _______ bstanley@ 

datablueprint.com. 

Juanita Billings is a senior member of the technical staff 

and a project lead specializing in legal support, systems 

analysis, and project management for Data Blueprint. Contact 

her at jbillings@datablueprint.com. 

___________________ 

Luke Anderson is a law partner specializing in technology 

and intellectual property matters with Duane Morris 

LLC in Atlanta, Georgia. Contact him at ________ 

landerson@ 

duanemorris.com. 

Selected CS articles and columns are available for free at 

http://ComputingNow.computer.org. 

A 


aG 

F

A 


aG 

F 

COVER FEATURE 

Arutyun I. Avetisyan, Institute for System Programming of the Russian Academy of Sciences 

Roy Campbell, Indranil Gupta, Michael T. Heath, and Steven Y. Ko, University of Illinois at 

Urbana-Champaign 

Gregory R. Ganger, Carnegie Mellon University 

Michael A. Kozuch and David O’Hallaron, Intel Labs 

Marcel Kunze, Karlsruhe Institute of Technology, Germany 

Thomas T. Kwan, Yahoo! Labs 

Kevin Lai, Martha Lyons, and Dejan S. Milojicic, HP Labs 

Hing Yan Lee and Yeng Chai Soh, Infocomm Development Authority of Singapore 

Ng Kwang Ming and Jing-Yuan Luke, Malaysian Institute of Microelectronic Systems 

Han Namgoong, Electronics and Telecommunications Research Institute, South Korea 

Open Cirrus is a cloud computing testbed 

that, unlike existing alternatives, federates 

distributed data centers. It aims to spur 

innovation in systems and applications 

research and catalyze development of an 

open source service stack for the cloud. 

There is growing interest in cloud computing 

within the systems and applications research 

communities. However, systems researchers 

often find it difficult to do credible work 

without access to large-scale distributed data 

centers. Application researchers could also benefit from 

being able to control the deployment and consumption 

of hosted services across a distributed cloud computing 

testbed. 

0018-9162/10/$26.00 © 2010 IEEE 

OPEN CIRRUS: 

A GLOBAL CLOUD 

COMPUTING 

TESTBED 


Pay-as-you-go utility computing services by companies 

such as Amazon and new initiatives by Google, IBM, 

Microsoft, and the National Science Foundation (NSF) 

have begun to provide applications researchers in areas 

such as machine learning and scientific computing with 

access to large-scale cluster resources. However, system 

researchers, who are developing the techniques and software 

infrastructure to support cloud computing, still have 

trouble obtaining low-level access to such resources. 

Open Cirrus (http://opencirrus.org) aims to address this 

problem by providing a single testbed of heterogeneous 

distributed data centers for systems, applications, services, 

and open source development research. The project is a 

joint initiative sponsored by Hewlett-Packard (HP), Intel, 

and Yahoo! in collaboration with the NSF, the University 

of Illinois at Urbana-Champaign (UIUC), the Karlsruhe Institute 

of Technology (KIT), the Infocomm Development 

Authority (IDA) of Singapore, the Russian Academy of 

APRIL 2010 

A 


aG 

F 

35

A 


aG 

F 

36 

COVER FEATURE 

Open source cloud stack (Zoni, Hadoop, Tashi) 

Shared infrastructure (>10*1,000 cores) 

Global services (sign-on, monitoring, store, sustainability dashboard, and so on) 

COMPUTER 

HP 

UIUC Intel 

CMU 

Yahoo! 

KIT 

RAS 

MIMOS 

Figure 1. Open Cirrus testbed. Each of the 10 current sites 

consists of a cluster with at least 1,000 cores and associated 

storage. The testbed očers a cloud stack consisting of 

physical and virtual machines and global services such as 

sign-on, monitoring, storage, and job submission. 

Sciences (RAS), the Electronics and Telecommunications 

Research Institute (ETRI) of South Korea, the Malaysian Institute 

of Microelectronic Systems (MIMOS), and Carnegie 

Mellon University (CMU). Additional members are expected 

to join Open Cirrus later this year. 

As Figure 1 shows, the current testbed is composed of 10 

sites in North America, Europe, and Asia. Each site consists 

of a cluster with at least 1,000 cores and associated storage. 

Authorized users can access any Open Cirrus site using the 

same login credential. 

MOTIVATION AND CONTEXT 

Open Cirrus has four main goals. 

First, the project aims to foster systems-level research 

in cloud computing. In the current environment, only big 

service providers such as Yahoo!, Google, Amazon, and Microsoft 

have access to large-scale distributed data centers 

to develop and test new systems and services. Most cloud 

computing researchers must typically rely on simulations 

or small clusters. Open Cirrus aims to help democratize 

innovation in this area by providing two unique features 

essential to systems-level research: 

Open Cirrus sites allow access to low-level hardware 

and software resources—for example, install OS, 

access hardware features, and run daemons. 

The testbed comprises heterogeneous sites in different 

administrative domains around the world, 

enabling researchers to leverage multiple data centers. 

Second, Open Cirrus seeks to encourage new cloud 

computing applications and applications-level research. 

Providing a platform for real-world applications and services 

is an important part of Open Cirrus. Particularly 

exciting are 

IDA 

ETRI 

the potential for developing new application models 

and using them to understand the necessary systemslevel 

support, and 

using the federated nature of Open Cirrus to provide 

a platform for new kinds of federated applications 

and services that run across multiple data centers. 

Third, Open Cirrus offers a collection of experimental 

data. Cloud computing researchers often lack data 

sets with which to conduct high-quality experimental 

evaluations. Open Cirrus sites will let researchers 

import, store, and share large-scale data sets such as 

Web crawls and data-center workload traces. With 

such facilities, Open Cirrus could become a “watering 

hole” where researchers with similar interests can exchange 

data sets and develop standard cloud computing 

benchmarks. 

Fourth, Open Cirrus aims to develop open source 

stacks and APIs for the cloud. To become widespread, 

cloud computing requires a nonproprietary and vendor-neutral 

software stack. Open Cirrus will serve as 

a platform that the open source community can use to 

design, implement, and evaluate such codes and interfaces 

for all cloud stack levels. Open source is as much 

about community as it is about software, and Open 

Cirrus seeks to become the foundation offia largerfiopen 

cloud community. 

The Open Cirrus sites are working together to provide 

a single federated testbed, as opposed to each site 

building and operating a separate cluster, for three 

reasons: 

collaborating on a single, larger effort will achieve 

greater impact than participants could individually; 

testing in the different site environments will improve 

the quality of software and services; and 

pooling resources will improve efficiency because the 

sites will be sharing innovations. 

One measure of efficiency is management cost. Figure 2 

shows ballpark cost figures gleaned from the current Open 

Cirrus sites. While the costs of running a cloud infrastructure 

increase with the number of sites, the savings from 

sharing software development and operational methods 

reduces overall costs. For example, several participating 

organizations are prominent developers of the software 

components in the Open Cirrus service architecture. By 

sharing these new systems and the lessons learned in deploying 

them, all of the sites benefit. 

ARCHITECTURE, DESIGN, 

AND IMPLEMENTATION 

Several high-level architectural choices drove the Open 

Cirrus design: 

A 


aG 

F

A 


aG 

F 

Systems versus application-only research. In contrast 

to clusters such as Google/IBM, Microsoft Windows 

Azure, and Amazon EC2/S3, Open Cirrus enables 

research using physical machines in addition to virtualized 

machines. This requires provisioning of the bare 

metal, enabling root access to provisioned servers, 

providing isolation at the network level, and reclaiming 

access in case of fraudulent or erroneous behavior. 

Federated versus unified sites. In contrast to a unified 

architecture such as PlanetLab, Open Cirrus federates 

numerous sites with various hardware, services, and 

tools. The sites exist on different continents, under 

different regulations and subject to different privacy 

concerns. Commonality is enabled by Open Cirrus 

global services under development, such as global 

sign-on and monitoring. Some local services may 

vary across sites, but common practices and regulations 

will promote consistent administration and 

oversight. 

Data-center focus versus centralized homogeneous 

infrastructure. Compared to a centralized approach 

such as Emulab, Open Cirrus revolves around multiple 

data centers. This data-center focus enables 

independent research while sharing resources. It has 

implications for security, enforcing authorizations 

between users and individual sites, and integration 

with existing organizational regulations. 

Service architecture design 

Design of the Open Cirrus service architecture is guided 

by a desire to create a unified and coherent resource, 

rather than several completely disjoint clusters. 

Direct access to physical resources. Systems research is 

supported by allowing direct access to physical resources 

on the machine. For example, researchers can have root 

passwords, install kernel images, and access processors, 

chipsets, and storage. However, some resources, particularly 

network resources needed for proper isolation such 

as virtual local area network (VLAN) switch configurations, 

may be virtualized or unavailable. 

Similar operating environments. Given that multiple 

organizations with different practices manage the Open 

Cirrus sites, it’s infeasible for these sites to have identical 

operating environments. However, it’s possible to create 

similar operating environments by defining a minimum 

set of services that every site must offer. 

Support for near-production use. While supporting 

cloud computing system software research is Open Cirrus’ 

central mission, robust research often requires access to 

real-world use cases. Therefore, Open Cirrus sites strive 

to offer high-quality services to researchers who aren’t 

necessarily conducting research at the systems level. This 

use provides the workloads, traces, and testing needed for 

insights into real-world use. 

Annual cost per site in $K 

$1,800 

$1,600 

$1,400 

$1,200 

$1,000 

$800 

$600 

$400 

$200 

$0 

1 5 10 20 50 

Number of sites 

Global services available from any site. A small set 

of global services is available from any Open Cirrus site. 

Examples include the global sign-on authentication service, 

global monitoring, and a moderate-scale storage service for 

configuration files, intermediate results, or binaries. 

Service stack architecture 

A typical Open Cirrus site consists of foundation, utility, 

and primary domain services, as Figure 3 shows. 

Zoni. The foundation service for the software architecture 

is Zoni. Fundamentally, Zoni is the software 

component responsible for managing physical resources 

in the cluster and is crucial to providing users with baremetal 

server access to conduct software system research. 

This component provides five key functions: 

allocation of server nodes; 

isolation of node groups, called domains; 

provisioning of key software in a domain; 

out-of-band server management; and 

debugging of allocated nodes. 

(Shared) Innovation on OC 

OC development 

Running/managing OC 

System administration 

Invest./maint. for 1K cores 

Figure 2. Open Cirrus management costs: annual cost per 

site for dičerent numbers of sites. While the costs of running 

a cloud infrastructure increase with the number of sites, the 

savings from sharing software development and operational 

methods reduces overall costs. 

Zoni maintains a database of all available cluster resources. 

User requests, which are primarily for some 

number of nodes, are satisfied by allocating resources 

from that inventory. Users request allocations of physical 

nodes for various reasons. One common reason is to conduct 

controlled performance measurements; another is 

for experimentation with system software that involves 

aspects of networking—for example, developing a new 

cluster management system may require controlling a 

Dynamic Host Configuration Protocol (DHCP) server. 

Experiments that belong to this latter group must be iso- 

APRIL 2010 

A 


aG 

F 

37

A 


aG 

F 

38 

COVER FEATURE 

lated from other activities in the cluster to maintain both 

experimental integrity and cluster stability. Node allocations 

isolated from the rest of the cluster in this way are 

domains. The current implementation creates domains 

by programming the cluster switches to create VLANs. 

To bootstrap a domain for remote users, Zoni must 

also provide a mechanism for provisioning software onto 

at least one of the allocated nodes. This feature is currently 

implemented through a process based on Preboot 

Execution Environment (PXE) booting. However, even 

though the software may be provisioned properly, there 

is no guarantee that it will behave as intended or boot at 

all. Consequently, Zoni also provides users with out-ofband 

debugging and management facilities. Currently, the 

key debugging facility is remote console access, and the 

management facility consists primarily of remote power 

control. Both are provided through the Intelligent Platform 

Management Interface (IPMI). 

Primary domain services. Naturally, not all cluster 

users are interested in managing a Zoni domain; some are 

interested in developing higher-level services, and some 

are interested in simply using the services offered. To 

serve these last two user groups, one domain in each site 

is designated the primary domain and provides a stable 

set of services for production use. 

To support users working with very large data sets, a 

cluster storage system, in particular the Hadoop distributed 

file system (HDFS), is used to aggregate the storage 

of all the nodes in the domain. A key property of HDFS is 

that it supports location-aware computing—that is, the file 

system exports the location of data blocks in such a way 

that runtime services can schedule a computing task that 

processes a block on the node that contains that block. 

This capability reduces pressure on the cluster network. 

To support a diverse set of user needs, the recommended 

primary domain services include a virtual 

machine management (VMM) layer, which provides a convenient 

mechanism for allocating resources to various 

users and services. Hadoop, for example, implements a 

map/reduce programming paradigm and offers a popular 

runtime system for building cloud services. The Maui 

Cluster Scheduler combined with the Torque resource 

manager is another popular runtime system that provides 

COMPUTER 

Foundation services (Zoni) Site utility services Primary domain services 

Domain 

isolation 

(VLAN) 

Physical 

machine 

control 

(IPMI) 

Networking 

(DNS, 

DHCP) 

Provisioning 

(PXE) 

Monitoring 

(Ganglia) 

Power 

management 

Accounting 

and 

billing 

Attached 

storage 

(NFS) 

Domain 

isolation 

(DLS) 

Resource 

telemetry 

(RTS) 

Application frameworks 

(Hadoop, MPI, Maui/Torque) 

Virtual machine management 

(AWS-compatible systems 

such as Tashi and Eucalyptus) 

Cluster storage 

(HDFS) 

Figure 3. Open Cirrus site services. A typical site consists of foundation, utility, and primary domain services. 

a job submission service. By providing the two services 

through a disjoint set of virtual machines, the primary 

domain’s administrator can dynamically adjust the pool 

of resources available to each service according to need 

simply by adding or removing VMs. Of course, users 

whose jobs aren’t accommodated by one of the existing 

application framework services may request a private 

pool of VMs. 

Different sites may select any VMM service as long as 

it supports the EC2 interface from Amazon Web Services 

(AWS). Good candidates include Tashi and Eucalyptus. 

Tashi is an open source cluster management system for 

cloud computing on massive Internet-scale data sets being 

developed as an Apache Incubator project; it is designed 

specifically to complement Open Cirrus. While Tashi is 

similar to other systems that manage logical clusters of 

VMs, it was developed to support research in coscheduling 

computation, storage, and power. 

Coscheduling of computation and storage using the 

location information provided by storage systems such as 

Hadoop must overcome additional challenges in a virtualized 

environment. In particular, instances of the Hadoop 

runtime executing within VMs may be unable to correctly 

assess the relative distances to data blocks whose locations 

HDFS described with nonvirtualized location information. 

Two additional services may overcome this problem: The 

Data Location Service (DLS) is a clearinghouse for data 

location information independent of a storage mechanism, 

and the Resource Telemetry Service (RTS) provides 

a means to obtain an abstract distance measure between 

two location identifiers. 1 

Example. Figure 4 illustrates how Open Cirrus services 

fit together. In this example, the cluster is partitioned 

into four domains. From left to right, the first domain is 

used for low-level systems research, where researchers 

install their own OS kernels and run their own experimental 

codes and services. The second domain runs 

a VMM system that provides users with virtual clusters 

of VMs that share physical nodes and storage in 

the Zoni domain. Users build their own services and 

applications on top of these virtual clusters. The third 

domain provides a dedicated storage service that applications 

running on the second partition use. The fourth 

A 


aG 

F

A 


aG 

F 

domain offers a sandbox for developing the 

next version of the Tashi cluster management 

component. 

Site utility services. To manage an Open 

Cirrus site easily, many additional, less critical 

services are required. For example, a monitoring 

service such as Ganglia not only enables the site 

administrator to monitor the cluster’s health, it 

also facilitates collection of cluster operational 

data that may inform future research projects. 

Some conventional network file system storage 

is convenient for storing user scripts, small 

data sets, and small output files. Site utilities 

also include facilities for tracking resources 

consumed by users and managing the cluster’s 

power consumption. 

Open Cirrus federation 

The Open Cirrus testbed provides a unique opportunity 

for experimenting with issues involving federation in 

cloud computing. To support these experiments, developers 

are adding a small set of “global” services—common 

services that run at participating sites to provide a common 

infrastructure. 

Global sign-on. While users are granted access to each 

site separately, this authentication clearinghouse service 

enables users to access all granting sites with the same 

credentials. 

Global monitoring. By collecting the monitoring output 

from each site into a central location, administrators can 

quickly assess the health of the entire testbed, users can 

determine which sites are busy, and cross-site statistics 

can be archived for future research. 

User directories. Mounting a user directory in a 

common location at each site facilitates the distribution 

of small files, such as execution scripts and configuration 

files, for conducting experiments. 

Global storage. Similarly, some data sets may be made 

available across all sites, either through replication or 

some other distribution technology, as appropriate. 

Scheduling. Ultimately, users may find it convenient to 

submit a job without having to specify where the job must 

run. A global scheduling component could automatically 

dispatch jobs to the optimal cluster according to some 

criteria: load balancing, minimizing data movement, minimizing 

energy consumption, and so on. 

Resource tracking. As jobs are scheduled across multiple 

sites, the resources that users consume may need to 

be tracked and credits exchanged between sites based on 

resource consumption. 

RESEARCH USING OPEN CIRRUS 

Table 1 summarizes basic characteristics of the current 

Open Cirrus sites. More important than physical charac- 

Services 

Virtual cluster 

Systems research Tashi 

Zoni service 

Applications 

Virtual cluster 

Storage service Tashi development 

Network-isolated 

Zoni domains 

Physical cluster 

Figure 4. Example service hierarchy possible in Open Cirrus, with the 

cluster partitioned into four domains. 

teristics are the types of collaboration that Open Cirrus 

enables. 

Approximately 100 research projects at 10 sites use 

Open Cirrus at the systems and applications levels. 

Systems-level projects include robust adaptive routing 

over redundant layer-2 networks, data computation 

overlay for aggregation and analyses, and Open Cirrus 

and PlanetLab federation (at HP); use of optical switches 

to break data-center networking bottlenecks, power-aware 

workload scheduling, and optimizing service energy-delay 

product (at Intel); and applying RAID techniques to HDFS, 

pipelining data between map and reduce stages of Hadoop 

jobs to improve user interaction, and deploying log-analysis 

techniques to improve performance of Hadoop clusters 

(at Yahoo!). 

Applications-level projects include computerized 

language translation, astrophysics, graph data mining, 

and computer design simulation (at CMU); real-time 

streaming applications such as gesture recognition 

from video and simulation of material that changes its 

shape under software control (at Intel); DNA sequencing, 

search, and annotation (at ETRI); and continuously 

extracting knowledge from webpages, natural-language 

processing, large-scale graph algorithms, analysis 

of Wikipedia group dynamics, statistical machine 

translation, large-scale document analysis, statistical 

machine-learning algorithms, and computational sustainability 

(at Yahoo!). 

In most of these research projects, Open Cirrus offers 

the benefits of running experiments at scale; leveraging 

commonly shared stack, services, and best practices; 

and creating synergies across the layers. Systems-level 

research—for example, networking, power, and cooling—leverages 

applications to better understand the 

cloud load, while applications improve in performance 

and scalability using results from the underlying systems 

research. 

APRIL 2010 

A 


aG 

F 

39

A 


aG 

F 

40 

COVER FEATURE 

Site Cores Servers 

OPEN CIRRUS ECONOMIC MODEL 

The emergence of each individual site in Open Cirrus 

and the expected growth of the federation are driven by 

the economy in today’s cloud computing environment. 

Explicit break-even points for the choice between renting 

versus owning a cloud infrastructure implicitly justify 

Open Cirrus’ economic rationale. 

Single site 

Consider a medium-sized organization—for example, a 

start-up or a university department—seeking to provide a 

Web service to a client population. The service will run in 

a cloud, accessing stored data and consuming CPU cycles. 

Suppose this service is identical to the UIUC Open Cirrus 

site: 128 servers (1,024 cores) and 524 Tbytes. Should the 

organization rent the infrastructure from a cloud provider, 

such as EC2 (Elastic Compute Cloud) and S3 (Simple Storage 

Service) from Amazon Web Services, or should it buy 

and maintain a cloud? 

At average AWS rates of US$0.12 per Gbyte/month 

and $0.10 per CPU-hour, renting the cloud infrastructure 

would incur a monthly storage cost of 524 × 1,000 

× $0.12, or $62,880; the total monthly cost would be 

$62,880 + 1,024 × 24 × 30 × $0.10 = $136,608. In the 

case of owning a cloud, amortized monthly costs would 

be split among the hardware, power, and network 45 

percent, 40 percent, and 15 percent, respectively. 2-4 If the 

service lifetime is M months, it would incur a monthly 

storage cost, assuming $300 1-Tbyte disks and scaling 

for power and networking, of 524 × $300/0.45/M, or 

$349,333/M; the total monthly cost, based on actual systems 

cost and the salary of one system administrator for 

about 100 servers, 3,4 would be $700,000/0.45/M + $7,500, 

or $1,555,555/M + $7,500. 

COMPUTER 

Public 

partition 

(servers) 

Table 1. Characteristics of current Open Cirrus sites. 

Memory 

size 

(Tbytes) 

Storage 

size 

(Tbytes) Spindles 

Network data 

rate Focus 

CMU 1,165 159 50 2.40 892 784 1 Gbps Tashi, distributed file systems, 

applications/data sets 

ETRI 1,024 256 200 0.50 128 256 1 Gbps Large data sets, cloud infrastructure 

HP 1,024 256 178 3.30 632 1,152 10 Gbps internal; 

1 Gbps x-rack 

Networking, federation 

IDA 2,400 300 100 4.80 59+ 600 1 Gbps Applications based on Hadoop, Pig 

Intel 1,364 198 198 1.77 610 746 1 Gbps Tashi, Zoni, MPI, Hadoop 

KIT 2,656 232 128 10.00 280 192 40 Gbps Applications with high throughput 

MIMOS 1,024 167 16 0.50 36 258 1 Gbps Platform, tools, testing, security 

UIUC 1,024 128 64 2.00 524 258 1 Gbps Data sets, cloud infrastructure 

RAS 1,136 142 600 9.10 36 142 1 Gbps Hadoop, Tashi, Zoni, Pig, MPI 

Yahoo! 3,200 480 400 2.40 1,200 1,600 1 Gbps Hadoop, Pig 

The break-even point for storage would be $349,000/M 

< $62,880, or M > 5.55 months; the overall break-even 

point would be $1,555,555/M + $7,500 < $136,608, or 

M > 12 months. Thus, if the service runs for more than 12 

months, owning the cloud infrastructure is preferable to 

renting it. Similarly, it’s better to own storage if you use it 

for more than six months. 

Clouds are typically underutilized. 2 With X percent 

resource utilization, the break-even time becomes 

12 × 100/X months. Given the typical hardware lifetime 

of 36 months, the break-even resource utilization is 

12 × 100/X < 36, or X > 33.3 percent. Even at the 

current 20 percent CPU utilization rates observed in industry, 

storage utilization greater than 47 percent would 

make ownership preferable, as storage and CPU account 

evenly for costs. 

Federated sites 

Federation can help absorb overloads due to spikes—for 

example, at conference deadlines—or underprovisioning. 2,4 

Figure 5 plots the costs incurred by a single underprovisioned 

cloud for three options: offloading only to AWS, 

offloading to five federated clouds and AWS, and offloading 

to 49 federated clouds and AWS. 

A federation of six sites can defer costs up to 250 percent 

overload, while with 50 sites the break-even point is roughly 

2,500 percent. (This assumes that other sites are utilized 

50 percent and not idle; otherwise, the break-even point 

would be 500 percent and 5,000 percent, respectively.) The 

detailed data and spreadsheet for this calculation are available 

at http://opencirrus.org. Note that this is only a starting 

step—the calculation can be expanded by accounting for 

the economic costs of disasters, such as massive failure, 

project cancellation, and start-up time. 

A 


aG 

F

A 


aG 

F 

RELATED WORK 

We can broadly divide existing cloud computing testbeds 

into those that mainly support applications research 

and those that can support systems research. Table 2 compares 

the most prominent testbeds. 

The Google/IBM cluster (www.google.com/intl/en/ 

press/pressrel/20071008_ibm_univ.html), TeraGrid, 5 and 

Microsoft Windows Azure platform (www.microsoft. 

com/windowsazure) all focus on supporting computing 

applications research. Thus, these testbeds don’t enable 

access to bare-metal hardware or root access to the OS; 

instead, services such as MPI and Hadoop are installed 

to ease access to the resources. For example, the Google/ 

IBM cluster is configured with the Hadoop service and 

targets data-intensive applications research such as largescale 

data analytics. TeraGrid is a multisite infrastructure 

mainly used for scientific research, and Microsoft provides 

a Windows Azure interface but no access to systems-level 

data. 

The Open Cloud Testbed (www.opencloudconsortium. 

org) focuses on cloud computing middleware research and 

is currently configured as a smaller-scale testbed with four 

32-node sites. 

Testbeds such as PlanetLab, 6 Emulab, 7 DETER (cyber- 

DEfense Technology Experimental Research), 8 ___________________________ 

___ 

and 

Amazon EC2 (http://aws.amazon.com) are designed to support 

systems research but with diverse goals. 

PlanetLab consists of a few hundred machines spread 

around the world and is mainly designed to support 

Characteristics Open Cirrus 

Type of 

research 

Systems and 

services 

Approach Federation of 

heterogeneous 

data centers 

Participants CMU, ETRI, HP, 

Intel, IDA, KIT, 

MIMOS, RAS, 

UIUC, Yahoo! 

Monthly cost in $K 

100,000 

10,000 

1,000 

100 

Table 2. Comparison of cloud computing testbeds. 

Google/ 

IBM cluster TeraGrid PlanetLab Emulab 

Dataintensive 

applications 

Cluster 

supported 

by Google 

and IBM 

Google, 

IBM, MIT, 

Stanford 

Univ., 

Univ. of 

Washington 

Distribution 10 sites Centralized—one 

data center 

in Atlanta 

Scientific 

applications 

Multisite 

heterogeneous 

clusters for 

supercomputing 

Many 

universities 

and 

organizations 

11 partners 

in US 

Existing data center 

Open Cirrus 6 

Open Cirrus 50 

10 

100 200 300 400 500 600 700 800 900 1,000 

Utilization (percent) 

Figure 5. Costs incurred by a single underprovisioned cloud 

for three options: oćoading only to Amazon Web Services 

(existing data center), oćoading to ve federated clouds 

(Open Cirrus 6) and AWS, and oćoading to 49 federated 

clouds (Open Cirrus 50) and AWS. 

wide-area networking and distributed systems research. 

Although it doesn’t provide access to bare-metal hardware, 

it does provide root access to the OS through a lightweight 

virtualization similar to FreeBSD jails. 

Emulab, the original Zoni service, is a single-site testbed 

where each user can reserve a certain number of machines 

(typically a few tens) and get exclusive access to bare-metal 

hardware. Emulab also provides mechanisms to emulate 

different network characteristics. Open Cirrus provides 

Emulab-like flexibility for systems research with federation 

and heterogeneity, which are crucial for cloud computing. 

Systems and 

services 

Nodes 

hosted by 

research 

institution 

Many 

universities 

and 

organizations 

More than 

700 nodes 

worldwide 

Open Cloud 

Testbed 

Systems Interoperability 

across 

clouds using 

open APIs 

Singlesite 

cluster 

with 

exible 

control 

Univ. of 

Utah 

More 

than 300 

machines 

Multisite heterogeneous 

clusters 

Amazon 

EC2 

Commercial 

use 

Raw access 

to virtual 

machines 

APRIL 2010 

LANL 

cluster 

Systems 

Reuse of 

LANL’s 

retiring 

clusters 

Four centers Amazon CMU, 

LANL, NSF 

480 cores distributed 

in 

four locations 

Several 

unified 

data 

centers 

Thousands 

of 

older but 

still useful 

nodes at 

one site 

A 


aG 

F 

41

A 


aG 

F 

42 

COVER FEATURE 

The DETER testbed is an installation of the Emulab 

software and is mainly used for security research—for 

example, collecting a large-scale worm trace. Consisting 

of two heterogeneous sites, DETER may be viewed as a 

federated Emulab installation. However, the two sites are 

tightly coupled: The controller resides in one site and controls 

physical resources in both sites. In Open Cirrus, all 

sites are loosely coupled. 

Amazon EC2 provides VMs on a pay-as-you-go basis. 

Although it allows complete control over the VMs, users 

can’t control network resources, reducing its flexibility as 

a systems research testbed. 

Finally, CMU’s Garth Gibson is leading an effort to recycle 

Los Alamos National Laboratory’s retiring clusters 

(typically with a few thousand machines) by making them 

available for systems research. 

Other related cloud computing efforts such as Reservoir 

(http://sysrun.haifa.il.ibm.com/hrl/reservoir) and 

RightScale (www.rightscale.com) aren’t testbeds per se. 

Reservoir (Resources and Servers Visualization without 

Barriers) is an EU-funded grid computing project that enables 

massive-scale deployment and management of IT 

services across administrative domains. RightScale is a 

cloud services management platform. 

Open Cirrus offers unique opportunities for 

cloud computing research that no existing 

alternatives offer. It federates heterogeneous 

sites, systems and applications research, and 

data sets. In addition, it provides an open 

source service stack with nonproprietary APIs for cloud 

computing. And through shared innovation, Open Cirrus 

will have a greater impact on research communities 

around the globe. 

While working on Open Cirrus during the past year, 

we realized the value of common stacks and services, 

but even more, we came to appreciate the benefits of a 

research community working together toward the same 

goals. Heterogeneity of the individual sites has contributed 

to the diversity of solutions and has strengthened our approaches, 

even though it does make global services more 

complex to develop, deploy, and maintain. 

Future work on Open Cirrus will revolve around 

increased use across the sites. In particular, we’re exploring 

applications that can be used on multiple sites 

to increase their performance, scale, and reliability. Another 

area of interest is standards, but we need more 

experience in using stacks and services before exploring 

standardization. 


Partial funding for the Open Cirrus UIUC site was provided by 

NSF. Funding for the Open Cirrus CMU/PDL site was provided 

by Intel, NSF, ARO, and the PDL Consortium. Open Cirrus is 

COMPUTER 

a trademark of Yahoo! Inc. Several people made significant 

contributions to Open Cirrus, including A. Chien, R. Gass, 

K. Goswami, C. Hsiung, J. Kistler, M. Ryan, C. Whitney, and 

especially J. Wilkes, who was instrumental in formulating 

the original Open Cirrus vision as well as the concept of a 

management layer like Zoni. 

References 

1. M.A. Kozuch et al., “Tashi: Location-Aware Cluster Management,” 

Proc. 1st Workshop Automated Control for 

Datacenters and Clouds (ACDC 09), ACM Press, 2009, pp. 

43-48. 

2. M. Armbrust et al., “Above the Clouds: A Berkeley View 

of Cloud Computing,” tech. report UCB/EECS-2009-28, 

EECS Dept., Univ. of California, Berkeley, 2009; www.eecs. 

________________________________ 

berkeley.edu/Pubs/TechRpts/2009/EECS-2009-28.pdf. 

3. J. Hamilton, “Internet-Scale Service Efficiency,” keynote 

at 2nd Large-Scale Distributed Systems and Middleware 

Workshop (LADIS 08), 2008; http://mvdirona.com/jrh/ 

TalksAndPapers/JamesRH_Ladis2008.pdf. 

_________________________ 

4. A. Greenberg et al., “The Cost of a Cloud: Research Problems 

in Data Center Networks,” ACM SIGCOMM Computer 

Communication Rev., Jan. 2009, pp. 68-73. 

5. C. Catlett et al., “TeraGrid: Analysis of Organization, 

System Architecture, and Middleware Enabling New Types 

of Applications,” High Performance Computing and Grids in 

Action, L. Grandinetti, ed., IOS Press, 2007, pp. 225-249. 

6. L. Peterson et al., “A Blueprint for Introducing Disruptive 

Technology into the Internet,” ACM SIGCOMM Computer 

Communication Rev., Jan. 2003, pp. 59-64. 

7. B. White et al., “An Integrated Experimental Environment 

for Distributed Systems and Networks,” ACM SIGOPS Operating 

Systems Rev., winter 2002, pp. 255-270. 

8. T. Benzel et al., “Design, Deployment, and Use of the 

DETER Testbed,” Proc. DETER Community Workshop Cyber 

Security Experimentation and Test, Usenix Assoc., 2007; 

__________________________________ 

www.isi.edu/deter/docs/200708-usecdw-deter-design- 

_______ deploy.pdf. 

Arutyun I. Avetisyan is deputy director of the Institute for 

System Programming of the Russian Academy of Sciences 

(ISP RAS), RAS’s representative in Open Cirrus. Contact 

him at __________ 

arut@ispras.ru. 

Roy Campbell is Sohaib and Sara Abbasi Professor in the 

Department of Computer Science at the University of Illinois 

at Urbana-Champaign (UIUC). Contact him at rhc@ 

________ illinois.edu. 

Indranil Gupta is an associate professor in the Department 

of Computer Science at UIUC, where he also heads 

the Distributed Protocols Research Group. Contact him at 

____________ 

indy@illinois.edu. 

Michael T. Heath is a professor and Fulton Watson Copp 

Chair in the Department of Computer Science at UIUC, 

where he’s also director of the Computational Science and 

Engineering Program. Contact him at ____________ 

heath@illinois.edu. 

Steven Y. Ko is a postdoctoral research associate in the Department 

of Computer Science at Princeton University who 

worked on Open Cirrus as a PhD student at UIUC. Contact 

him at steveko@cs.princeton.edu. 

_________________ 

A 


aG 

F 

____

A 


aG 

F 

Gregory R. Ganger is a professor in the Department of 

Electrical and Computer Engineering at Carnegie Mellon 

University (CMU), where he’s also director of the Parallel 

Data Lab. Contact him at ______________ 

ganger@ece.cmu.edu. 

Michael A. Kozuch is a principal researcher for Intel Labs 

Pittsburgh working in the area of computer architecture 

and system software. Contact him at _____________ 

michael.a.kozuch@ 

intel.com. 

David O’Hallaron is the director of Intel Labs Pittsburgh 

and an associate professor in the Department of Computer 

Science and the Department of Electrical and Computer 

Engineering at CMU. Contact him at ______________ 

david.ohallaron@intel. 

___ com. 

Marcel Kunze heads the Cloud Computing research group 

at the Steinbuch Centre for Computing, Karlsruhe Institute 

of Technology, Germany, and is a technical lead in Open 

Cirrus. Contact him at _______________ 

marcel.kunze@kit.edu. 

Thomas T. Kwan is director of research operations at 

Yahoo! Labs, where he manages multiple cloud computing 

research partnerships for Yahoo!. Contact him at ______ tkwan@ 

yahoo-inc.com. 

Kevin Lai is a research scientist in the Social Computing 

Lab at HP Labs and has worked on various projects at the 

intersection of operating systems and incentive engineering. 

Contact him at klai@hp.com. 

________ 

Martha Lyons is a Distinguished Technologist at HP Labs 

focused on incubating and delivering innovative capabilities 

to HP’s customers. Contact her at _______________ 

martha.lyons@hp.com. 

Dejan S. Milojicic is a scientist and senior researcher manager 

in the Strategy and Innovation Office at HP Labs and director of 

Open Cirrus. Contact him at _______________ 

dejan.milojicic@hp.com. 

Hing Yan Lee is program director of the National Grid 

Office at the Infocomm Development Authority (IDA) of 

Singapore. Contact him at _________________ 

lee_hing_yan@ida.gov.sg. 

Yeng Chai Soh is a professor in the School of Electrical and 

Electronic Engineering at Nanyang Technological University, 

Singapore, and is also affiliated with IDA. Contact him 

at _____________ 

eycsoh@ntu.edu.sg. 

Ng Kwang Ming is director of the Grid Computing Lab at the 

Malaysian Institute of Microelectronic Systems (MIMOS). 

Contact him at ________________ 

kwang.ming@mimos.my. 

Jing-Yuan Luke is a staff engineer in the Grid Computing 

Lab at MIMOS. Contact him at _____________ 

jyluke@mimos.my. 

Han Namgoong is a director in the Cloud Computing Research 

Department of the Software Research Laboratory at 

the Electronics and Telecommunications Research Institute 

(ETRI), South Korea. Contact him at nghan@etri.re.kr. 



COMPUTING LIVES 

www.computer.org/annals/computing-lives 

___________________________________________________ 

___________ 

The “Computing Lives” podcast series of selected articles from the IEEE Annals of the History 

of Computing cover the breadth of computer history. This series features scholarly accounts 

by leading computer scientists and historians, as well as fi rsthand stories by computer pioneers. 

APRIL 2010 

A 


aG 

F 

43

A 


aG 

F 

44 

COVER FEATURE 

Edward Walker, University of Texas at Austin 

Storage clouds are online services for 

leasing disk storage. A new modeling 

tool, formulated from empirical data 

spanning many years, lets organizations 

rationally evaluate the benefit of using 

storage clouds versus purchasing hard 

disk drives. 

Hard disk drives provide storage for a broad range 

of devices, from mobile phones to large IT 

server farms. In 2008, approximately 590 million 

hard disk drives were shipped worldwide, 1 

largely driven by the vast amount of information 

our digital society is generating. For example, a report 

from researchers at the University of California, Berkeley, 

estimates that 92 percent of the five exabytes (10 18 ) of 

new information created in 2002 was stored on magnetic 

media, primarily hard disk drives. 2 

At the same time, the business of selling infrastructure 

as a service through the Internet is growing. This technology 

trend, also known as cloud computing, lets individuals 

and organizations outsource their IT requirements to 

remote data centers, paying for only what they use. The 

cloud computing industry’s worth was estimated at more 

than $16 billion in 2008, and it’s expected to grow to $42 

billion by 2012. 3 

Several online services currently lease storage infrastructure. 

These storage clouds let anyone with a credit 

card purchase storage capacity online, paying a monthly 

fee for the storage they use. __________ 

Amazon.com’s S3 service 

(http://aws.amazon.com/s3), for example, lets users store 

COMPUTER 

TO LEASE OR NOT 

TO LEASE FROM 

STORAGE CLOUDS 

Walter Brisken and Jonathan Romney, National Radio Astronomy Observatory 

arbitrary objects of up to 5 Gbytes each in their online 

storage repositories. The S3 service uses a tiered pricing 

structure, with storage getting cheaper as more is used. As 

of late 2008, Amazon.com reported that users stored more 

than 40 billion objects in the S3 service. 4 

With the significant growth of society’s storage requirements, 

and the availability of pay-per-use online storage 

services, when should a consumer consider using storage 

clouds? We focus on the problem of resolving this buy-orlease 

storage decision. 

BUY OR LEASE 

An organization’s decision to buy or lease long-term 

assets is often a critical dilemma. The literature describes 

numerous models to assist firms in resolving this acquisition 

decision. 5-7 However, there is a significant lack of work 

in applying these models to compare the cost of purchasing 

versus leasing storage from cloud services. Important 

studies have examined the cost of performing scientific 

experiments using the Amazon.com cloud services, 8,9 but 

these prior works are difficult to generalize outside the 

scope of the applications used in the studies. 

We believe our proposed model for evaluating the benefits 

of purchasing versus leasing storage from a cloud 

service is the first reported in the literature, offering a 

method for deriving a reasonable estimate of the best possible 

outcomes from the alternative investment choices. 

Furthermore, our model, which we formulated using 

empirical disk price data we’ve been collecting weekly 

from Pricewatch.com over many years, is agnostic to the 

application using the storage cloud. Our empirical data 

tracks the lowest prices for serial advanced technology 

attachment (SATA) disk drives for sale online since the 


A 


aG 

F

A 


aG 

F 

technology’s introduction in 2003. This data is significant 

because SATA is the predominant drive technology used 

in desktop and nearline systems. 

Most organizations use financial models as only one data 

point in deciding whether to buy or lease an asset. Other 

factors also influence their decision. 7 However, our proposed 

model bridges a critical gap in the literature and, more 

importantly, the market. Our model helps users understand 

the premium they’re paying for what seems to be an arbitrarily 

priced online storage market. Our research agenda 

aims to provide a quantitative framework for rationally reasoning 

about the cost of leasing infrastructure from cloud 

services, and extends our prior work on leasing from compute 

clouds. 10 Importantly, the systematic understanding of 

the real cost of leasing assets from cloud services lets users 

make rational decisions with innovative pricing structures. 

This promotes market transparency and ultimately supports 

a competitive product market. 

PRELIMINARIES 

A key principle in economic finance is the time value 

of money. Basically, this principle states that an investor 

always prefers to receive some fixed amount of money 

today rather than in the future. Hence, when making 

buy-or-lease decisions, investors often compare future 

cash flows in an investment over time, discounted to their 

present value by some interest rate. 5-7 The discounting 

interest rate used reflects the risk involved in raising the 

capital needed to invest. 

In equation form, the simplified standard capitalbudgeting 

format for calculating a purchased asset’s net 

present value (NPV) is as follows: 

NPV P = 

N 

 

T = 0 

P P C T T 

(1+ I ) K T 

S 

+ E , N (1+ I ) K 

where P is the annual profit resulting from the purchased 

T 

P asset in year T; C is the asset’s expected annual operat- 

T 

ing cost at year T; I is the firm’s cost of capital, defined 

K 

as the interest rate of its outstanding debt used to finance 

the purchase; 11 N is the asset’s productive life in years; S is 

the asset’s salvage value after N years; and E is the asset’s 

purchase (capital) cost. 

Similarly, the equation for calculating a leased asset’s 

NPV is as follows: 

NPV L = 

L P C T T 

( ) T 

N 

 

T = 0 

1+ I K 

 

LT ( ) T 

N 

 

T = 0 

1+ I R 

L where C is the leased asset’s expected annual operat- 

T 

ing cost at year T; L is the lease payment at year T; and 

T 

I is the interest rate for financing the lease payments. In 

R 

this formulation, the lease’s financing rate is generally 

regarded as smaller than the cost of capital, I , because of 

K 

, 

Table 1. Parameter terms in the buy-or-lease decision model. 

Term Description 

Cost of electric utility ($/kilowatt hour) 

Size of purchased disk drives (Gbytes) 

Proportional difierence between human efiort in main - 

taining a purchased versus a leased storage infrastructure 

ϒ Used disk depreciation factor on salvage ([0.0, 1.0]) 

C Disk controller unit cost ($) 

H Annual human operator salary ($) 

T 

I Risk-free interest rate (percent) 

F 

Κ Current per-Gbyte storage price ($/Gbyte) 

LT Pc PD VT Expected annual per-Gbyte lease payment ($/Gbyte/year) 

Disk controller power consumption (kW) 

Disk drive power consumption (kW) 

Expected storage requirement in year T (Gbytes) 

the involved payment structure’s predictability. 5 

With this NPV formulation for asset purchase and lease, 

investors can make the buy-or-lease decision using the 

following criteria: If the incremental NPV (NPV) 0 

buy; if NPV < 0 lease, where NPV = NPV P – NPV L . 

DECISION MODEL 

We derive a buy-or-lease decision model that calculates 

the comparative NPVs from storage purchase versus 

lease. Our model accounts for the expected capital expenditure 

from SATA hard disk drive purchase, replacement, 

and end-of-use salvage. It also accounts for the expected 

operational expenditure for utility consumption and 

human operator cost. The “Derivation of the Decision 

Model” sidebar describes how we derived our decision 

model. We summarize it as follows: If NPV 0 buy; if 

NPV < 0 lease, where 

C − E + L T T T 

ΔNPV = 

( 1+ I F ) T 

N 

S 

∑ 

+ − C 

N 

T = 0 

( 1+ I F ) 

S =γ ∗ Ω∗⎡⎢ V ⎤ ∗ K ∗e T ⎥Ω −0.438T 

C =−ρ∗H − 365 ∗24 

T T ( ⎡⎢ ⎤⎥Ω ) 

( )∗δ ∗ P + P ∗ V C D T 

( ⎡⎢ ⎤ − ⎡⎢ V ⎤ ⎥Ω T −1 ⎥Ω )∗Ω∗K ∗e −0.438T 

E T = 1.03∗ V T 

Table 1 lists the model parameters. We assume ⎡V⎤ fl is 

an operator returning the minimum number of fl-sized 

disk drives needed to store V Gbytes of data. The derived 

terms S, C T , and E T represent the expected end-of-life disk 

salvage value, the operating cost in year T, and the capital 

cost in year T, respectively. 

EXAMPLE APPLICATION 

We used our decision model to evaluate the advantages 

of purchasing versus leasing storage from a hypothetical 

storage cloud vendor. 

. 

APRIL 2010 

A 


aG 

F 

(1) 

45

A 


aG 

F 

46 

COVER FEATURE 

We calculate the incremental net present value (ΔNPV) as 

follows: 

ΔNPV = NPV − NPV P L 

⇒ 

PT − C P 

T 

( 1+ IK ) T 

N 

L 

S 

P − C L T T 

T 

∑ + − E − + 

T T 

T =0 ( 1+ IK ) ( 1+ IK ) ( 1+ IR ) T 

N 

N 

∑ ∑ 

T =0 

T =0 

⇒ 

C L P − CT 

T 

( 1+ IK ) T 

N 

S 

L T 

∑ + + 

N 

T =0 ( 1+ IK ) ( 1+ IR ) T 

N 

∑ − E. 

T =0 

Assuming that profit from storage is equal in both the buy and 

lease cases—that is, using 1 Tbyte of storage from a purchased 

disk or from a storage cloud results in the same level of 

productivity—the above simplification results in the removal of 

the profit term P T . 

Capital cost 

Three components make up the cost of purchasing storage E: 

The consumer needs a disk controller to house the purchased 

disks. 

The consumer purchases disk drives in blocks based on 

increasing storage needs. 

The consumer must periodically replace disks due to 

failure. 

These components contribute to the future cash flow E as 

follows: 

E = 

COMPUTER 

DERIVATION OF THE DECISION MODEL 

( ( ⎡⎢ V ⎤ − ⎡⎢ V ⎤ T ⎥Ω T −1⎥Ω 

)∗Ω+R T )∗G T 

( ) T 

E T 

1+ I K 

⇒ E = + C 

T 

( 1+ IK ) 

E = V T ( ( ⎡⎢ ⎤ − ⎡⎢ V ⎤ T ⎥Ω T −1⎥Ω 

)∗Ω+R T )∗G , T 

where Ω represents the hard disk drive size in Gbytes; V T is the 

storage requirement in Gbytes at year T; ⎡V T ⎤ Ω is an operator that 

returns the minimum number of Ω-sized disk drives needed to 

store V T ; G T is the predicted cost per Gbyte of disk storage at year 

T; R T is the disk replacement in Gbytes at year T; and C is the disk 

controller cost. 

The important insight in this formulation is that the capital 

cost isn’t all incurred at the start of the project, unlike in traditional 

NPV models. Rather, it’s a time-varying formula in which 

users can grow their storage systems as their requirements 

evolve. We modify ΔNPV to reflect this growth in capital cost: 

ΔNPV = 

N 

∑ 

T =0 

S 

+ 

1+ IK L P C − CT − ET 

T 

( ) 

( ) T 

1+ I K 

Operating cost 

− C 

LT + 

N 

( ) T 

N 

∑ 

T =0 

1+ I R 

P We estimate the operating cost of purchased storage, C , by T 

calculating the electric utility cost associated with running the 

+ C 

disk controller and the disk units, plus the cost of a human 

operator to manage the system/data. These cost components 

P contribute to C as follows: 

T 

P C = (365⋅24)∗δ ∗ PC + P T 

( ∗⎡⎢ V ⎤ D T ⎥Ω )+α∗H , T 

where δ is the utility cost ($/kilowatt per hour); P is the control- 

C 

ler’s power requirement in kWs; P is the power requirement in 

D 

kWs per disk drive; and α is the proportion of the human operator 

cost, H , required to maintain the system/data at year T. 

T 

L The operating cost for leased storage, C , only includes the 

T 

cost of a human operator to manage the data. Thus, we calculate 

L L 

C as CT = β∗HT , where β is the proportion of the human opera- 

T 

tor cost required to maintain the data on the leased storage at 

year T. Substituting ρ for (α – β), we can modify ΔNPV to reflect the 

operating costs as follows: 

ΔNPV = 

N 

∑ 

T =0 

−ρ∗H T − 365⋅24 

( )∗δ ∗( P + P ∗⎡⎢ V ⎤ C D T ⎥Ω )− ET ( ) T 

L T 

1+ I K 

S 

− C + + 

. 

N 

( 1+ IK ) ( 1+ IR ) T ∑ 

T =0 

Approximating the best outcomes 

N 

Consumers often find it difficult to estimate their capital cost 

and the lease financing rates available to them. We therefore suggest 

an approximate method of evaluating the buy-or-lease 

decision by deriving the best possible outcome in both the purchase 

and lease cases. 

From the well-known economic Law of One Price, 1 we can 

derive the upper bounds for NPV P and NPV L by substituting I K and 

I R with the risk-free interest rate I F . The “Corollary for Estimating 

the Upper-bound Present Value” sidebar explains this in more 

detail. In turn, we can estimate the risk-free interest rate, I F , from 

the published return on an instrument such as government treasury 

bills. We can therefore derive an approximation of the 

decision criteria using these best NPVs, letting us further simplify 

the currently derived version of ΔNPV to: 

ΔNPV = 

N 

∑ 

T =0 

( )∗δ ∗( P + P ∗⎡⎢ V ⎤ C D T ⎥Ω )− E + L T T 

( ) T 

−ρ∗H T − 365∗24 

S 

− C + 

. 

( 1+ IF ) N 

Disk price trends 

1+ I F 

We haven’t yet provided a function for the term G , which we need 

T 

to calculate E and to estimate the cost of disk storage at year T. 

T 

Since April 2003, we’ve been collecting SATA disk price data 

from Pricewatch.com on a weekly basis. We collected price data 

for all available disk drive sizes each week—250 Gbytes, 500 

Gbytes, and so on—regardless of manufacturer or model. Figure 

A plots the 10 lowest prices each week for SATA disk drives across 

all the available drive sizes. Approximating the observed exponential 

trend line using regression analysis, we obtain the formula 

′ − 0.0012 X G = 1.2984e . 

X 

A 


aG 

F

A 


aG 

F 

′ The function G predicts the cost per Gbyte 

X 

of SATA disk storage X days from 20 April 2003, 

′ with G = 1.2984 when X = 0. We can therefore 

X 

approximate the function G by assuming that 

T 

the future disk price trend conforms to the 

equation K ∗ eC ∗ T , where K represents the lowest 

storage price per Gbyte available to the consumer 

at T = 0; and T represents the number of 

years in the future. We therefore derive G as G T T 

= K ∗ e –0.0012∗ 365 ∗ T⇒ G = K ∗ e T –0.438∗ T . 

Disk replacement rates 

Any realistic cost model for disk storage ownership 

must estimate the disk replacement cost. 

A recent large-scale study of disk failures measured 

the annualized replacement rate (ARR) of 

disk drives in real data centers. 2 The study 

observed ARRs in the range of 0.5 to 13.5 percent, 

with the most commonly observed ARRs in 

the 3 percent range. 

In our model, we approximate R T with this 

empirical approximation of the disk replacement 

rate by using the formula R T = 0.03 * Ω * 

⎡V T ⎤ Ω . In this formula, the constant 0.03 represents the observed 

3 percent disk replacement rate. 2 Thus, we can simplify E T to: 

–0.438∗ T 

E = (( ⎡V ⎤ – ⎡V ⎤ ) ∗ Ω + 0.03 ∗ Ω ∗⎡V⎤ ) ∗ K ∗ e T T Ω T – 1 Ω T Ω 

⇒ E = (1.03 ∗⎡V⎤ – ⎡V ⎤ ) ∗ Ω ∗ K · e T T Ω T – 1 Ω –0.438∗ T . 

Disk salvage value 

$/GByte 

We assume a hard disk drive can be sold in the used market for 

some salvage value at the end of its life. To predict this salvage 

value, we leverage the future disk price prediction formula, discounting 

the predicted price by some depreciation factor, γ, in the 

We assume this vendor lets users purchase raw disk 

storage over the Internet. We’re only interested in the storage 

cloud’s pricing structure for the end user, and we don’t 

assume any particular service access technology. Table 2 

shows the assumed tiered monthly pricing structure. For 

illustrative purposes, we don’t consider data uploading or 

downloading costs. 

Single-user computers 

Single-user computers make up the vast majority of all 

shipped disk storage capacity. 2 For our study, we assume 

the same user owns and operates the single-user computer. 

This user’s storage requirement grows at a moderate rate 

of 100 Gbytes per year. Therefore, regardless of where 

data is stored, we assume the level of effort in managing 

the system/data is approximately equal in both cases 

because of the low storage volume involved—that is, 

= 0. We assume the user must purchase a new disk 

controller (C = $1,000), specified to consume 0.5 kW of 

power. Also, as storage is required, the user will purchase 

2.5 

2.0 

1.5 

1.0 

0.5 

y = 1.2984e – 0.0012 X 

Disk price 

Exponential (disk price) 

0 200 400 600 800 1,000 1,200 1,400 1,600 1,800 2,000 

Start date: 20 April 2003 

Days 

Figure A. Weekly SATA disk price data collected from Pricewatch.com from 

20 April 2003 to 19 August 2008. 

range [0, 1]. In equation form, this gives us the salvage value S = γ 

* Ω * ⎡V T ⎤ Ω ∗ K ∗ e –0.438 ∗ T , simplifying into ΔNPV as shown by 

Equation 1 in the main text. 

References 

1. O.A. Lamont and R.H. Thaler, “Anomalies: The Law of One 

Price in Financial Markets,” J. Economic Perspective, vol. 17, no. 

4, 2003, pp. 191-202. 

2. B. Schroeder and G. Gibson, “Disk Failures in the Real World: 

What Does an MTTF of 1,000,000 Hours Mean to You?” Proc. 

5th Usenix Conf. File and Storage Technologies (FAST 07), 

Usenix Assoc., 2007, pp. 1-16. 

COROLLARY FOR ESTIMATING THE 

UPPER-BOUND PRESENT VALUE 

We derive the following corollary from the economic Law of 

One Price, which states, “In an efficient market, identical 

goods will have only one price.” 

Corollary 1. The upper bound of the present values for the purchase 

and lease price equilibrium in an efficient market is derived 

from the risk-free interest rate, I . F 

Informal proof: In an efficient market, an arbitrager has no 

opportunity to make a risk-free profit. Assuming an efficient market 

in which financing at the risk-free rate is possible for a subset of 

agents, the proof is by contradiction. Thus, if the present value of 

the lease price is higher than that derived from the risk-free rate, I , F 

an arbitrager can purchase disks from monies borrowed at the riskfree 

rate, lease at this higher price, and pocket the risk-free profit. 

Also, if the present value of the purchase price is higher than that 

derived from the risk-free rate, I , an arbitrager can purchase a disk 

F 

from monies borrowed at the risk-free rate, sell the disk at some 

future instance at this higher price, and pocket the risk-free profit. 

APRIL 2010 

A 


aG 

F 

47

A 


aG 

F 

48 

Year 

(T) 

ΔNPV (thousands of dollars) 

–50 

COVER FEATURE 

0 

–100 

–150 

–200 

–250 

–300 

0 2 4 6 

Operational years 

8 10 12 

Figure 1. The calculated NPV values for a medium-size enterprise deciding 

whether to buy or lease storage with life expectancy from 0 to 10 years. 

ΔNPV (thousands of dollars) 

150 

100 

50 

0 

–50 

–100 

–150 

–200 

0 2 4 6 8 10 12 

Operational years 

Figure 2. The calculated NPVs for a large-size enterprise deciding whether to 

buy or lease storage with life expectancy from 0 to 10 years. 

500-Gbyte disk drives ( = 500), with a present cost of 

$150 per drive (K = $0.30), each specified to consume 0.01 

kW of power. Finally, we assume the electric utility cost 

is $0.04 per kilowatt hour (kWh), and the end-of-life disk 

salvage depreciation factor is 0.1. 

Table 3 shows the worksheet for the buy-or-lease deci- 

COMPUTER 

Table 2. Hypothetical storage cloud pricing structure. 

Default Storage > 50 Tbytes Storage > 100 Tbytes Storage > 500 Tbytes 

$0.15/Gbyte/month $0.14/Gbyte/month $0.13/Gbyte/month $0.12/Gbyte/month 

Storage 

requirement 

in Gbytes (V T ) 

Purchased 

drives 

Table 3. Buy-or-lease decision model worksheet for a single-user computer. 

Operating 

cost (C T ) 

Capital 

cost 

(E T ) 

Lease 

payment 

(L T ) Recurring cost (CPV T ) 

sion model for an expected storage life 

expectancy of 0 to 6 years. We compute the 

values for the “purchased drive” column 

using V T 500 – V T–1 500 , and the NPVs using 

Equation 1. In addition, we calculate the 

recurring costs (CPV T ) and salvage values 

(SPV T ) representing terms in the formula: 

ΔNPV = CPV + SPV – C 

T T 

C – E + I T T T 

CPV = T 

( 1 + I K ) T 

T 

∑ 

T – 0 

SPV . 

T = 

S 

( 1 + I K ) T 

Salvage value 

(SPV T ) 

Incremental 

net present 

value (NPV) 

0 100 1 ($178.70) $154.50 $98 ($236) $15.00 ($1,221) 

1 200 0 ($178.70) $2.90 $278 ($141) $9.58 ($1,131) 

2 300 0 ($178.70) $1.87 $458 $131 $6.12 ($863) 

3 400 0 ($178.70) $1.21 $638 $575 $3.91 ($421) 

4 500 0 ($178.70) $0.78 $818 $1,188 $2.50 $190 

5 600 1 ($182.21) $17.79 $998 $1,947 $3.19 $950 

6 700 0 ($182.21) $0.65 $1,178 $2,884 $2.04 $1,886 

Notes: C = $1,000, = 500 Gbytes, Κ = $0.30 ($150 per 500 Gbytes), = 0, = 0.1, = $0.04, P = 0.5 kW, P = 0.01 kW, and I = 1 percent. 

C D K 

The worksheet shows that for an 

expected storage life expectancy of less 

than four years, leasing is always the preferred 

option. This is because of the high 

cost of the disk controller the user purchases 

at the outset. If the storage system 

will likely exist for more than four years, 

however, purchasing storage is the superior 

choice. 

Medium-size enterprises 

Medium-size enterprises—those with 

clusters of tens to hundreds of servers— 

made up approximately 50 percent of all 

data centers installed in the US from 2000 to 2007. 12 We 

assume our medium-size enterprise has a storage requirement 

that grows at 1 Tbyte a year to serve a moderate-size 

data center of tens or low hundreds of servers. 

For such an enterprise, the human operator burden 

in owning and operating a storage cluster will be larger 

A 


aG 

F

A 


aG 

F 

than that of leasing from the storage cloud vendor. We 

therefore assume the proportional difference in the operator’s 

level of effort required to manage the system/data 

(H T = $70,000 per year) is = 0.5. We assume that 

the firm must purchase an enterprise-class RAID disk 

controller (C = $2,000), specified to consume 0.7 kW 

of power. Also, as storage is needed, the enterprise will 

purchase 1-Tbyte disk drives ( = 1,000), with a present 

cost of $300 per drive (K = $0.30), each specified 

to consume 0.01 kW of power. Finally, we assume the 

electric utility cost is $0.04 per kWh, and the end-of-life 

disk salvage depreciation factor is 0.1. 

Figure 1 shows the calculated NPV values for a storage 

life expectancy of 0 to 10 years. In all the operational lifetimes 

examined, the model shows that leasing is always 

preferable to purchasing storage. In this case, the clear 

recommendation to the medium-size enterprise is to lease 

storage from the storage cloud vendor. 

Large-size enterprises 

Next, we look at the benefits of purchasing versus leasing 

storage for a large-size enterprise—for example, a 

data center with thousands of servers. In this scenario, we 

assume the large enterprise’s storage requirement grows 

at 10 Tbytes per year. 

In this case, the human operator burden in owning and 

operating a storage cluster is even larger than that of leasing 

from the storage cloud vendor. We therefore assume 

the proportional difference in the operator’s level of effort 

required to manage the system/data (H T = $70,000/year) 

is = 1.0. We also assume that the firm must purchase an 

enterprise-class RAID disk controller (C = $2,000), specified 

to consume 0.7 kW of power. Furthermore, we assume 

the controller has a peak capacity of 100 Tbytes, and the 

firm will purchase additional controllers as the storage 

need arises. 

For the actual storage, the enterprise will purchase 

1-Tbyte disk drives ( = 1,000), with a present cost of 

$300 per drive (K = $0.30), each specified to consume 0.01 

kW of power. Finally, as before, we assume the electric utility 

cost is $0.04 per kWh, and the end-of-life disk salvage 

depreciation factor is 0.1. 

Figure 2 shows the calculatedNPVs for a storage life 

expectancy of 0 to 10 years. As the graph shows, leasing 

storage is advantageous up to a nine-year storage life 

expectancy. After that, it becomes more advantageous for 

the enterprise to purchase and maintain a storage cluster. 

Thus, the final decision to buy or lease storage will depend 

on the expected use of the storage and data. For example, 

if the storage is destined for use by a server cluster with 

a five-year life expectancy, the enterprise should lease 

storage. However, if the storage is destined for a long-term 

archival system with an indefinite life expectancy, the 

enterprise should purchase storage instead. 

LATENCY IS NOT ZERO 

A common flawed assumption in designing distributed 

systems is the notion that latency is zero. In fact, latency isn’t 

zero for cloud services. Accessing storage from across the 

commodity Internet can incur a substantial cost in terms 

of I/O latency. Our model doesn’t account for this latency. 

However, future extensions can incorporate this factor by 

estimating the profit parameter, P T , which we assumed to 

be equal when deriving our current model. We can use this 

profit parameter to reward services with faster response 

times. For example, an enterprise might naïvely consider a 

service that’s two times faster to be more productive, and 

hence two times more profitable. For the moment, we leave 

this substantive extension to a future time. 

Our primary purpose in this article is to stimulate 

discussion, debate, and future work in the quantitative 

modeling of the cloud computing industry. 

To this end, we propose a model to assist consumers, 

researchers, and policy makers in estimating the 

benefit of leasing from storage clouds. 

Ultimately, an organization’s buy-or-lease decision will 

depend on their anticipated parameters in the analysis. 

Our model simply provides a first stepping-stone for rational 

decision making to prevail in the cloud computing 

market. 


This article is based on work supported in part by US National 

Science Foundation grant 0721931. 

References 

1. K. Chandar, “SSD & HDD Market Tracker,” iSuppli 

research report, 2009; www.isuppli.com/Abstract/ 

________________________________ 

ABSTRACT - SSD_HDD Market Tracker 2009.pdf. 

2. P. Lyman and H.R. Varian, “How Much Information?” 

Oct. 2003, School of Information Management and Systems, 

Univ. of Calif., Berkeley; ______________ 

www2.sims.berkeley. 

_____________________________ 

edu/research/projects/how-much-info-2003. 

3. N. Leavitt, “Is Cloud Computing Really Ready for 

Prime Time?” Computer, Jan. 2009, pp. 15-20. 

4. A. Henry, “Keynote Address: Cloud Storage FUD 

(Failure, Uncertainty, and Durability),” presented at 

the 7th Usenix Conf. File and Storage Technologies, 

2009; www.usenix.org/media/events/fast09/tech/ 

____________ 

videos/henry.mov. 

5. R.W. Johnson and W.G. Lewellen, “Analysis of Leaseor-Buy 

Decision,” J. Finance, vol. 27, no. 4, 1972, pp. 

815-823. 

6. G.B. Harwood and R.H. Hermanson, “Lease-or-Buy 

Decisions,” J. Accountancy, vol. 142, no. 3, 1976, pp. 

83-87. 

APRIL 2010 

A 


aG 

F 

49

A 


aG 

F 

50 

COVER FEATURE 

7. P.F. Anderson and J.D. Martin, “Lease vs. Purchase 

Decisions: A Survey of Current Practice,” Financial 

Management, vol. 6, no. 1, 1977, pp. 41-47. 

8. M. Palankar et al., “Amazon S3 for Science Grids: A 

Viable Solution?” Proc. Int’l Workshop Data-Aware 

Distributed Computing, ACM Press, 2008, pp. 55-64. 

9. E. Deelman et al., “The Cost of Doing Science on the 

Cloud: The Montage Example,” Proc. ACM/IEEE Conf. 

Supercomputing, IEEE Press, pp. 1-16. 

10. E. Walker, “The Real Cost of a CPU Hour,” Computer, 

Apr. 2009, pp. 35-41. 

11. W.G. Lewellen, The Cost of Capital, Wadsworth Publishing, 

1970. 

12. US Environmental Protection Agency, Energy 

Star Program, Report to Congress on Server and 

Data Center Energy Efficiency, Public Law 109- 

431, Aug. 2007; www.energystar.gov/ia/partners/ 

___________________________________ 

prod_development/downloads/EPA_Datacenter_ 

Report_Congress_Final1.pdf. 

___________________ 

Edward Walker is a researcher with the Texas Advanced 

Computing Center at the University of Texas at Austin. 

His research interests include distributed systems, storage 

systems, and operating systems. Walker received a PhD in 

Welcomes Your Contribution 

COMPUTER 

Computer 

magazine 

looks ahead 

to future 

technologies 

computer science from the University of York, UK. He is a 

member of IEEE and the ACM. Contact him at ewalker@ ______ 

computer.org. 

Walter Brisken is a scientist at the National Radio Astronomy 

Observatory. His research interests include pulsars, 

the ionized interstellar medium, and very long baseline 

interferometry. Brisken received a PhD in physics from 

Princeton University. He is a member of the American 

Astronomical Society. Contact him at _____________ 

wbrisken@nrao.edu. 

Jonathan Romney is a scientist at the National Radio 

Astronomy Observatory. His research interests include 

active galactic nuclei and astronomical instrumentation. 

Romney received a PhD in astronomy from the California 

Institute of Technology. He is a member of the American 

Astronomical Society, the International Astronomical 

Union, and the International Union of Radio Science (URSI). 

Contact him at jromney@aoc.nrao.edu. 

________________ 

Selected CS articles and columns are available for free 

at http://ComputingNow.computer.org. 

Computer, the flagship publication of the IEEE 

Computer Society, publishes peer-reviewed 

technical content that covers all aspects of 

computer science, computer engineering, 

technology, and applications. 

Articles selected for publication in Computer 

are edited to enhance readability for the nearly 

100,000 computing professionals who receive 

this monthly magazine. 

Readers depend on Computer to provide current, 

unbiased, thoroughly researched information on 

the newest directions in computing technology. 

To submit a manuscript for peer review, 

see Computer’s author guidelines: 

www.computer.org/computer/author.htm 

A 


aG 

F

A 


aG 

F 

COVER FEATURE 

Karthik Kumar and Yung-Hsiang Lu, Purdue University 

The cloud heralds a new era of computing 

where application services are provided 

through the Internet. Cloud computing 

can enhance the computing capability of 

mobile systems, but is it the ultimate solution 

for extending such systems’ battery 

lifetimes? 

Cloud computing 1 is a new paradigm in which 

computing resources such as processing, 

memory, and storage are not physically present 

at the user’s location. Instead, a service 

provider owns and manages these resources, 

and users access them via the Internet. For example, 

Amazon Web Services lets users store personal data via 

its Simple Storage Service (S3) and perform computations 

on stored data using the Elastic Compute Cloud (EC2). 

This type of computing provides many advantages 

for businesses—including low initial capital investment, 

shorter start-up time for new services, lower maintenance 

and operation costs, higher utilization through virtualization, 

and easier disaster recovery—that make cloud 

computing an attractive option. Reports suggest that there 

are several benefits in shifting computing from the desktop 

to the cloud. 1,2 What about cloud computing for mobile 

users? The primary constraints for mobile computing are 

limited energy and wireless bandwidth. Cloud computing 

can provide energy savings as a service to mobile users, 

though it also poses some unique challenges. 

0018-9162/10/$26.00 © 2010 IEEE 

CLOUD COMPUTING 

FOR MOBILE USERS: 

CAN OFFLOADING 

COMPUTATION 

SAVE ENERGY? 


SAVING ENERGY FOR MOBILE SYSTEMS 

Mobile systems, such as smart phones, have become 

the primary computing platform for many users. Various 

studies have identified longer battery lifetime as the most 

desired feature of such systems. A 2005 study of users in 

15 countries 3 found longer battery life to be more important 

than all other features, including cameras or storage. A 

survey last year by ChangeWave Research 4 revealed short 

battery life to be the most disliked characteristic of Apple’s 

iPhone 3GS, while a 2009 Nokia poll showed that battery 

life was the top concern of music phone users. 

Many applications are too computation intensive to 

perform on a mobile system. If a mobile user wants to use 

such applications, the computation must be performed in 

the cloud. Other applications such as image retrieval, voice 

recognition, gaming, and navigation can run on a mobile 

system. However, they consume significant amounts of 

energy. Can offloading these applications to the cloud save 

energy and extend battery lifetimes for mobile users? 

Low-power design has been an active research topic for 

many years. In IEEE Xplore, searching “low” and “power” 

in the document title produces more than 5,000 results. 

There are four basic approaches to saving energy and extending 

battery lifetime in mobile devices: 

Adopt a new generation of semiconductor technology. 

As transistors become smaller, each transistor 

consumes less power. Unfortunately, as transistors 

become smaller, more transistors are needed to provide 

more functionalities and better performance; as 

a result, power consumption actually increases. 

APRIL 2010 

A 


aG 

F 

51

A 


aG 

F 

52 

COVER FEATURE 

Communication D 

COMPUTER 

Never 

offioad 

Computation C 

Depends 

on 

bandwidth 

B 

Always 

offioad 

Figure 1. Očoading is benećcial when large amounts of 

computation C are needed with relatively small amounts of 

communication D. 

Avoid wasting energy. Whole systems or individual 

components may enter standby or sleep modes to 

save power. 

Execute programs slowly. When a processor’s clock 

speed doubles, the power consumption nearly octuples. 

If the clock speed is reduced by half, the 

execution time doubles, but only one quarter of the 

energy is consumed. 

Eliminate computation all together. The mobile system 

does not perform the computation; instead, computation 

is performed somewhere else, thereby extending 

the mobile system’s battery lifetime. 

We focus on the last approach for energy conservation. 

Ofioading computation to save energy 

Sending computation to another machine is not a new 

idea. The currently popular client-server computing model 

enables mobile users to launch Web browsers, search 

the Internet, and shop online. What distinguishes cloud 

computing from the existing model is the adoption of virtualization. 

Instead of service providers managing programs 

running on servers, virtualization allows cloud vendors 

to run arbitrary applications from different customers on 

virtual machines. 

Cloud vendors thus provide computing cycles, and users 

can use these cycles to reduce the amounts of computation 

on mobile systems and save energy. Thus, cloud computing 

can save energy for mobile users through computation 

offloading. 5 Virtualization, a fundamental feature in cloud 

computing, lets applications from different customers run 

on different virtual machines, thereby providing separation 

and protection. 

Energy analysis for computation ofioading 

Various cost/benefit studies focus on whether to offload 

computation to a server. 6,7 The following example provides 

a simple analysis for this decision. 

Suppose the computation requires C instructions. Let 

S and M be the speeds, in instructions per second, of the 

cloud server and the mobile system, respectively. The same 

task thus takes C/S seconds on the server and C/M seconds 

on the mobile system. If the server and mobile system exchange 

D bytes of data and B is the network bandwidth, it 

takes D/B seconds to transmit and receive data. The mobile 

system consumes, in watts, P c for computing, P i while 

being idle, and P tr for sending and receiving data. (Transmission 

power is generally higher than reception power, 

but for the purpose of this analysis, they are identical.) 

If the mobile system performs the computation, the 

energy consumption is P c × (C/M). If the server performs 

the computation, the energy consumption is [P i × (C/S)] + 

[P tr × (D/B)]. The amount of energy saved is 

P c × C 

M − P i 

× C 

S − P tr 

D 

× . (1) 

B 

Suppose the server is F times faster—that is, S = F × M. 

We can rewrite the formula as 

C 

M × Pc − P ⎛ ⎞ i 

⎝ 

⎜ F ⎠ 

⎟ − P D 

× . (2) 

tr B 

Energy is saved when this formula produces a positive 

number. The formula is positive if D/B is sufficiently small 

compared with C/M and F is sufficiently large. The values 

of M, P , P , and P are parameters specific to the mobile 

i c th 

system. For example, an HP iPAQ PDA with a 400-MHz 

(M = 400) Intel XScale processor has the following values: 

P fi 0.9 W, P fi 0.3 W, and P fi 1.3 W. 

c i tr 

If we use a four-core server, with a clock speed of 3.2 

GHz, the server speedup F may be given by (S/M) fi [(3.2 

× 1,024 × 4 × X)/400], where X is the speedup due to 

additional memory, more aggressive pipelining, and so 

forth. If we assume X = 5, we obtain the value of F fi 160. 

The value of F can increase even more with cloud computing 

if the application is parallelizable, since we can 

offload computation to multiple servers. If we assume that 

F = 160, Equation 2 becomes 

C 

400 

⎛ 0.3 ⎞ 

× 0.9 − 

⎝ 

⎜ 

160⎠ 

⎟ − 1.3 × D 

B 

D 

≈ 0.00225 × C − 1.3 × . (3) 

B 

For offloading to break even, we equate Equation 3 to 

zero and obtain 

B ≈ 577.77 × o D 

, (4) 

C 

where B o is the minimum bandwidth required for offloading 

to save energy, determined by the ratio of (D/C). If (D/C) 

is low, then offloading can save energy. Thus, as Figure 

1 shows, offloading is beneficial when large amounts of 

computation C are needed with relatively small amounts 

of communication D. 

A 


aG 

F

A 


aG 

F 

Sample applications bene ting from ofioading 

Two sample applications illustrate the benefits of 

offloading: a chess game and image retrieval. 

Chess is one of the world’s most popular games. A 

chessboard has 8 × 8 = 64 positions. Each player controls 

16 pieces at the beginning of the game. Chess is 

Markovian, meaning that the game is fully expressed by 

the current state. Each piece may be in one of the 64 possible 

locations and needs 6 bits to represent the location. 

(This is an overestimate: Some pieces have restrictions— 

for example, a bishop can move to only half of the board, 

that is, 32 possible locations). To represent a chess game’s 

current state, it is sufficient to state that 6 bits × 32 pieces 

= 192 bits = 24 bytes; this is smaller than the size of a 

typical wireless packet. 

The amount of computation for chess is very large; Claude 

Shannon and Victor Allis estimated the complexity of chess 

to exceed the number of atoms in the universe. Chess can be 

parallelized, 8 making the value of F in Equation 2 very large. 

Since the amount of computation C is extremely large, and D 

is very small, chess provides an example where offloading is 

beneficial for most wireless networks. 

An image retrieval application retrieves images similar 

in content to a query from an image collection. The 

program accomplishes this by comparing numerical representations 

of the images, called features. The features 

for the image collection can be computed in advance; 

for a query, the program computes its features during 

retrieval and compares these with the image collection. 

Since most of the computation is done in advance, less 

computation is performed online, and the value of C is 

small. D is large since considerable data must be sent. As 

a result, even if the values of F become fl, D/B might still 

be too large when compared to C/M in Equation 2. Thus, 

offloading saves energy only if B is very large—that is, 

at high bandwidths. 

The “Mobile Image Processing” sidebar has more detail 

on the advantages of mobile devices offloading image retrieval 

to the cloud. 

Making computation ofioading more attractive 

Analysis indicates that the energy saved by computation 

offloading depends on the wireless bandwidth B, the 

amount of computation to be performed C, and the amount 

of data to be transmitted D. Existing studies thus focus on 

determining whether to offload computation by predicting 

the relationships among these three factors. 

However, there is a fundamental assumption underlying 

this analysis with the client-server model: Because the 

server does not already contain the data, all the data must 

be sent to the service provider. The client must offload the 

program and data to the server. For example, typically a 

newly discovered server for computation offloading does 

not already contain a mobile user’s personal image collec- 

MOBILE IMAGE PROCESSING 

Mobile devices such as cell phones and PDAs are becoming 

increasingly popular. Most of these devices are equipped 

with cameras and have several gigabytes of ash storage 

capable of storing thousands of images. With such large image 

collections, two functionalities become important: accessing 

specićc sets of images from the collection, and transmitting the 

images over a wireless network to other devices and servers for 

storage. 

For accessing a specićc set of images, content-based image 

retrieval (CBIR) can be a better alternative than manually browsing 

through all of them. For example, a user might want to view 

all images containing a specićc person or captured at a specićc 

location. Mobile image retrieval allows the user to obtain the 

relevant pictures by comparing images and eliminating the 

irrelevant matches on the mobile system. 

Several studies propose performing CBIR on mobile 

devices. 1-4 Because these mobile devices are battery powered, 

energy conservation is important. 2-4 It is energy e cient to par - 

tition CBIR between the mobile device and server depending on 

the wireless bandwidth. 3 As the bandwidth increases, očoad - 

ing image retrieval saves more energy. 

Most of the energy consumption for očoaded applications 

is due to transmission. For image retrieval, transmitting the 

images over a wireless network consumes signićcant amounts 

of energy. The images may be preprocessed on the mobile 

device before transmission5 to reduce the transmission energy. 

This reduction in transmission energy is achieved by reducing 

the ćle sizes. However, the amount of energy saved depends on 

the wireless bandwidth and the image contents. 

Preprocessing the images saves energy if the reduction in 

transmission energy compensates for the energy spent due to 

preprocessing. If the wireless bandwidth is high, the value of the 

former reduces. Moreover, di erent images may have di erent 

values of the latter based on their contents. Hence preprocessing 

must be adaptive based on the wireless bandwidth and the 

image contents. Wireless transmission energy is the most signić - 

cant bottleneck to energy savings in mobile cloud computing, 

and such techniques will become increasingly signićcant as it 

becomes more popular. 

References 

1. J. Yang et al., “A Fast Image Retrieval System Using Index 

Lookup Table on Mobile Device,” Proc. 19th Int’l Conf. Pattern 

Recognition (ICPR 08), IEEE Press, 2008, pp. 265-271. 

2. C. Zhu et al., “iScope: Personalized Multimodality Image 

Search for Mobile Devices,” Proc. 7th Int’l Conf. Mobile Systems, 

Applications, and Services (Mobisys 09), ACM Press, 

2009, pp. 277-290. 

3. Y-J. Hong, K. Kumar, and Y-H. Lu, “Energy-Efficient Content- 

Based Image Retrieval for Mobile Systems,” Proc. IEEE Int’l 

Symp. Circuits and Systems (ISCAS 09), IEEE Press, 2009, pp. 

1673-1676. 

4. D. Chen et al., “Tree Histogram Coding for Mobile Image 

Matching,” Proc. 2009 Data Compression Conf., IEEE CS Press, 

2009, pp. 143-152. 

5. Y. Nimmagadda, K. Kumar, and Y-H. Lu, “Energy-Efficient 

Image Compression in Mobile Devices for Wireless Transmission,” 

Proc. IEEE Int’l Conf. Multimedia and Expo (ICME 09), IEEE 

Press, 2009, pp. 1278-1281. 

APRIL 2010 

A 


aG 

F 

53

A 


aG 

F 

54 

COVER FEATURE 

(a) 

(b) 

User 

User 

tion. However, cloud computing changes that assumption: 

The cloud stores data and performs computation on it. For 

example, services like Google’s Picasa and Amazon S3 

can store data, and Amazon EC2 can be used to perform 

computation on the data stored using S3. 

This results in a significant change in the value of D for 

most applications. There is no longer a need to send the 

data over the wireless network; it suffices to send a pointer 

to the data. Also, the value of F is elastic: Large numbers of 

processors can be obtained on the cloud. This increases 

the energy savings in Equation 2: A very small D and very 

large F imply that energy can always be saved. 

CHALLENGES AND POSSIBLE SOLUTIONS 

Does this make cloud computing the “ultimate” solution 

to the energy problem for mobile devices? Not 

quite. While cloud computing has tremendous potential 

to save energy, designers must consider several issues 

including privacy and security, reliability, and handling 

real-time data. 

Privacy and security 

In cloud computing, Web applications and data replace 

traditional stand-alone programs, which are no longer 

stored in users’ computers. Shifting all data and computing 

resources to the cloud can have implications for privacy 

and security. Because the data is stored and managed in 

the cloud, security and privacy settings depend on the IT 

management the cloud provides. 

A bug or security loophole in the cloud might result in 

a breach of privacy. For example, in March 2009, a bug 

in Google caused documents to be shared without the 

owners’ knowledge, 9 while a July 2009 breach in Twitter 

allowed a hacker to obtain confidential documents. 10 

Cloud service providers typically work with many thirdparty 

vendors, and there is no guarantee as to how these 

vendors safeguard data. For example, a phishing attack in 

2007 duped a staff member for salesforce.com into reveal- 

COMPUTER 

Data 

Data 

E (data) E (data) 

Encrypt Internet 

E(data) 

Encrypt 

E (data) E (data) 

Internet 

Cloud 

storage 

Data 

Decrypt Data 

Cloud 

storage 

Figure 2. Two encryption scenarios for cloud computing. (a) Data remain decrypted at the cloud storage site, preventing 

unauthorized access through the Internet; the cloud vendor cannot access the data either. (b) Data are decrypted by the cloud 

vendor to enable necessary operations on the data. 

ing a password; 13 the attacker then used the password to 

access confidential data. 

Another potential privacy violation is the “tracking” 

of individuals through location-based navigation data 

offloaded to the cloud. Moreover, data stored at one location 

may not be secure at another due to different access 

rules. For example, the EU has stricter privacy protection 

than the US, where laws such as the Patriot Act give government 

agencies and other entities more power to access 

personal information. Since the storage locations are typically 

unknown to the user, determining what laws apply to 

safeguarding data can be difficult. This may raise serious 

security concerns for a company that stores its trade secrets 

in the cloud, or a user who uploads a patentable idea 

from a mobile system to the cloud. 

Clearly, some types of data cannot be stored in the 

cloud without considering these privacy and security implications. 

One possible solution is to encrypt data before 

storage. However, encryption alone will not solve the problem. 

Figure 2 shows two encryption scenarios. In Figure 2a, 

the data remain encrypted at the cloud storage site. This 

can prevent unauthorized access even when the storage is 

breached in the cloud; the cloud vendor cannot access the 

data either. In Figure 2b, the cloud vendor decrypts data to 

perform operations on that data. For example, in the case 

of a document, the cloud vendor must know which words 

are used to check spelling; for a spreadsheet, the cloud 

application must know the numbers for calculation. This 

is the general case for cloud services. 

Another possible privacy and security solution is to 

use a technique called steganography, as described in the 

“Data Outsourcing and Privacy” sidebar. Steganographic 

techniques can be used to transform the data so that operations 

can be performed without exposing them. 

Performing encryption or steganographic techniques 

before sending data to the cloud requires some additional 

processing C p on the mobile system and consumes additional 

energy. Equation 2 thus can be rewritten as 

A 


aG 

F

A 


aG 

F 

DATA OUTSOURCING AND PRIVACY 

Data outsourcing refers to sending data outside users’ immediate computing 

environment. This is an increasingly important concern for cloud computing. As 

government agencies shift to the cloud (for example, www.apps.gov and www.data. 

gov), information like healthcare, criminal, and tax records move to the cloud. This 

makes security and privacy a critical concern. 

Multimedia content like images and videos have signićcant redundancy. This 

makes it possible to hide data in multimedia using steganography. 1 Figure A shows 

an example of steganography: images (a) and (c) look identical, but image (c) contains 

image (b) hidden in it. Applying appropriate transformations to image (c) can 

obtain image (b). Steganographic techniques can be used to transform the data 

before storage so that operations can still be performed on the data. 

(a) (b) (c) 

Figure A. An example of steganography: images (a) and (c) look identical, 

but image (c) contains image (b) hidden in it. Applying appropriate 

transformations to image (c) can obtain image (b). 

For text documents and spreadsheets, however, there is little redundancy. This 

makes it di cult to transform the data and still perform operations. For example, 

it is hard to spell check a text document that has alphabets mapped onto symbols. 

If the mapping is one-to-one and onward, alphabet frequency can be easily used 

to determine the words. Similarly, if the elements in a spreadsheet are transformed, 

performing meaningful mathematical operations on them is di cult. 

Performing computation on encrypted or steganographic data still remains an 

open research problem. 

Developing a hybrid cloud is an interim solution to the data outsourcing problem. 

A hybrid cloud is “a cloud computing environment in which an organization 

provides and manages some resources in-house and has others provided externally” 

(searchcloudcomputing.com). _________________ 

The claim is that this is more feasible 

currently because large enterprises already have the substantial investments in 

their infrastructure required to provide resources in-house. In addition, many 

organizations would prefer to keep sensitive data under their own control to 

ensure privacy and security. 

Reference 

1. F.A.P. Petitcolas, R.J. Anderson, and M.G. Kuhn, “Information Hiding—A Survey,” 

Proc. IEEE, Special Issue on Protection of Multimedia Content, vol. 87, no. 7, 1999, 

pp. 1062-1078. 

C 

M × Pc − P ⎛ ⎞ i 

⎝ 

⎜ F ⎠ 

⎟ − Ptr D 

× 

B − Pc × C p 

, (5) 

M 

where P c × (C p /M) is the additional energy required to protect privacy and 

security. If this value is significant, cloud computing might not save energy 

for the mobile user. Instead, it may be more energy efficient to perform 

operations on the mobile system, since C p = 0 when the device performs 

the computation. 

Reliability 

Another potential concern with mobile 

cloud computing is reliability. 11 A mobile 

user performing computation in the cloud 

depends on the wireless network and 

cloud service. Dependence on the wireless 

network implies that cloud computing 

may not even be possible, let alone energy 

efficient, when connectivity is limited. 

This is typical in regions like national 

parks; a user may thus not be able to organize, 

retrieve, or identify any images 

captured in the park. Mobile cloud computing 

is also difficult in locations such 

as the basement of a building, interior of 

a tunnel, or subway. In these cases, where 

the value of B in Equation 2 can become 

very small or even zero, cloud computing 

does not save energy. 

Dependence on the cloud for important 

computations could lead to problems 

during service outages. Several service 

providers such as Google, Amazon, and 

T-Mobile have experienced such outages, 

as Table 1 shows. These can significantly 

reduce the value of F in Equation 2. 

Data storage presents another reliability 

problem. In October 2009, both 

T-Mobile’s and Microsoft’s mobile Sidekick 

service crashed, and all customers 

lost their data and contacts. One option 

in such scenarios involves an independent 

backup of data with an alternate service 

provider, which might increase the value 

of D in Equation 2. 

Real-time data 

Some applications—including chess, 

searching newly captured images for 

content-based image retrieval, mobile surveillance, 

and context-aware navigation 

—have real-time data. In such scenarios, 

D in Equation 2 is no longer a pointer to 

the data; it refers to the actual data. For 

applications like chess, the value of D is 

small and hence offloading can still save 

energy. When the value of D is large, 

offloading may not save energy. In such 

cases, performing the computation on the 

mobile system may be more energy efficient. 

A possible solution is partitioning 

computation between the mobile system 

and the cloud to reduce energy consumption. 

Such a solution may include partially 

APRIL 2010 

A 


aG 

F 

55

A 


aG 

F 

56 

COVER FEATURE 

processing the real-time data on the mobile system. If the 

processed data are smaller in size, sending the processed 

data to the server reduces the wireless transmission energy. 

Our analysis suggests that cloud computing can 

potentially save energy for mobile users. However, 

not all applications are energy efficient 

when migrated to the cloud. Mobile cloud 

computing services would be significantly 

different from cloud services for desktops because they 

must offer energy savings. The services should consider 

the energy overhead for privacy, security, reliability, and 

data communication before offloading. 


We thank Jibang Liu for the steganography images used in 

this article. 

References 

1. M. Creeger, “CTO Roundtable: Cloud Computing,” ACM 

Queue, June 2009, pp. 1-2. 

2. Google Tech Talk, “Away with Applications: The Death 

of the Desktop,” 4 May 2007; http://video.google.com/ 

__________________________ 

videoplay?docid=-6856727143023456694. 

3. CNN.com, “Battery Life Concerns Mobile Users,” 23 Sept. 

2005; www.cnn.com/2005/TECH/ptech/09/22/phone. 

____ study. 

4. J. Paczkowski, “Iphone Owners Would Like to Replace 

Battery,” All Things Digital, 21 Aug. 2009; http:// 

digitaldaily.allthingsd.com/20090821/iphone-ownerswould-like-to-replace-battery-att. 

COMPUTER 

Table 1. Service provider outages. 11,12 

Outage date Vendor Service 

July 2008 Amazon S3 outage 

Feb 2008 Amazon S3 outage 

August 2008 Google Gmail service outage 

October 2009 T-Mobile/Microsoft Sidekick loses users’ data 

build your career 

IN COMPUTING 

5. K. Yang, S. Ou, and H.H. Chen, “On Effective Offloading 

Services for Resource-Constrained Mobile Devices Running 

Heavier Mobile Internet Applications,” IEEE Comm. 

Magazine, vol. 46, no. 1, 2008, pp. 56-63. 

6. C. Wang and Z. Li, “Parametric Analysis for Adaptive Computation 

Offloading,” ACM SIGPLAN Notices, vol. 39, no. 6, 

2004, pp. 119-130. 

7. R. Wolski et al., “Using Bandwidth Data to Make Computation 

Offloading Decisions,” Proc. IEEE Int’l Symp. Parallel 

and Distributed Processing (IPDPS 08), 2008, pp. 1-8. 

8. M. Newborn, “A Parallel Search Chess Program,” Proc. ACM 

Ann. Conf. Range of Computing: Mid-80’s Perspective, ACM 

Press, 1985, pp. 272-277. 

9. J. Kincaid, “Google Privacy Blunder Shares Your Docs 

without Permission,” TechCrunch, 7 Mar. 2009; ____ http:// 

___________________________________ 

techcrunch.com/2009/03/07/huge-google-privacy-blunder- 

_______________________ 

shares-your-docs-without-permission. 

10. R. McMillan, “Hacker: I Broke into Twitter,” PCWorld. 

com, 1 May 2009; www.pcworld.com/businesscenter/ 

______________________________ 

article/164182/hacker_i_broke_into_twitter.html. 

11. M. Armbrust et al., “Above the Clouds: A Berkeley View of 

Cloud Computing,” tech. report UCB/EECS-2009-28, EECS 

Dept., Univ. of California, Berkeley, 2009. 

12. T-Mobile Forums, “A Message from Our Chief Operations 

Officer, Jim Alling,” 6 Oct. 2009; http://forums.t-mobile. 

com/t5/Previous-Sidekick-Models/A-Message-From-Our- 

_____________________________ 

Chief-Operations-Officer-Jim-Alling/m-p/200661. 

13. R. McMillan, “Salesforce.com Warns Customers of Phishing 

Scam,” PCWorld, 6 Nov. 2007; http://www.pcworld. 

___________________________________ 

com/businesscenter/article/139353/salesforcecom_warns_ 

customers_of_phishing_scam.html. 

______________________ 

Karthik Kumar is a PhD candidate in the School of Electrical 

and Computer Engineering at Purdue University. 

His research interests include energy conservation for 

mobile systems and resource allocation for scalable computing. 

Kumar is a student member of IEEE. Contact him 

at kumar25@purdue.edu. 

_______________ 

Yung-Hsiang Lu is an associate professor in the School of 

Electrical and Computer Engineering at Purdue University. 

His research focuses on energy efficiency of computing 

systems. Lu received a PhD in electrical engineering from 

Stanford University. He is a senior member of IEEE. Contact 

him at yunglu@purdue.edu. 

_____________ 

www.computer.org/buildyourcareer 

A 


aG 

F

A 


aG 

F 


Linking Software 

Development 

and Business 

Strategy through 

Measurement 

Victor R. Basili, Mikael Lindvall, Myrna Regardie, and Carolyn Seaman, 

Fraunhofer Center for Experimental Software Engineering 

Jens Heidrich, Jürgen Münch, Dieter Rombach, and Adam Trendowicz, 

Fraunhofer Institute for Experimental Software Engineering 

The GQM + Strategies approach extends the goal/question/metric 

paradigm for measuring the success or failure of goals and strategies, 

adding enterprise-wide support for determining action on the basis 

of measurement results. An organization can thus integrate its 

measurement program across all levels. 

Most organizations that develop software try 

to retain their competitive edge by reducing 

software-related risks. Effective risk management 

requires aligning business goals 

with development strategies and translating 

the results into a quantitative project management 

plan. In addition, organizations must also justify cost and 

resources for software and system development and other 

IT services. Often, such justification demands a concrete 

demonstration of how such development will contribute 

to the organization’s overall business goals—the top-level 

goals the organization desires to accomplish. 

Working against both alignment and justification is the 

lack of methods for linking business goals and softwarerelated 

efforts. Without a deep understanding of how software 

development fits into their organization’s business 

objectives, decision makers can neither plan development 

nor evaluate the organization-wide success of development 

strategies. 

0018-9162/10/$26.00 © 2010 IEEE 


To address this need, we have developed GQM + Strategies, 

which uses measurement to link goals and strategies on all 

organizational levels. The approach is based on rationales for 

deciding when and how to transform goals into operations 

and how to evaluate the success of strategies with respect 

to those goals. The “Why Business Alignment?” sidebar describes 

our motivation for developing this approach. 

As the name implies, GQM + Strategies is based on the 

goal/question/metric (GQM) paradigm 1 but adds the ability 

to create measurement programs that ensure alignment 

between goals and strategies from the highest strategic 

business levels to individual development projects and vice 

versa. 2 Although we derived the approach from software 

development experiences, it is not necessarily applicable 

solely to this domain. 

GOAL-ORIENTED APPROACHES 

Software development organizations often encounter 

problems when instituting measurement programs, such 

APRIL 2010 

A 


aG 

F 

57

A 


aG 

F 

58 


as collecting too much data, failing to gather useful or correct 

data, and not having a mechanism for analyzing the 

data in a way that contributes to sound strategic decisions. 

This leads to numerous problems, including decreased 

cost-effectiveness of the measurement program and disillusionment 

about metrics on the part of developers and 

managers. The end result is often the eventual failure of 

the measurement program as a whole. 

COMPUTER 

WHY BUSINESS ALIGNMENT? 

With the increasing dependence on software and other 

information technology (IT) has come a commensurate 

growth in software system size and complexity. This growth in 

turn has magnified the cost, schedule, and quality concerns 

that have always plagued software development. For decades, 

software engineering researchers and practitioners 

have attempted to reduce development cost, shorten development 

time, and increase quality. Although great strides 

have been made in all three areas, software growth along all 

dimensions (size, complexity, pervasiveness, criticality, and so 

on) has outpaced the ability to control all the developmentrelated 

factors. 

Ečorts to dećne and understand these relationships have made 

one notion crystal clear: Issues related to software cost, schedule, 

and quality are inextricably linked with the larger issues that face 

the businesses developing the software—regardless of the specićc 

business objectives. Some businesses sell what they develop, 

either as contracted custom software or as commercial shrinkwrapped 

applications. Others sell a product or service, of which 

software is a signićcant component. Still others might develop 

software only to support their organizational IT infrastructure so 

that they can develop and market products more ečectively. 

Finally, some have no commercial interest at all, such as nonproćt 

organizations, government entities, or educational institutions. 

Although all these organizational conćgurations challenge their 

development projects in dičerent ways, the point is that all have a 

larger business context, which encompasses larger business goals, 

strategies, and success measures. All software development 

projects must consider context factors—the variables that represent 

the organizational environment and ačect the kind of models 

and data the project can use. 

The problem is not that businesses fail to achieve their objectives, 

but rather that they do not always state these objectives 

explicitly or clearly enough to verify that they have indeed achieved 

those objectives. It is even less clear how a business translates its 

objectives into its lower organizational levels and into individual 

projects. At present, no methodology bridges the gap between 

business strategies and their project-level implementation. 

Quantitative data is a prerequisite to understanding the relationships 

between the business and project-level goals and 

verifying the achievement of objectives—one reason that software 

development maturity requires the use of quantitative measures. 

However, popular software improvement strategies, such as Capability 

Maturity Model Integration and the IT Infrastructure Library, 

are not directly and explicitly linked to business value. Consequently, 

the investment in collecting data does not result in the 

expected benećts, and the contribution of project performance to 

the achievement of strategic goals remains unclear. 

To address these problems, researchers have developed 

goal-oriented approaches—so called because they 

use goals, objectives, strategies, or other mechanisms to 

guide the choice of data to be collected and systematically 

analyzed. The GQM approach, for example, provides a 

method for an organization or a project to define goals, 

refine those goals into specifications of the data to be collected, 

and then analyze and interpret the resulting data in 

light of the original goals. Implicit in the GQM approach is 

the use of interpretation models, which help practitioners 

interpret the resulting data in a specific context. The GQM 

approach has served the software industry for several decades 

in defining measurement programs, but it does not 

explicitly provide support for motivating and integrating 

the measurement of goals at various organizational levels, 

such as project goals, business objectives, and corporate 

strategies. It also does not encourage users to explicitly 

document assumptions, estimated unknowns that can 

affect data interpretation. 

Another goal-oriented approach, Balanced Scorecard 

(BSC), 3 links strategic objectives and measures through a 

scorecard, which consists of four perspectives: financial, 

customer, internal business processes, and learning and 

growth. BSC does not dictate a static set of measures, but 

rather serves as a framework for strategic measurement 

and management. As such, it is a tool for defining strategic 

goals from multiple perspectives beyond a purely financial 

focus, but it does not address lower-level goal setting. 

Practical Software Measurement (PSM) 4 offers detailed 

guidance on software measurement, including providing 

a catalog of specific measures and information on how an 

organization can apply those measures in projects. PSM 

also includes a process for choosing measures according 

to the issues and objectives relevant to a software development 

project. For the most part, however, approaches such 

as BSC and PSM do not support the alignment of objectives 

at different organizational levels, or this is simply not the 

focus of these approaches. 

Researchers have also proposed various combinations 

of GQM, BSC, and PSM 5-8 that recognize the need to link 

higher- and lower-level goals. However, because these 

approaches do not support different goals at explicitly 

linked organizational levels, it is difficult to feed analytical 

results and interpretations back up the chain. In 

contrast, GQM + Strategies creates mappings between the 

data related to goals at different levels so that insights 

gained relative to one goal at one level—whether project 

or departmental, or a business objective—can support and 

contribute to satisfying goals at other organizational levels 

without requiring that each level share the same goals. 

Some application domains have specific requirements 

that emphasize the need to link business goals to lowerlevel 

properties. Organizations are becoming increasingly 

aware, for example, that the IT infrastructure itself im- 

A 


aG 

F

A 


aG 

F 

Goal + Strategies element GQM graph 

inffiuences 

Context/ 

assumption 

inffiuences 

Figure 1. GQM + Strategies components. The primary components are the Goal + Strategies element and the GQM graph. The 

Goal + Strategies element includes a single goal and derived strategies, as well as all context information and assumptions that 

explain how goals and strategies link. The GQM graph refiects a single GQM goal, the corresponding questions and metrics, and an 

interpretation model. 

poses significant risks, which has led to several regulatory 

constraints in both the IT governance and IT service domains, 

such as the Sarbanes-Oxley Act in the US. 9 

The solutions proposed by models in these domains 

offer connections between predefined sets of goals and attributes 

of the IT infrastructure. They are sharply focused 

on a particular domain, such as IT governance, and define 

a very detailed model that fits that domain quite well. To 

apply such an approach, however, the organization must 

strictly follow the prescribed model. Release 4 of the Control 

Objectives for Information and Related Technology 

(CoBIT) standard (www.isaca.org), for example, is based 

on a process model that subdivides IT into four domains 

and 34 processes. 

Approaches such as CoBIT and the IT Infrastructure 

Library (ITIL) provide no mechanism for adapting and tailoring 

the solution, nor do they support addressing the 

specific context or documenting inherent assumptions. 

No clearly defined interpretation model indicates if an 

overall strategy is working or has to be changed to avoid 

business failure. 

COMPONENTS OF AN 

INTEGRATED APPROACH 

Goal 

Strategy 

realized 

by a 

set of 

leads to 

a set of 

> made 

measurable 

through 

< measures 

achievement 

of 

GQM + Strategies aims to address the weaknesses of existing 

goal-oriented approaches by providing both explicit 

links among organizational levels and the flexibility to 

tailor the approach to the organization’s specific needs and 

objectives. Eight conceptual components form the basis for 

constructing a consistent model. Figure 1 illustrates how 

these components interrelate. 

GQM goal 

Goal + Strategies element GQM graph 

made measurable through 

Question 

Question 

Interpretation model 

is part of 

Metric 

Metric 

Metric 

Business goals: goals the organization wishes to accomplish 

in general to achieve its strategic objectives. 

Context factors: environmental variables that represent 

the organizational environment and affect the 

kind of models and data that can be used. 

Assumptions: estimated unknowns that can affect the 

interpretation of the data. 

Strategies: a set of possible approaches for achieving a 

goal that may be refined by a set of concrete activities. 

Level i goals: a set of lower-level goals inherited from 

the previous level’s (i 1) goals as part of the previous 

level’s goal strategy—an example is a goal related to 

a project that is part of a selected software strategy. 

Interpretation models: models that help interpret 

data to determine if the goals at all levels have been 

achieved. 

Goal + Strategies element: a single goal and derived 

strategies (including a set of concrete activities), as 

well as all context information and assumptions that 

explain how the goal and corresponding strategies 

are linked. 

GQM graph: a single GQM goal that measures a 

Goal + Strategies element and its corresponding questions, 

metrics, and interpretation models. 

All organizations have business goals, such as improving 

customer satisfaction, garnering market share, and 

reducing production costs, and they devise strategies 

to deal with them, taking into account the context and 

making explicit any assumptions. Strategies specify how 

to achieve the goal and should, in turn, define lower-level 

APRIL 2010 

A 


aG 

F 

59

A 


aG 

F 

60 


Figure 2. A GQM + Strategies model for ABC, a company that provides its customers with information services through the Web. 

GQM + Strategies enforces the explicit documentation of the relevant context factors and assumptions that are necessary for 

understanding and evaluating each goal. A Goal + Strategies element consists of a goal and an associated strategy (bottom of each 

goal box). Each element, in turn, is associated with a GQM graph (green rectangle to the right of the goal) representing questions and 

metrics as well as an interpretation model that evaluates if the goal was achieved. 

goals for various parts of the organization, such as software, 

hardware, and marketing. An organization might 

devise a separate set of strategies to deal with these lowerlevel 

goals. The number of goal and strategy levels depends 

on the organization’s internal structure. 

Assumptions and context information about the organization 

strongly influence the definition of goals and 

strategies. Measuring the effectiveness and the accomplishment 

of goals and the effectiveness of strategies at 

all levels is critical for business success. And making goals 

connected and transparent at all levels helps in communicating 

and achieving them. What most organizations need 

is support for these concepts. 

The components in Figure 1 provide this support by allowing 

multiple goal levels and multiple strategies for each 

goal (a GQM + Strategies element), as well as a measurement 

plan (GQM graph). The GQM + Strategies element provides 

the basic components for deriving the goals and strategies 

as influenced by the context; the basic known facts about 

the market, the organization, the product, and so on; and 

the assumptions—factors that are assumed or guessed 

but not known for sure. A key element is making the context 

and assumptions explicit so that they can be verified. 

The relevant context and assumptions aid in defining the 

rationale for choosing specific goals and strategies. A set 

COMPUTER 

Business level 

Software level 

Project level 

GQM + Strategies elements 

Goal 1: Increase profit 

from software service usage 

Strategy 1: Deliver 

added functionality 

Goal 2: Deliver 5% new 

functionality every 6 months 

within 10% of budget 

Strategy 2: Use MoSCoW 

and Cocomo 

Goal 3: Apply MoSCoW 

and Cocomo effectively 

Strategy 3: Conduct 

training, determine tools, 

perform pilot study 

GQM goals 

G1: Evaluate 

trend in profit 

G2: Evaluate 

functional 

growth of each 

release 

G3: Evaluate 

effectiveness 

of MoSCow 

and Cocomo 

Questions Metrics Interpretation model 

Q1: What is 

the current 

profit? 

Q2: What is 

the profit for 

year 2 and 

year 3? 

Q3: How 

many M 

requirements 

in each release? 

Q4: How 

long between 

releases? 

Q5: Cost 

estimation 

accuracy? BV: budget 

variance 

Q6: How 

extensive 

was the 

training? 

Q7: What is 

cost of 

training? 

P0: current 

annual 

profit 

PX: annual 

profit for 

year x 

MR: number 

of customerrequested 

requirements 

(M) 

implemented 

RD: release 

duration 

NT: number 

of trainees 

TC: hours 

spent in 

training 

If P2 0 

and 

P3 2 

and … 

then goal is 

satisfied 

of predefined goals and strategies can become part of an 

organization-specific experience base and allow different 

parts of the organization to select and adapt predefined 

goals and strategies. 

Associated with each GQM + Strategies element is a 

measurement plan that uses the GQM approach measurement 

and evaluation framework to specify how to 

evaluate the goal, what data to collect, and how to interpret 

that data. The nodes of each GQM graph consist of 

a measurement goal, which describes what knowledge 

needs to be gained from the measurement activity; a set 

of questions to be answered; the metrics and data items 

required to answer the questions; and an interpretation 

model that specifies how the data items are to be 

combined and what the criteria are for determining the 

goal’s success. 

The nodes are related semihierarchically. A goal can 

have several associated measurement goals, each of 

which is the basis for an entire GQM graph. However, different 

GQM structures are likely to use some of the same 

questions and metrics, and interpretation models might 

combine data from different GQM structures, thus optimizing 

metrics collection. The results of the lower-level 

interpretation models feed into the higher-level ones to 

provide feedback on lower-level goal achievement. 

A 


aG 

F 

… 

…

A 


aG 

F 

With the complete model, an organization can 

not only define measurement consistent with 

larger, upper-level organizational concerns, but 

also interpret and roll up the resulting measurement 

data at each level. 

AN APPLICATION 

To illustrate the features of our approach, we 

consider a fictitious but representative type of 

organization, ABC, which provides information 

services to its customers through the Web. Customers 

do not buy the software, but rather pay a 

service fee for access to information and to software 

that searches, analyzes, and presents that 

information. Thus, the business model implies that 

the number of customer accesses to ABC’s system 

determines the company’s revenue. Figure 2 represents 

a model of this application. 

Business goals 

Although the approach is flexible enough to accommodate 

several different entry points, the starting point of the 

GQM + Strategies process in this example is a business goal. 

As Goal 1 shows, one of ABC’s business goals is to increase 

profit from software service use. The GQM + Strategies approach 

enforces the explicit documentation of the relevant 

context factors and assumptions necessary for understanding 

and evaluating each goal. For Goal 1, one such 

context factor is that how often customers access ABC 

software products determines the amount of revenue generated 

at ABC. 

The GQM + Strategies goal template documents other 

details. Each template includes the desired magnitude of 

improvement, the time frame for achieving the goal, the 

scope of responsibility for achieving the goal, and any 

constraints or conflicting goals as well as relationships to 

other goals. Figure 3 shows a partial template for Goal 1. 

The GQM graph in Figure 2 is based on the measurement 

goal, G1. In full GQM notation, G1 is 

Analyze the trend in profit for the purpose of evaluation 

with respect to a 10 percent increase in annual income per 

year from the point of view of ABC’s management in the 

context of the ABC organization. 

This goal leads to questions Q1 and Q2: What is the current 

profit (measured by P 0 ), and what is the profit for each 

succeeding year (as measured by P x )? The decision maker 

analyzes the results using the criteria incorporated in the 

interpretation model (far right), which says that 

Starting in year 2, if the profit for the current year (P ) is 2 

at least 10 percent (1.1 times) higher than the profit for the 

initial preceding year (P ), then the goal has been satisfied. 

1 

Activity: Increase 

Focus: Net income 

Object: ABC Web services 

Magnitude: 10% per year 

Time frame: Annually, beginning in 2 years 

Scope: Development groups assessed at CMMI level 2 or higher 

Constraints: Available resources, ability to sustain CMMI levels 

Relations: CMMI-related goals 

Figure 3. Business goal “Increase pro t from software service usage” 

expressed in the GQM + Strategies goal template. Business goals are at the 

top level of the GQM + Strategies model. They become measurable through 

the derivation of one or more GQM graphs, which are then associated 

with other GQM graphs at other organizational levels. 

The full interpretation model, not shown for simplicity, 

includes an “else” part related to the effectiveness of the 

chosen strategies. 

The strategies associated with each goal come from a 

list of potential strategies that the GQM + Strategies user 

has enumerated, taking into account influencing context 

factors. Possible strategies for meeting Goal 1 are to deliver 

added capabilities to encourage heavier system use, 

increase rates charged to customers, or reduce development 

costs. The user opted for the strategy of delivering 

added functionality in the product releases at regular and 

frequent intervals. At this point, the user must also make 

explicit the assumption that this added functionality will 

lead to increased customer satisfaction, which will in turn 

lead to heavier use. 

The assumption is that ABC has enough projects with 

a Capability Maturity Model Integration maturity level 

greater than 1 that, if just those projects provide a 15 percent 

improvement, ABC will see a 10 percent improvement 

overall. In general, all context information and assumptions 

are attached to goals, strategies, and their relations 

(as Figure 1 shows) and are documented in a corresponding 

graph (not shown). 

Software-development goals 

At the next level down in the model is a goal derived 

from the strategy (or strategies) chosen at the top level. 

Goal 2 is to deliver a new release of the software every six 

months that incorporates at least 5 percent more functionality 

than the previous release and to keep the cost 

of this increased functionality to within 10 percent of the 

current budget. 

These are really two interrelated but separate software 

goals, so the user defines them in separate GQM + Strategies 

goal templates. Figure 4 elaborates the first of these goals 

(increasing functionality by 5 percent). At this level, the 

goal is specific to software development, so it is treated as 

APRIL 2010 

A 


aG 

F 

61

A 


aG 

F 

62 


Activity: Deliver 

Focus: More usable functionality, for example, M (must) type requirements 

from the backlog of customer-requested requirements 

Object: Each release of ABC Web services software 

Magnitude: 5% more functionality than the prior release 

Time frame: Every 6 months, beginning in 2 years 

Scope: Web services development projects with CMMI level 2 or higher 

Constraints: Available resources, ability to sustain CMMI levels, ability to estimate 

cost, and schedule for a release 

Relations: Achievement of cost and schedule estimate accuracy, ability to improve 

CMMI levels of development groups 

Figure 4. Software goal template for the goal “Deliver 5 percent new 

functionality every six months.” A separate but related software goal, 

which would have its own template, would address the 10 percent cost 

variance. Both these goals are derived from a higher-level strategy that 

addresses the business goal of increasing revenue from increased use of 

ABC’s software products. 

for x = 2, 3, … 

if Px ≥ 1.1 * Px–1 then 

the goal has been satisfied, 

else if functionality was increased appropriately, then 

either some assumption is incorrect or we have chosen the wrong level of strategy. 

Figure 5. Re ning the interpretation model from Figure 2 by adding 

an “else” part. Full interpretation—the exact reason that a particular 

goal was not satis ed—depends on goals at lower levels. This iterative 

re nement ensures that goals at all levels are linked and makes goal 

interrelationships explicit. 

a goal at the software level, as opposed to one at the business 

level. In general, the name accorded the lower levels 

depends on the organization applying GQM + Strategies and 

the number of levels to be modeled. 

The software goal template asks for the same categories 

of information as the business goal template in Figure 

3. The user defines the measurement and interpretation 

model for this goal and refines the interpretation model 

for the business goal. As Figure 5 shows, the refinement of 

Figure 2’s interpretation model involves adding an “else” 

part that recognizes that full interpretation depends on the 

lower-level goals. For example, if the functionality was not 

increased by 5 percent, perhaps the strategy associated 

with Goal 2 was not effective. The interpretation model becomes 

increasingly detailed, with more conditional logic, 

at each lower level in the overall GQM + Strategies model. 

The organization must decide to develop and carry out a 

strategy for accomplishing a software goal. The strategy for 

Goal 2 in Figure 2 is to adopt an approach such as rating the 

importance of different requirements, as in the MoSCoW 

prioritization approach 10 for requirements and release planning, 

and to adopt the Constructive Cost Model (Cocomo) 11 

for cost estimation. The user must explicitly document an 

COMPUTER 

important context factor relevant to this strategy: 

Someone with expertise in the MoSCoW approach 

(a consultant who recommends it) is available, but 

ABC has no staff with such experience. 

Three assumptions relevant to this strategy, 

which the user must also document, follow: 

The organization can estimate the percentage 

of functionality delivered, for example, by 

using a proxy such as additional lines of code 

delivered, number of function points delivered, 

or a formula based on an actual requirements 

count. 

The difficulty and importance of requirements 

are weighted in some way (hard, medium, 

easy) to provide input to the cost model. 

The backlog of customer-requested requirements 

continues to grow. 

As with the business goal, the software goal 

has an associated set of GQM structures that 

define how to evaluate the goal (G2 in Figure 2). 

In full GQM notation, G2 is 

Analyze each six-month release for the purpose 

of evaluation with respect to incorporation 

of 5 percent new functionality as compared to the 

previous release from the point of view of the Web 

services project manager in the context of the ABC 

organization. 

This goal leads to questions Q3 and Q4: How many new 

requirements of must (M) importance were included in 

each release? How much time elapsed between releases? 

The interpretation of this software goal’s achievement is 

If, at each six-month milestone, the growth in functionality 

of a release is 5 percent, then the level 2 goal is 

satisfied for this release; else, assumptions about MoSCoW 

are incorrect or we have not chosen the correct strategy. 

At this point, the user can also further refine the business 

goal’s interpretation. If the business goal is satisfied 

(ABC’s profit increased by 10 percent), but Goal 2 is not, 

then the assumptions are wrong. The delivery of a particular 

requirement alone might have caused the gain, for 

example. 

Project-specific goals 

The last goal level in Figure 2 consists of goals derived 

from the previous level’s strategy that apply to a particular 

software development group or project. In the sample 

application, Goal 3 is to apply the MoSCoW and Cocomo 

approaches effectively. The relevant development group 

A 


aG 

F

A 


aG 

F 

has developed a strategy for this goal that involves training 

personnel, acquiring tools that will assist in applying these 

methods, and piloting these methods on a single project. A 

relevant assumption is that training for these approaches 

can be targeted to a few specific individuals, so the impact 

of the training on cost and schedule is reasonable. 

The GQM graph at this level, only part of which is shown 

in Figure 2, involves evaluating the effectiveness of using 

MoSCoW and Cocomo and of the training and other tools. 

The GQM + Strategies user then uses these results to revisit 

the interpretation of higher-level goals. Some of the questions 

and metrics defined at the second level are reusable 

at subsequent levels—a benefit of basing GQM + Strategies 

on the GQM approach. 

SUPPORTING STRATEGIC MEASUREMENT 

In all cases, specifying a strategy for goal achievement 

creates explicit links among goals at the various levels, from 

business objectives to project operations, which is critical 

to strategic measurement. In the example, the highest-level 

goal was to increase profit, the next-level goal was to provide 

new functionality in short releases, and the lowest-level 

goal was to use specific application methods to achieve the 

previous level goals. The linking strategy specified that ABC 

will achieve its increased profit goal by providing customers 

with more functionality. Often such links are implicit, but 

making them explicit has many benefits. 

Templates 

Templates also support strategic measurement by guiding 

users in defining all goal types at the necessary detail 

level. In the example, the full template for the business 

goal partially described in Figure 3 includes more extensive 

information about the target increase in income, 

the time frame, and any constraints or conflicting goals. 

GQM + Strategies includes templates for all goal types in the 

measurement model. 

Tracking assumptions 

Although the ABC example did not show it, our approach 

also builds in the ability to track context factors 

and assumptions at each level (through the requirement 

to document them along with all strategy documentation) 

so that users can more easily assess the effects of 

changes in either context or assumptions. In the example, 

the approach required that users document the assumption 

about the training required for MoSCoW and Cocomo 

so that, if the assumption turns out to be false, the model 

will indicate what elements the assumption affects and, 

consequently, what requires reevaluating. 

Interpretation models 

Interpretation models tie together measurement goals, 

context factors, assumptions, and data, thus making it 

easier to correctly and usefully interpret measurement 

results. This idea stems from the original GQM approach, 

but we have broadened it to allow interpretation models 

at each level to inform not only that level, but also higher 

levels as the data is aggregated and rolled up. In the ABC 

example, the results of applying the interpretation model 

at the lowest level will yield information about how the 

piloting of MoSCoW and Cocomo went, as well as information 

about the training and tools. The information from the 

lower level can help diagnose any problems encountered at 

the next level up. Further, the results of the interpretation 

at the second level can inform the analysis at the top level. 

At the top level, if profit does not increase as expected, the 

analytical results from the second level will help determine 

if the problem is due to higher costs, inadequate functionality 

delivered, late releases, or some other cause. 

Our approach explicitly links goals 

at different levels, from business 

objectives to project operations, 

which is critical to strategic 

measurement. 

Flexibility 

Our approach is flexible enough to accommodate other 

software measurement approaches and business strategy 

approaches. ABC might have used BSC to define its business 

goals and strategies and then used that description as 

a starting point for applying GQM + Strategies. PSM might 

have been useful later in applying our approach, when ABC 

needed to define the metrics. 

Shared measurement planning and experience 

Because all goal levels are linked, measurement planning 

and results are organization-wide rather than limited 

to a single project or department. Often the results are 

lower costs and better return on the investment in a measurement 

program, greater likelihood that the program 

will succeed, more effective risk identification and management, 

and greater compliance with software process 

improvement models such as CMMI. 

A substantial benefit is that sharing makes it more 

feasible to build a measurement experience base that 

can become a valuable corporate asset. Such a base can 

make project measurement and planning easier over time 

and consequently decrease project costs. It might begin 

with the set of generic experiences already available in 

GQM + Strategies and add value as the organization populates 

it with organization-specific models. 

APRIL 2010 

A 


aG 

F 

63

A 


aG 

F 

64 


Any part of the ABC model in Figure 2 is reusable. The 

assumptions and context factors are particularly important 

because they capture the properties that might not 

hold when the organization reuses the model for a different 

situation. However, because the model requires the explicit 

capture of assumptions and context factors and links these 

to particular goals and strategies, the organization can 

clearly see which parts of the model need to be reevaluated 

when an assumption or context factor changes. The 

result is the flexibility to adapt goals and strategies to particular 

market needs and to analyze the consequences to 

the organization. 

APPLICATION SCENARIOS 

We have already taken steps to apply the GQM + Strategies 

approach in various industrial settings. Although our approach 

focuses on software development, it is finding 

equal utility in broader contexts, as the applications in 

Table 1 illustrate. 

A few studies are ongoing to get more insights into 

the deployment of GQM + Strategies. So far, most of the 

studies have reached the phase of initially setting up a 

GQM + Strategies grid, but they have not yet progressed to 

the point of using and maintaining it. 

The GQM + Strategies approach provides many 

features for organizations that want to create a 

software measurement program that is consistent 

with and contributes to the achievement of 

goals at all organizational levels. Our sample 

application illustrates the use of GQM + Strategies in a 

single context with three goal levels. Other situations 

might call for additional levels, but generally the process 

of capturing multilevel goals includes a business goal 

level, a set of lower-level goals, and a project-specific goal 

level. Each of these levels, in turn, can have multiple peer 

goals. Regardless of how many levels the organization 

COMPUTER 

Table 1. Ongoing applications of the GQM + Strategies approach. 

Business Domain Application 

European telecommunications 

company 

Telecommunications Drive strategic improvement programs, support paradigm 

shift toward purpose-driven metrics 

European automotive supplier Automotive Support CMMI’s Measurement and Analysis process area 

European network testing 

company 

International software 

company 

Telecommunications Evaluate cost, benefit, and schedule for modernizing 

existing product suite 

Embedded systems used in 

telecommunications 

Increase the visibility at all organizational levels of how 

strategic decisions impact operations 

Asian insurance company Information systems Align strategies and goals for new business domain 

Asian systems engineering 

organization 

Joint research project to 

develop a common software 

platform 

Safety-critical software for aerospace 

application 

Support of complex, dynamic business processes 

in a variety of domains, including logistics, 

retail, and customized industrial facilities 

Increase visibility of goals and strategies and derived 

measurement goals to enhance supplier collaboration 

Align project objectives and business objectives of involved 

research and industry partners 

requires, all levels must be based on well-defined goals, 

each with an associated strategy, documented context 

factors and assumptions, and a measurement and evaluation 

framework. 

The most important benefit of applying GQM + Strategies 

is the resulting transparency of measurement motivations 

and goals at different organizational levels, which makes 

it easier to identify goal relationships and conflicts and 

facilitates communication for organizational segments, 

such as marketing and software development. In the ABC 

example, without the GQM + Strategies model, project personnel 

might have misdirected their training efforts or 

chosen the wrong tools. The model helped project personnel 

understand why they were being asked to implement 

MoSCoW and Cocomo, which was likely to produce more 

focused training or tailoring. This is one of the many benefits 

of using our approach in a short development cycle. 


GQM + Strategies is registered trademark No. 

302008021763 at the German Patent and Trade Mark 

Office; international registration number IR992843. 

References 

1. V. Basili, G. Caldiera, and D. Rombach, “Goal, Question, 

Metric Paradigm,” Encyclopedia of Software Engineering, 

vol. 1, J.J. Marciniak, ed., John Wiley & Sons, 1994, pp. 

528-532. 

2. V. Basili et al., “Bridging the Gap between Business Strategy 

and Software Development,” Proc. Int’l Conf. Information 

Systems, Association for Information Systems Electronic 

Library; http://aisel.aisnet.org/icis2007/25. 

3. R. Kaplan and D. Norton, “The Balanced Scorecard—Measures 

That Drive Performance,” Harvard Business Rev., 

Jan.-Feb. 1992, p. 71. 

4. US Dept. of Defense and US Army, Practical Software and 

Systems Measurement: A Foundation for Objective Project 

Management, v. 4.0c, Mar. 2003; www.psmsc.com. 

A 


aG 

F

A 


aG 

F 

5. S.A. Becker and M.L. Bostelman, “Aligning Strategic and 

Project Measurement Systems,” IEEE Software, May/June 

1999, pp. 46-51. 

6. L. Buglione and A. Abran, “Balanced Scorecards and GQM: 

What Are the Differences?” Proc. 3rd European Software 

Measurement Conf., Federation of European Software Measurement 

Assoc., 2000, pp. 18-20. 

7. A.J. Bianchi, “Management Indicators Model to Evaluate 

Performance of IT Organizations,” Proc. Int’l Conf. Management 

of Engineering and Technology, vol. 2, IEEE Press, 

2001, pp. 217-229. 

8. D. Card, “Integrating Practical Software Measurement and 

the Balanced Scorecard,” Proc. 27th Ann. Int’l Computer 

Software and Applications Conf., IEEE CS Press, 2003, p. 

362. 

9. Sarbanes-Oxley Act of 2002, Public Law No. 107-204, 116 

Stat. 745, codified in sections of 11, 15, 18, 28, and 29 in 

US Code, 30 July 2002. 

10. D. Clegg and R. Barker, Case Method Fast-Track: A RAD 

Approach, Addison-Wesley Professional, 1994. 

11. B.W. Boehm et al., Software Cost Estimation with COCOMO 

II, Prentice Hall, 2000. 

Victor R. Basili is a professor in the Department of Computer 

Science at the University of Maryland and chief 

scientist of the Fraunhofer Center for Experimental 

Software Engineering. His research interests include measuring, 

evaluating, and improving the software process 

and product via empirical studies. Basili received a PhD in 

computer science from the University of Texas. Contact him 

at basili@fc-md.umd.edu. 

_______________ 

Mikael Lindvall is a senior scientist and division director 

at the Fraunhofer Center for Experimental Software 

Engineering. His research interests include agile methods, 

software process improvement, software architectures, and 

experience and knowledge management. Lindvall received 

a PhD in computer science from Linköping University, 

Sweden. Contact him at mikli@fc-md.umd.edu. 

Myrna Regardie is a senior engineer at the Fraunhofer 

Center for Experimental Software Engineering. Her research 

interests include software process improvement, 

measurement, and project and knowledge management. 

Regardie received a BS in mathematics from Juniata College, 

Huntingdon, Pennsylvania. Contact her at ________ 

mregardie@ 

fc-md.umd.edu. 

___________ 

Carolyn Seaman is an associate professor of information 

systems at the University of Maryland, Baltimore County, 

and a scientist at the Fraunhofer Center for Experimental 

Software Engineering. Her research interests include 

software evolution, management of software projects, 

and qualitative research methods in software engineering. 

Seaman received a PhD in computer science from the 

University of Maryland. Contact her at ___________ 

cseaman@fc-md. 

umd.edu. 

______ 

_______________ 

Jens Heidrich is a department head at the Fraunhofer Institute 

for Experimental Software Engineering. His research 

interests include project management, quality assurance, 

and measurement. Heidrich received a PhD in computer 

science from the University of Kaiserslautern, Germany. 

Contact him at jens.heidrich@iese.fraunhofer.de. 

______________________ 

Jürgen Münch is division manager for quality management 

at the Fraunhofer Institute for Experimental Software Engineering. 

His research interests include quality assurance, 

process management, and measurement. Münch received 

a PhD in computer science from the University of Kaiserslautern. 

Contact him at ______________________ 

juergen.muench@iese.fraunhofer. 

__ de. 

Dieter Rombach is a professor and chair of the Software 

Engineering Department at the University of Kaiserslautern 

and executive director of the Fraunhofer Institute 


interests include software methodologies, modeling and 

measurement of the software process and resulting products, 

software reuse, and distributed systems. Rombach 

received a PhD in computer science from the University 

of Kaiserslautern. Contact him at _______________ 

dieter.rombach@iese. 

fraunhofer.de. 

_________ 

Adam Trendowicz is a researcher at the Fraunhofer Institute 


interests include software cost modeling, measurement, 

and process improvement. Trendowicz received a PhD in 

computer science from the University of Kaiserslautern. 

Contact him at adam.trendowicz@iese.fraunhofer.de. 

________________________ 



Join the 

IEEE 


www.computer.org 

APRIL 2010 

A 


aG 

F 

65

A 


aG 

F 

66 

COMPUTER SOCIETY CONNECTION 

Ken Birman Wins 

Kanai Award 

Ken Birman 

is a founding 

member of 

TRUST, a 

consortium 

that explores 

challenges in trustworthy computing. 

Kenneth P. Birman, N. 

Rama Rao Professor 

at Cornell University, 

recently received the 

IEEE Computer Society’s 2009 

Tsutomu Kanai Award. The award 

recognizes major contributions to 

state-of-the art distributed computing 

systems and their applications. 

Birman was recognized “for funda- 

COMPUTER 

mental and practical contributions 

to distributed computing, fault tolerance, 

reliability and distributed 

systems management.” 

Birman’s work has focused on the 

development of trustworthy distributed 

computing systems. Early in his 

career, he developed the Isis Toolkit, a 

reliable group communication system 

that introduced the virtual synchrony 

model for fault tolerance. The widely 

adopted Isis was at the core of such 

mission-critical systems as the 

French air traffic control system, the 

New York Stock Exchange, and the US 

Navy’s Aegis-class warships. 

Birman’s group subsequently 

developed a series of systems that 

explored challenges of extreme 

COMPUTING 

THEN 

Learn about computing history 

and the people who shaped it. 

http://computingnow. 

computer.org/ct 

scale using gossip and peer-topeer 

protocols. These included 

Horus, Ensemble, Bimodal Multicast, 

the Astrolabe platform, and 

the Gossip Objects platform. Ideas 

and technology from these efforts 

have helped shape modern cloud 

computing systems, including 

the communication layer of IBM’s 

flagship WebSphere product, Microsoft’s 

cluster management platform, 

and Amazon’s data-center management 

systems. 

Birman became a Fellow of the 

ACM in 1998 and won the 2009 

IEEE Transactions on Parallel and 

Distributed Systems Outstanding 

Achievement Award. 

Tsutomu Kanai Award 

The Tsutomu Kanai Award was 

established in 1997 by an endowment 

from Hitachi in honor of its 

president. The award consists of 

a crystal model, certificate, and 

$10,000 honorarium. The IEEE 

Computer Society Awards Committee 

considers the seminal nature 

of the achievements, their practical 

impact, breadth, and depth, as 

well as the quality of the nomination. 

The awards honor technical 

achievements as well as service 

to the computer profession and to 

the Society. Birman will accept his 

award at the Computer Society’s 

2010 awards ceremony in Denver. 

The deadline to make a nomination 

for the 2010 Tsutomu Kanai 

Award is 15 October. For more information, 

visit www.computer.org/ 

portal/web/awards/kanai. 

________________ 


A 


aG 

F

A 


aG 

F 

Francine Berman Wins Kennedy Award 

Francine Berman 

was one of 

two founding 

principal 

investigators 

on the National 

Science 

Foundation’s TeraGrid Project. 

rancine Berman, vice 

president for research at 

Rensselaer Polytechnic 

Institute, recently received 

the IEEE Computer Society’s Ken 

Kennedy Award for outstanding 

contributions to programmability 

or productivity in high-performance 

computing. She is a pioneer in grid 

computing and a leading advocate 

for the development of a national 

cyberinfrastructure for the access, 

use, stewardship, and preservation 

of digital data. Berman’s work has 

had a major impact on the direction 

of computational science and 

the cyberinfrastructure. Her citation 

reads “For her influential leadership 

in the design, development and 

deployment of national-scale cyber 

infrastructure, her inspiring work 

as a teacher and mentor, and her 

exemplary service to the high- 

performance community.” 

Berman is co-chair of the Blue 

Ribbon Task Force on Sustainable 

Digital Preservation and Access, 

an international group focusing 

on the economic sustainability of 

digital information that must be 

accessed and preserved for many 

decades. 

In 2001, Berman became director 

of both the San Diego Supercomputer 

Center and the National Partnership 

for Advanced Computational Infrastructure, 

a consortium of more than 

40 national and international partners 

who worked together to create a 

comprehensive national computing 

infrastructure. 

Berman is a founding member and 

co-chair of the Computing Research 

Association’s Committee on the Status 

of Women in Computing Research 

and currently serves on the Anita 

Borg Institute for Women and Technology 

Board of Trustees. 

Before moving to RPI, Berman 

held the High Performance Computing 

Endowed Chair in the 

Jacobs School of Engineering at 

the University of California, San 

Diego. In 2000, she was named an 

ACM Fellow for pioneering work in 

application scheduling for parallel 

distributed computing. Berman 

received a BA in mathematics from 

the University of California, Los 

Angeles, and an MA and PhD in 

Top Educators Honored 

he IEEE Computer Society 

sponsors an active 

and prestigious awards 

program as part of its 

mission to promote the free exchange 

of ideas among computer professionals 

around the world and to recognize 

its members for their outstanding 

accomplishments. Several noted 

educators recently received two 

Computer Society awards that honor 

achievement in education. 

COMPUTER SCIENCE 

& ENGINEERING 

UNDERGRADUATE 

TEACHING AWARD 

The IEEE Computer Society 

Computer Science & Engineering 

Undergraduate Teaching Award is 

presented each year for outstanding 

contributions to undergraduate education 

through teaching and service, 

for helping to maintain interest in 

the field, and for making a statement 

mathematics from the University 

of Washington. 

Ken Kennedy Award 

The IEEE Computer Society Ken 

Kennedy Award was established 

in memory of the founder of Rice 

University’s nationally ranked computer 

science program and one of 

the world’s foremost experts on 

high-performance computing. A certificate 

and $5,000 honorarium are 

awarded jointly by the ACM and the 

Computer Society for outstanding 

contributions to programmability 

or productivity in high-performance 

computing together with significant 

community service or mentoring 

contributions. 

about the importance with which 

the Society views undergraduate 

education. 

Judy Robertson 

Judy Robertson of Heriot-Watt University, 

the 2009 award winner, was 

honored “for outstanding contributions 

to the undergraduate education 

through teaching and the innovative 

use of pioneering technologies in 

teaching.” Robertson is the principal 

investigator of a grant funded by the 

UK’s Engineering and Physical Sciences 

Research Council that supports 

high school teachers who use gamemaking 

projects with their students. 

She received a BS and PhD in computer 

science and artificial intelligence 

from the University of Edinburgh. 

Elizabeth Burd 

Elizabeth Burd of Durham University, 

the 2008 winner, was recognized 

“for outstanding contributions to 

APRIL 2010 

A 


aG 

F 

67

A 


aG 

F 

68 

COMPUTER SOCIETY CONNECTION 

Microsoft and IEEE have announced 

a collaboration that aims to 

increase the number of students 

engaged in technical pursuits. The IEEE 

Computer Society, which has long 

očered its student members access to 

free Microsoft software, is participating 

in the collaboration. 

“As an information provider to technical 

professionals, this collaboration 

with Microsoft will give us an expanded 

ability to introduce students to more 

internships, scholarships, and contests,” 

said Violet Doan, the Computer Society’s 

membership director. 

“Engineers make tremendous contributions 

to society by creating technologies 

that drive global economic 

growth,” said Walid Abu-Hadba, vice 

president of developer and platform 

evangelism at Microsoft. “The Microsoft-IEEE 

collaboration will provide 

aspiring engineers around the world 

undergraduate education through 

teaching and the organization of 

programs to promote excellence in 

undergraduate teaching.” Burd is 

the leader of Durham’s Technology 

Enhanced Learning Research Group. 

Her most recent grant, supported by 

the Teaching and Learning Research 

Program, focuses on investigating 

the use of multitouch software in 

classrooms. 

Call for Papers | General Interest 

I 

EEE Micro seeks general-interest submissions 

for publication in upcoming issues. These 

works should discuss the design, performance, 

or application of microcomputer and microprocessor 

systems. Of special interest are articles on 

performance evaluation ion and 

workload character- 

COMPUTER 

MICROSOFT AND IEEE COOPERATE 

TO SUPPORT STUDENTS 

with tools and resources to convert their 

innovative thinking into marketplace 

innovation by equipping today’s technical 

students for tomorrow’s jobs.” 

Microsoft and IEEE will work together 

to provide IEEE student members 

with access to Microsoft servers, 

development tools, and eLearning, 

including privileged access to more 

than 300 software titles as part of the 

MSDN Academic Alliance, a subscription 

service traditionally available only 

through educational institutions. An 

added benećt is the opportunity for 

science, technology, engineering, and 

math students to explore job opportunities 

with Microsoft partners and 

customers through the Microsoft Students 

to Business program. 

Learn more about IEEE Computer Society 

student activities at www.computer. 

org/portal/web/studentactivities/ 

____ home. 

TAYLOR L. BOOTH 

AWARD 

Recipients of the Taylor L. Booth 

Award are presented with a bronze 

medal and $5,000 honorarium in 

recognition of an outstanding record 

in computer science and engineering 

education. A successful candidate 

must meet at least two of the following 

criteria in the computer science 

and engineering field: 

ization. Summaries of work in progress and descriptions 

of recently completed works are most 

welcome, as are tutorials. Micro does not accept 

previously published material. 

Check our author center (www.computer.org/mc/ 

_________ 

mi cro/author.htm) for word, fi gure, and reference 

limits. All submissions pass through peer review 

co consistent istent with 

other professional-level technical 

pu publications, ublications, and eediting 

for clarity, readability, and 

co conciseness. Contac Contact IEEE Micro at _____ micro-ma@ 

computer.org with aany 

questions. 

Achieving recognition as a 

teacher of renown. 

Writing an influential text. 

Leading, inspiring, or providing 

significant education content 

during the creation of a curriculum 

in the field. 

Inspiring others to a career in 

computer science and engineering 

education. 

Michael Heath 

Michael Heath, chair of computer 

science at the University of Illinois 

at Urbana-Champaign, was selected 

in 2009 “for contributions to computational 

science and engineering 

education, curriculum and scholarship.” 

Heath is director of both the 

computational science and engineering 

program and the Center for 

Simulation of Advanced Rockets at 

UIUC. He was named an ACM Fellow 

in 2000. 

James P. Cohoon 

and Jack W. Davidson 

James P. Cohoon and Jack W. 

Davidson of the University of 

Virginia were recognized as winners 

in 2008 “for sustained effort 

to transform introductory computer 

science education through 

lab-based multimedia pedagogy 

coupled with examples that attract 

a diverse student body.” Cohoon’s 

research involves algorithms, probabilistic 

search, genetic algorithms, 

simulated annealing, and diversity 

in computer science education. 

Davidson performs research in 

programming languages, computer 

security, embedded systems, and 

computer architecture. 

Taylor L. Booth founded the 

University of Connecticut’s 

Booth Engineering Center 

for Advanced Technology and was a 

candidate for IEEE Computer Society 

president at the time of his death in 

1986. For more information on Computer 

Society awards, visit ____ 

www. 

computer.org/portal/web/awards. 

______________________ 

A 


aG 

F

A 


aG 

F 

CALL AND CALENDAR 

CALLS FOR ARTICLES FOR 

IEEE CS PUBLICATIONS 

IEEE Micro seeks contributions 

on a wide range of topics related to 

multicore processing for a September/ 

October 2010 special issue on European 

multicore processing projects. 

Suitable topics include advancements 

in multicore architecture, interconnection 

networks, and design-space 

exploration. 

Articles are due by 19 April. 

Visit www.computer.org/portal/web/ 

________________ 

computingnow/micfp5 to view the 

complete call for papers. 

IT Professional seeks papers for a 

January/February 2011 issue focusing 

on green IT. 

Green IT is, and will continue to 

be, a hot topic due to increasing 

awareness of the harmful effects of 

green gas emissions, new stringent 

environmental legislation, rising concern 

about electronic waste disposal 

practices that damage the environment, 

and corporate-image concerns 

that are pushing businesses and individuals 

to adopt green practices. 

Articles are due by 1 July. Visit 

www.computer.org/portal/web/ 

_______________ 

computingnow/itcfp1 to view the 

complete call for papers. 

CALLS FOR PAPERS 

ICEBE 2010, IEEE Int’l Conf. on 

e-Business Eng., 20-22 October, 

Shanghai; papers due 20 April; ____ 

http:// 

conferences.computer.org/icebe/2010/ 

cfp.htm 

____ 

CW 2010, Int’l Conf. on Cyberworlds, 

20-22 October, Singapore; papers due 

30 April; ________________ 

www3.ntu.edu.sg/SCE/ 

cw2010/cfp.htm 

___________ 

0018-9162/10/$26.00 © 2010 IEEE 

CALENDAR 

MAY 

2-3 May: VG 2010, IEEE/EG Int’l 

Symp. on Volume Graphics, Norrkoping, 

Sweden; www.eurographics2010. 

se/vg2010 

_______ 

3-6 May: ASYNC 2010, IEEE Int’l 

Symp. on Asynchronous Circuits 

and Systems, Grenoble, France; ____ http:// 

asyncsymposium.org/async2010 

17-20 May: CCGrid 2010, IEEE Int’l 

Symp. on Cluster Computing and 

the Grid, Southbank, Australia; ____ www. 

manjrasoft.com/ccgrid2010 

28-31 May: CSO 2010, 3rd Int’l Joint 

Conf. on Computational Sciences 

and Optimization, Huangshan Mountain, 

China; www.gip.hk/cso2010 

JUNE 

9-11 Jun: EuroVis 2010, IEEE Eurographics 

Symp. on Visualization, 

Talence, France; www.eurovis.org 

9-12 Jun: CCC 2010, IEEE Int’l Conf. 

on Computational Complexity, Cambridge, 

Massachusetts; http://facweb. 

cs.depaul.edu/jrogers/complexity 

______________________ 

SUBMISSION INSTRUCTIONS 

The Call and Calendar section lists conferences, symposia, and workshops that the IEEE 

Computer Society sponsors or cooperates in presenting. 

Visit www.computer.org/conferences for instructions on how to submit conference 

or call listings as well as a more complete listing of upcoming computer-related 

conferences. 


28-30 Jun: WET ICE 2010, IEEE 

Int’l Workshops on Enabling 

Technologies: Infrastructures for 

Collaborative Enterprises, Larissa, 

Greece; http://wetice.org 

AUGUST 

17-19 Aug: MASCOTS 2010, Int’l 

Symp. on Modeling, Analysis and 

Simulation of Computer and Telecommunications 

Systems, Miami; 

http://mascots2010.cis.fiu.edu 

20-22 Aug: SocialCom 2010, IEEE Int’l 

Conf. on Social Computing, Bloomington, 

Minnesota; www.iisocialcom. 

org/conference/socialcom2010 

23-26 Aug: ICPR 2010, 20th Int’l 

Conf. on Pattern Recognition, Istanbul; 

www.icpr2010.org 

24-26 Aug: EMS 2010, Int’l Conf. on 

Eng. Management and Service Sciences, 

Wuhan, China; www.scirp.org/ 

_________ 

conf/ems2010 

SEPTEMBER 

12-18 Sep: ICSM 2010, Int’l Conf. on 

Software Maintenance, Timisoara, 

Romania; http://icsm2010.upt.ro 

20-24 Sep: Cluster 2010, IEEE Int’l 

Conf. on Cluster Computing, Heraklion, 

Crete; www.cluster2010.org 

OCTOBER 

3-8 Oct: MODELS 2010, IEEE Model- 

Driven Eng. Languages and Systems, 

Oslo, Norway; http://models2010.ifi. 

uio.no 

APRIL 2010 

A 


aG 

F 

____ 

69

A 


aG 

F 

70 

COMPUTER 

CALL AND CALENDAR 

EVENTS IN 2010 

May 

2-3 ..........................VG 2010 

3-6 ......................ASYNC 2010 

17-20 ....................CCGrid 2010 

28-31 .......................CSO 2010 

June 

9-11 .................... EuroVis 2010 

9-12 ........................CCC 2010 

28-30 ..................WET ICE 2010 

August 

17-19 .................MASCOTS 2010 

20-22 . . . . . . . . . . . . . . . . SocialCom 2010 

23-26 ...................... ICPR 2010 

24-26 .......................EMS 2010 

20-22 Oct: CW 2010, Int’l Conf. on 

Cyberworlds, Singapore; _______ 

www3.ntu. 

____________ 

edu.sg/sce/cw2010 

20-22 Oct: ICEBE 2010, IEEE Int’l 

Conf. on e-Business Eng., Shanghai; 

http://conferences.computer. 

org/icebe 

2 Free Sample Issues! 

A $26 value 

The magazine of computational tools 

and methods for 21st century science. 

Cloud Computing 

Computational Astrophysics 

Computational Nanoscience 

Computational Engineering 

Geographical Information 

Systems 

http://cise.aip.org | www.computer.org/cise 

Send an e-mail to jbebee@aip.org to receive the two 

most recent issues of CiSE. (Please include your mailing address.) 

Recent Peer-Reviewed Topics: 

SC 2010 

stablished 21 years ago, the SC conference has built a community of participants 

E that includes researchers, scientists, computing center stač members, IT and data 

center managers, application developers, computer builders, program managers, 

journalists, and congressional stačers. 

The technical program addresses virtually every area of scientific and engineering 

research as well as technological development, innovation, and education. Its presentations, 

tutorials, panels, and discussion forums have contributed to breakthroughs in 

many new and innovative areas of computing. 

SC 2010 is sponsored by the IEEE Computer Society Technical Committee on Scalable 

Computing in cooperation with ACM SIGARCH. The conference takes place 13-19 November 

in New Orleans. Visit www.sc-conference.org for complete conference details. 

23-26 Oct: FOCS 2010, Symp. on 

Foundations of Computer Science, 

Las Vegas; http://theory.stanford. 

_________ 

edu/focs2010/ 

NOVEMBER 

1-3 Nov: SRDS 2010, IEEE Int’l Symp. 

on Reliable Distributed Systems, 

New Directions 

Petascale Computing 

Reproducible Research 

Software Engineering 

New Delhi, India; www.scs.ryerson. 

________________ 

ca/iwoungan/SRDS2010 

13-19 Nov: SC 2010, Int’l Conf. for 

High-Performance Computing, Networking, 

Storage, and Analysis, 

New Orleans; www.sc-conference. 

org 

______________ 

MEMBERS 

$47/year 

for print & online 

A 


aG 

F 

___

A 


aG 

F 

PURPOSE: The IEEE Computer Society is the world’s largest 

association of computing professionals and is the leading 

provider of technical information in the field. 

MEMBERSHIP: Members receive the monthly magazine 

Computer, discounts, and opportunities to serve (all activities 

are led by volunteer members). Membership is open to all IEEE 

members, affiliate society members, and others interested in the 

computer field. 

COMPUTER SOCIETY WEB SITE: www.computer.org 

OMBUDSMAN: To check membership status or report a change 

of address, call the IEEE Member Services toll-free number, 

+1 800 678 4333 (US) or +1 732 981 0060 (international). Direct 

all other Computer Society-related questions—magazine delivery 

or unresolved complaints—to ____________ 

help@computer.org. 

CHAPTERS: Regular and student chapters worldwide provide the 

opportunity to interact with colleagues, hear technical experts, 

and serve the local professional community. 

AVAILABLE INFORMATION: To obtain more information on any 

of the following, contact Customer Service at +1 714 821 8380 or 

+1 800 272 6657: 

Membership applications 

Publications catalog 

Draft standards and order forms 

Technical committee list 

Technical committee application 

Chapter start-up procedures 

Student scholarship information 

Volunteer leaders/staff directory 

IEEE senior member grade application (requires 10 years 

practice and significant performance in five of those 10) 

PUBLICATIONS AND ACTIVITIES 

Computer: The flagship publication of the IEEE Computer Society, 

Computer, publishes peer-reviewed technical content that 

covers all aspects of computer science, computer engineering, 

technology, and applications. 

Periodicals: The society publishes 13 magazines, 18 transactions, 

and one letters. Refer to membership application or request 

information as noted above. 

Conference Proceedings & Books: Conference Publishing 

Services publishes more than 175 titles every year. CS Press 

publishes books in partnership with John Wiley & Sons. 

Standards Working Groups: More than 150 groups produce 

IEEE standards used throughout the world. 

Technical Committees: TCs provide professional interaction in 

more than 45 technical areas and directly influence computer 

engineering conferences and publications. 

Conferences/Education: The society holds about 200 

conferences each year and sponsors many educational activities, 

including computing science accreditation. 

Certifications: The society offers two software developer 

credentials. 

For more information, visit www.computer.org/certification. 

revised 20 Jan. 2010 

EXECUTIVE COMMITTEE 

President: James D. Isaak* 

President-Elect: Sorel Reisman* 

Past President: Susan K. (Kathy) Land, CSDP* 

VP, Standards Activities: Roger U. Fujii (1st VP)* 

Secretary: Jeffrey M. Voas (2nd VP)* 

VP, Educational Activities: Elizabeth L. Burd* 

VP, Member & Geographic Activities: Sattupathu V. Sankaran† 

VP, Publications: David Alan Grier* 

VP, Professional Activities: James W. Moore* 

VP, Technical & Conference Activities: John W. Walz* 

Treasurer: Frank E. Ferrante* 

2010–2011 IEEE Division V Director: Michael R. Williams† 

2009–2010 IEEE Division VIII Director: Stephen L. Diamond† 

2010 IEEE Division VIII Director-Elect: Susan K. (Kathy) Land, CSDP* 

Computer Editor in Chief: Carl K. Chang† 

* voting member of the Board of Governors † nonvoting member of the Board of Governors 

BOARD OF GOVERNORS 

Term Expiring 2010: Piere Bourque; André Ivanov; Phillip A. Laplante; 

Itaru Mimura; Jon G. Rokne; Christina M. Schober; Ann E.K. Sobel 

Term Expiring 2011: Elisa Bertino, George V. Cybenko, Ann DeMarle, 

David S. Ebert, David A. Grier, Hironori Kasahara, Steven L. Tanimoto 

Term Expiring 2012: Elizabeth L. Burd, Thomas M. Conte, Frank E. 

Ferrante, Jean-Luc Gaudiot, Luis Kun, James W. Moore, John W. Walz 

EXECUTIVE STAFF 

Executive Director: Angela R. Burgess 

Associate Executive Director; Director, Governance: Anne Marie Kelly 

Director, Finance & Accounting: John Miller 

Director, Information Technology & Services: Carl Scott 

Director, Membership Development: Violet S. Doan 

Director, Products & Services: Evan Butterfield 

Director, Sales & Marketing: Dick Price 

COMPUTER SOCIETY OFFICES 

Washington, D.C.: 2001 L St., Ste. 700, Washington, D.C. 20036 

Phone: 

Email: ____________ 

hq.ofc@computer.org 

Los Alamitos: 10662 Los Vaqueros Circle, Los Alamitos, CA 90720-1314 

Phone: +1 714 821 8380 

Email: ___________ 

help@computer.org 

Membership & Publication Orders: 

Phone: 

Email: ___________ 

help@computer.org 

Asia/Pacific: Watanabe Building, 1-4-2 Minami-Aoyama, 

Minato-ku, Tokyo 107-0062, Japan 

Phone: 

Email: _____________ 

tokyo.ofc@computer.org 

IEEE OFFICERS 

President: Pedro A. Ray 

President-Elect: Moshe Kam 

Past President: John R. Vig 

Secretary: David G. Green 

Treasurer: Peter W. Staecker 

President, Standards Association Board of Governors: 

W. Charlston Adams 

VP, Educational Activities: Tariq S. Durrani 

VP, Membership & Geographic Activities: Barry L. Shoop 

VP, Publication Services & Products: Jon G. Rokne 

VP, Technical Activities: Roger D. Pollard 

IEEE Division V Director: Michael R. Williams 

IEEE Division VIII Director: Stephen L. Diamond 

President, IEEE-USA: Evelyn H. Hirt 

Next Board Meeting: 

11 June 2010, Denver, CO, USA 

A 


aG 

F

A 


aG 

F 

72 

CAREER OPPORTUNITIES 

SOFTWARE ENGINEER: High Performance 

Computing, Maxeler Technologies. 

Maxeler Technologies provides 

software and hardware systems which 

accelerate high performance scientific 

computing applications by an order of 

magnitude. Tier 1 companies currently 

use Maxeler consulting and hardware 

solutions to gain a competitive advantage. 

We’re currently seeking software 

engineers with experience of high performance 

software development (C/ 

C++/Java/FORTRAN) for positions based 

in London. See http://www.maxeler. 

com/careers/ for more information or to 

apply. 

HEWLETT-PACKARD COMPANY has an 

opportunity for the following position in 

Cupertino, CA. PPM Developer: Reqs exp 

with: HP Project & Portfolio Mgmt s/w 

with the following modules: Demand 

Mgmt, Project Mgmt, Program Mgmt, 

Resources Mgmt, Finance Mgmt, Change 

Mgmt;Oracle PL/SQL; Shell Scripting; 

Process design; Organization change 

mgmt. Also requires Bachelors degree in 

CS, CE or rel field of study and 4 yrs. exp 

in job offered or rel. Send resume & refer 

to Job#CUPAHO2.: Please send resumes 

COMPUTER 

with job number to Hewlett-Packard 

Company, 19483 Pruneridge Ave., MS 

4206, Cupertino, CA 95014. No phone 

calls please. Must be legally authorized 

to work in the U.S. without sponsorship. 

EOE. 

PROJECT MANAGER, Senior Solutions 

Architect New York, NY. Utilize knowl. of 

complete SDLC & OO, JSF & Struts. Concepts 

to lead & participate in the design, 

development & maint. of complex software 

systems. Design & develop Java/ 

J2EE/SOA applicatns using RSA, WID & 

WPS, BPEL, BPM, Webservices, SOAP/ 

JMS. Familiar w/Datapower SOA Appliance. 

BS/MS-Engineering or foreign 

equivalent + related exp. Resumes: UTC 

Associates, Inc., 80 Wall St, Suite 1118, 

NY, NY 10005. 

HP ENTERPRISE SERVICES, LLC is accepting 

resumes for the following position 

in Oklahoma City, OK: Services 

Information Developer (Ref. #HPESOCF- 

WA1). SW design, develop, implementation, 

consulting & testing. Project requirement 

analysis. Requires Master’s 

or foreign degree equivalent in Comp 

NEW TENURE TRACK FACULTY POSITION 

IN THE SCHOOL OF SYSTEMS AND ENTERPRISES 

AT STEVENS INSTITUTE OF TECHNOLOGY 

Faculty Position in Software Engineering 

The School of Systems and Enterprises at Stevens Institute of Technology invites applications 

for a full-time, tenure-track faculty position in Software Engineering. We are especially interested 

in candidates with research interests in modeling, simulation and visualization of complex 

systems; and synergies with systems engineering and security systems engineering. The 

successful candidate will have the ability to develop an innovative and impactful research 

program in this area. Furthermore, an enthusiasm for teaching and conducting collaborative 

research is essential. 

An application, formatted as a single .pdf file, should include the candidate’s curriculum 

vitae, a statement of research interest, a statement of teaching interest, three scientific manuscripts, 

and up to 5 references. 

This appointment will be at the Associate or Assistant Professor Rank, dependent on scholarly 

achievement and accomplishments. Applications should be submitted to: 

Jon Wade, Ph.D., Professor and Associate Dean of Research 

School of Systems and Enterprises, Stevens Institute of Technology 

1 Castle Point on Hudson, Hoboken, NJ 07030 

Email: jon.wade@stevens.edu Tel: 775.636.5811 

_____________ 

This position will remain open until filled. Screening of candidates will start in 

April 2010. Stevens Institute of Technology is an equal opportunity/affirmative 

action employer and is dedicated to recruiting a diverse faculty community. 

We welcome all qualified applicants to apply, including women, minorities, 

veterans, and individuals with disabilities. 

www.stevens.edu/sse 

Cisco Systems, Inc. is accepting resumes for 

the following position: 

Richfield, OH 

Software Engineer 

(Ref#: RICH1) 

Responsible for the definition, design, development, 

test, debugging, release, enhancement 

and maintenance of networking software. 

Please mail resumes with reference number to 

Cisco Systems, Inc., Attn: J51W, 170 W. Tasman 

Drive, Mail Stop: SJC 5/1/4, San Jose, CA 95134. 

No phone calls please. Must be legally 

authorized to work in the U.S. without sponsorship. 

EOE. 

www.cisco.com 



Malvern, PA 

Solution Consultant 

(Ref#: MAL1) 

Provide specific solution, technology, and 

product consulting for software product. 

Please mail resumes with reference number 

to Cisco Systems, Inc., Attn: J51W, 170 W. 

Tasman Drive, Mail Stop: SJC 5/1/4, San Jose, 

CA 95134. No phone calls please. Must be 

legally authorized to work in the U.S. without 

sponsorship. EOE. 

www.cisco.com 


A 


aG 

F

A 


aG 

F 

Sci, Comp Eng, Electrical Eng, or related 

field + 4 yrs of exp in job offered, or as 

a programmer analyst, engineer, developer, 

or related occupation. Oracle SQL; 

Javascript; SOAP/Web Services; XML; 

AJAX; ASP.NET; C#. Please mail resumes 

with reference # to: Ref. # HPESOCF- 

WA1, Hewlett-Packard Company; 19483 

Pruneridge Avenue, MS 4206, Cupertino, 

CA 95014. No phone calls. Must be legally 

authorized to work in the U.S. without 


TECHNICAL CONSULTANT (Milpitas, 

CA) Provide prof’l computer consulting 

svcs in form of systems analysis, dsgn & 

dvlpmt, systems integration &/or testing 

consulting. Min req MS in Comp Sci. 

Resume: HR, GSPANN Technologies, Inc, 

362 Fairview Way, Milpitas, CA 95035. 

HEWLETT-PACKARD COMPANY is accepting 

resumes for a Software Designer 

in Vancouver, WA. (Ref. #VANSWD11). 

Design, develop, maintain, test, and provide 

quality/performance assurance of 

embedded software for enterprise class 

print and finishing devices. Mail resumes 

with reference # to: Ref. #VANSWD11, 

Hewlett-Packard Company, 19483 

Pruneridge Avenue, MS 4206, Cupertino, 

CA 95014. No phone calls please. Must 

be legally authorized to work in the U.S. 

without sponsorship. EOE. 

ADJILITY CONSULTING in Short Hills, 

NJ seeks Documentum Systems Engineer 

for Documentum imaging systems 

including configuration & issue troubleshooting 

& configuring application for 

system optimizations. Req Bachelor’s & 

5 yrs exp with EMC Documentum Content 

Server, EMC Documentum Captiva, 

EMC Documentum .Net & Java development, 

2 yrs exp with Windows platforms 

& Unix platforms, working knowledge of 

infrastructure topics and EMC Documentum 

System Administrator certification. 

Requires up to 100% travel to client locations 

w/in U.S. Email resume to _____ 

38.adjil- 

___________ 

ity@hiredesk.com. 

.NET DEVELOPER - Dsgn, dvlp, test 

& implmt applic s/w utilizing knowl of 

& exp w/ .NET 2.0/3.0/3.5, C#, VB.NET, 

AJAX, Classic ASP, VB 6, C++, IIS, Oracle 

8i/9i, SQL Server 2000/2005/2008, 

Java, Tomcat, JavaScript, Web Services, 

Cisco Systems, Inc. is accepting resumes for the following 

positions in San Jose/Milpitas/Santa Clara, CA 

IT Analyst (Ref#: SJ6) 

Responsible for evaluating and documenting business needs, recommending 

business process and Information Technology solution alternatives and 

coordinating delivery of technical solutions to clients. 

Network Consulting Engineer (Ref#: SJ9) 

Responsible for the support and delivery of Advanced Services to company’s 

major accounts. 

Manager, Advanced Services (Ref#: SJ52) 

Supervise the activities of a technical team with responsibility for results in 

terms of customer satisfaction and delivery assurance. 

Please mail resumes with reference number to Cisco Systems, Inc., Attn: J51W, 170 

W. Tasman Drive, Mail Stop: SJC 5/1/4, San Jose, CA 95134. No phone calls please. 

Must be legally authorized to work in the U.S. without sponsorship. EOE. 

www.cisco.com 

Georgia State University 

Department of Computer Science 

The Department of Computer Science of Georgia 

State University invites applications for an anticipated 

tenure-track position for Assistant or Associate 

Professor beginning the Fall semester, 2010, 

pending budgetary approval. Earned Ph.D. in Computer 

Science, or a closely related discipline, and 

excellent records of publications and funding are 

required. Preference is for individual with specialty 

in distributed systems and networking. An offer 

of employment will be conditional on background 

verification. 

Applicants should send: letter of interest, C.V. without 

birth date, but with citizenship status, and three 

letters of recommendation and transcripts of all 

graduate work to: 

Dr. Yi Pan, Chair 

Department of Computer Science 

Georgia State University 

34 Peachtree Street, Suite 1450 

Atlanta, Georgia, 30303 

Applications can also be sent via email to ___ pan@ 

______ cs.gsu.edu and will be accepted until position is 

filled. 

Georgia State University, a Research University of 

the University System of Georgia, is an AA/EEO 

employer. 

Assistant Professor – 

Computer Engineering 


The Department of Electrical and Computer Engineering 

(ECE) at the University of Pittsburgh 

(http://www.engr.pitt.edu/electrical/about/jobs. 

___ html) is searching for a tenure stream assistant 

professor in the area of Computer Engineering. 

The start date for this position is approximately 

Sept. 1, 2010. The preferred candidate will fill a 

core need in the area of “Computer Engineering,” 

with a demonstrated research focus is on one of the 

following Swanson School of Engineering Focus 

Areas: “Energy,” “Bioengineering,” “Manufacturing” 

(including RFID), “Sustainability,” or “Nanoscience 

and Engineering.” ECE partners with the 

Department of Computer Science to offer BS, MS 

and PhD degrees in Computer Engineering. 

To ensure full consideration, applications must be 

received by April 15, 2010, although later applications 

may be considered if time allows. Please 

send a resume (or CV), contact information for at 

least four references, plus a letter summarizing your 

teaching and research interests to: 

Chair, ECE Search Committee 

348 Benedum Engineering Hall, 


Pittsburgh, PA 15261 

or alternatively to: __________ 

eesearch@pitt.edu 

The University of Pittsburgh is an affirmative action, 

equal opportunity employer. 

APRIL 2010 73 

A 


aG 

F

A 


aG 

F 

74 

CAREER OPPORTUNITIES 



San Bruno/San Francisco, CA 

Quality Assurance 

Engineer 

(Ref#: SBR4) 

Responsible for all aspects of Quality Assurance 

and testing for multiple projects. 

Please mail resumes with reference number to 

Cisco Systems, Inc., Attn: J51W, 170 W. Tasman 

Drive, Mail Stop: SJC 5/1/4, San Jose, CA 95134. 


authorized to work in the U.S. without sponsorship. 

EOE. 

www.cisco.com 

Advertiser Page 

APC 1 

Cisco 72-75 

Georgia State University 73 

IEEE Computer Society CV3, CV4 

ISADS 2011 12 

Stevens Institute of Technology 72 

University of Massachusetts 9 

University of Pittsbrugh 73 

Classified Advertising 72-75 

Advertising Personnel 

Marion Delaney 

IEEE Media, Advertising Dir. 

Phone: +1 415 863 4717 

Email: ____________ 

md.ieeemedia@ieee.org 

Marian Anderson 

Sr. Advertising Coordinator 

Phone: +1 714 821 8380 

Fax: +1 714 821 4010 

Email: ______________ 

manderson@computer.org 

Sandy Brown 

Sr. Business Development Mgr. 

Phone: +1 714 821 8380 

Fax: +1 714 821 4010 

Email: ____________ 

sb.ieeemedia@ieee.org 

COMPUTER 

Entrust Truepass, Red Dot, PVCS, Win- 

Runner; Req MS Comp Sci, Eng or rel. 

Mail resumes to HR-National Computer 

Systems, 6-70 Towne Center Dr , North 

Brunswick, NJ 08902. 


opportunity for the following position in 

Houston, TX. Engineering Project Manager: 

Reqs: Ability to use lab equipmt 

incl Oscilloscopes, DMMS (Digital Multimeters), 

Pwr Meters, Gain/Phase analyzers. 

Exp. w/impedance analyzers; Data 

acquisition systms. Also requires Bachelors 

degree in Elec. Engrng, or rel field 

of study and 3 yrs. exp in job offered or 

rel. Send resume & refer to Job# HOUK- 

MI2.: Please send resumes with job number 

to Hewlett-Packard Company, 19483 

Pruneridge Ave., MS 4206, Cupertino, 

CA 95014. No phone calls please. Must 

be legally authorized to work in the U.S. 



opportunity for the following position 

in Houston, TX. Systems/Software Engineer: 

Reqs: 7yrs s/w dvlpmt exp on Linux 

&/or UNIX; C/C++; SCSI, SES and SAS; FC, 

SATA & RAID storage; SMI-S & CIM; SM- 

BIOS & IPMI; UNIX make, shell scripts & 

Perl; version cntrl systems; OOAD, design 

COMPUTER 

Advertising Sales 

Representatives 

Recruitment: 

Mid Atlantic 

Lisa Rinaldo 

Phone: +1 732 772 0160 

Fax: +1 732 772 0164 

Email: _______ 

lr.ieeemedia@ 

ieee.org 

New England 

John Restchack 

Phone: +1 212 419 7578 

Fax: +1 212 419 7589 

Email: ______ 

j.restchack@ 

ieee.org 

Southeast 

Thomas M. Flynn 

Phone: +1 770 645 2944 

Fax: +1 770 993 4423 

Email: _____ flynntom@ 

mindspring.com 

Midwest/Southwest 

Darcy Giovingo 

Phone: +1 847 498 4520 

Fax: +1 847 498 5911 

Email: _______ 

dg.ieeemedia@ 

ieee.org 

Northwest/Southern CA 

Tim Matteson 

Phone: +1 310 836 4064 

Fax: +1 310 836 4067 

Email: _______ 

tm.ieeemedia@ 

ieee.org 

Japan 

Tim Matteson 

Phone: +1 310 836 4064 

Fax: +1 310 836 4067 

Email: _______ 

tm.ieeemedia@ 

ieee.org 

Europe 

Heleen Vodegel 

Phone: +44 1875 825700 

Fax: +44 1875 825701 

Email: _____ impress@ 

impressmedia.com 

patterns & UML; HP-UX internals; SOAP/ 

XML; multi-threaded applctns; Java; 

storage bnchmrkng tools. Also requires 

Bach degree in CS, CE or rel & 7 yrs exp 

in job offered or rel. Send resume & refer 

to Job #HOUPPU2. Please send resumes 

with job number to Hewlett-Packard 

Company, 19483 Pruneridge Ave., MS 

4206, Cupertino, CA 95014. No phone 

calls please. Must be legally authorized 

to work in the U.S. without sponsorship. 

EOE. 

HP ENTERPRISE SERVICES, LLC is accepting 

resumes for the following positions: 

TECHNOLOGY CONSULTANT IN 

FARMINGTON HILLS, MI. (Ref. #HPESF- 

HVPA1). Create & implement standard 

deployment image for using Windows 

XP & Windows 7. Script solutions, independent 

of deployment tools used, to 

deploy standard image across globe. 

Requires Master’s or foreign degree 

equiv. in Electrical Engg, Comp Engg, 

Comp Sci, or related + 2 yrs experience 

in job offered, or as programmer analyst, 

intern, or related. Hewlett Packard 

Configuration Automation Enterprise; 

Visual Basic Scripting; Microsoft Windows 

Image Mgmt; Microsoft Windows 

Software Distribution; Microsoft Active 

Directory. 

Product: 

US East 

Dawn Becker 

Phone: +1 732 772 0160 

Fax: +1 732 772 0164 

Email: _______ 

db.ieeemedia@ 

ieee.org 

US Central 

Darcy Giovingo 

Phone: +1 847 498 4520 

Fax: +1 847 498 5911 

Email: ___________ 

dg.ieeemedia@ieee.org 

US West 

Lynne Stickrod 

Phone: +1 415 931 9782 

Fax: +1 415 931 9782 

Email: ___________ 

ls.ieeemedia@ieee.org 

Europe 

Sven Anacker 

Phone: +49 202 27169 11 

Fax: +49 202 27169 20 

Email: _____ sanacker@ 

__________ 

intermediapartners.de 

A 


aG 

F

A 


aG 

F 

BUSINESS CONSULTANT IN GREENS- 

BURG, PA. (Ref. #HPESSCH1). Interact 

with client face to face on daily basis. Innovate 

& formulate business solutions & 

“technology enabled” business models 

& new ways of doing business to create 

breakthrough positions in marketplace. 

Master’s or foreign degree equiv. in 

Comp. Sci, Comp. Engg., Electrical Engg, 

or related + 2 yrs of exp in job offered, 

or as SW consultant, SW engineer, or related. 

SAP Project Mgmt; SAP ABAP; SAP 

Netweaver XI; SAP EDI / ALE / IDOCS; SAP 

Scripts – Smartforms; SAP Workflow. 

Please mail resumes with reference # 

to: Ref. #, HP Enterprise Services; 19483, 

Pruneridge Avenue, MS 4206, 95014. 


authorized to work in the U.S. without 


SR. DEVELOPER - eCommerce Industries, 

Inc. (ECI) has an opening in Big Sandy, 

TX for Sr. Developer. Position involves 

designing & developing complex software 

applications. Design, plan, develop, 

test & document accounting & office 

products software, applying knowledge 

of programming techniques & computer 

systems. Analyze user request for new 

or modified functions for our financial/ 

accounting system analyze & develop 

specifications for the new design, to determine 

feasibility, cost & time required 

compatibility with the current system 

& computer capabilities. Consult with 

end users to identify current operating 

procedures & clarifying program objectives. 

Study relevant manuals, periodicals 

& technical reports to learn ways to develop 

software that meet user requirements. 

Formulate plans outlining steps 

required to develop programs, using 

structured analysis, designing & object 

oriented methodology. Prepare flowcharts 

& class diagrams to illustrate the 

sequence of steps a program must follow 

& describing the logical operations 

involved. Convert project specifications, 

using flowcharts & class diagrams into a 

sequence of detailed instructions & logical 

steps for coding into office product 

software language which is written in 

C++, JAVA, Windows, UNIX, LINUX, SQL 

Server, .Net, Visual Basic, SQL, Web appli- 

cations development networking. Code & 

software module design. Write documentation 

to describe program development, 

logic, coding & corrections & rectifying 

bugs. Develop understanding of the project 

scope & requirements & question any 

areas that are unclear. Identify potential 

problem areas in software design. Participate 

in training of new members of the 

department. Assist Customer Support 

& Technical Support in providing a customer 

response or assume responsibility 

for calls that have been escalated to Software 

Development. Requires Bachelor’s 

in Computer Science or CIS & 3 yrs exp. in 

JAVA programming OR Master’s in Computer 

Science or CIS & 1 year exp. in industry 

or as a graduate research assistant 

in JAVA programming. Submit your resume 

to ___________________ 

developmentjobs@ecisolutions. 

___ com. An EOE. 

PROGRAMMER - eCommerce Industries, 

Inc. (ECI) has an opening in Fort 

Wayne, IN for Programmer. Position involves 

planning, developing, testing & 

documenting accounting & office products 

software, applying knowledge of 

programming techniques & computer 

systems. Analyzing user request for new 

or modified functions for our financial/ 

accounting system to determine feasibility, 

cost & time required, compatibility 

with the current system & computer capabilities. 

Consulting with end users to 

identify current operating procedures & 

clarifying program objectives. Studying 

relevant manuals, periodicals & technical 

reports to learn ways to develop software 

that meet user requirements. Preparing 

flowcharts & class diagrams to illustrate 

sequence of steps a program must follow 

& describing logical operations involved. 

Converting project specifications into a 

sequence of detailed instructions & logical 

steps for coding into office product 

software language which is written in 

FoxPro & Microsoft .NET languages (C#, 

Visual Basic) using SQL server. Writing 

documentation to describe program development, 

logic, coding & corrections 

& rectifying bugs. Requires Bachelor’s 

in Computer Science or equivalent & 2 

yrs exp. in computer programming. Exp. 

to include Visual Basic & FoxPro. Submit 

your resume to _____________ 

britdevjobs@ecisolutions.com. 

An EOE. 

______ 

SUBMISSION DETAILS: Rates are $445.00 per column inch ($500 minimum). 

Eight lines per column inch and average five typeset words per line. Free 

online listing on careers.computer.org with print ad. Send copy at least one 

month prior to publication date to: Marian Anderson, Classified Advertising, 

Computer Magazine, 10662 Los Vaqueros Circle, Los Alamitos, CA 90720-1314; 

(714) 821-8380; fax (714) 821-4010. Email: _________________ 

manderson@computer.org. 



San Bruno/San Francisco, CA 

Customer Support Engineer 

(Ref#: SBR2) 

Responsible for providing technical support 

regarding the company's proprietary systems 

and software. 





legally authorized to work in the U.S. 


www.cisco.com 



Boston, MA 

Systems Engineering Manager 

(Ref#: BOS1) 

Provide technical direction and business 

guidance to the regional sales team. 





legally authorized to work in the U.S. 


www.cisco.com 

APRIL 2010 75 

A 


aG 

F

A 


aG 

F 

76 

COMPUTER 

AI REDUX 

Probabilistic Analysis 

of an Ancient 

Undeciphered Script 

Rajesh P.N. Rao, University of Washington 

Probabilistic methods for analyzing sequences are providing 

new insights into the 4,000-year-old undeciphered script of the 

Indus civilization. 

In the latter half of the 19th 

century, railway workers in 

British India found an almost 

inexhaustible supply of precisely 

cut baked bricks at Harappa, 

a small town located in present-day 

Pakistan. They proceeded to use the 

bricks as ballast for laying down 100 

miles of railroad track. Little did they 

know that these bricks were more 

than 4,000 years old, products of 

South Asia’s oldest urban civilization. 

The Indus civilization, so named 

because its first important sites were 

discovered along the Indus River, 

flourished from 2600 to 1900 BC. At 

its peak, it encompassed more than 

1 million square kilometers and was 

larger than the roughly contemporaneous 

Egyptian and Mesopotamian 

civilizations. Its cities were laid out 

in a grid-like pattern with a sophisticated 

water management and 

drainage system that would be the 

envy of many towns today. Citizens 

of the Indus civilization were highly 

enterprising, traveling to lands as far 

away as the Persian Gulf and Mesopotamia 

(present-day Iraq) to trade. 

Remarkably, there is no evidence 

that powerful kings or elites ruled the 

Indus cities, as in other Bronze Age 

civilizations. No extravagant royal 

palaces, pyramids, or ziggurats have 

been found. What archaeologists 

have unearthed in large numbers are 

tiny seals like those shown in Figure 

1a, most around 1” × 1” in size. Each 

typically depicts an expertly crafted 

animal, with a short text of signs 

at the top. These texts, which also 

appear on miniature tablets, copper 

plates, tools, weapons, and pottery, 

constitute the Indus script, one of the 

last remaining undeciphered scripts 

of the ancient world. 

THE INDUS SCRIPT 

Figure 1b shows a small subset of 

the approximately 400 signs in the 

Indus script. The number of signs 

is more than in purely alphabetic 

or syllabic scripts, which typically 

contain a few dozen signs, but less 

than in logographic scripts such 

as Chinese, which contain large 

numbers of signs representing 

entire words. Researchers have 

therefore suggested that, like other 

ancient scripts such as Sumerian 

and Mayan, the Indus script was 

logosyllabic in nature, each sign 

representing either a word or a 

syllable. 

What the Indus signs actually 

mean remains a mystery, although 

the number of books claiming to have 

deciphered the script could occupy 

several bookshelves. None of these 

claims have been widely accepted. 

The major impediments to decipherment 

include 

the brevity of existing Indus 

texts—the average text length is 

about five signs while the longest 

text consists of 17 signs; 

our almost complete lack of 

knowledge of the language 

spoken by the Indus people; and 

the lack of a bilingual document 

such as the Rosetta Stone, 

which was instrumental in 

deciphering the Egyptian hieroglyphic 

script. 

Given such formidable obstacles, 

efforts to decipher the script have 

ranged from inspired guesswork to 

ideology-driven speculation. 

An alternate, more objective 

approach is to first analyze the script’s 

syntactic structure, in the hope that 


A 


aG 

F

A 


aG 

F 

such an analysis could eventually lead 

to decipherment. Are the symbols in 

Indus texts randomly ordered or do 

they follow specific rules? Do particular 

symbols have particular positions 

within texts? How much flexibility 

does the script allow when composing 

a string of symbols? How do the 

Indus script’s syntactical properties 

compare with those of other ancient 

and modern languages and scripts? 

Researchers are investigating such 

questions using statistics, probabilistic 

reasoning, and machine learning. 

EARLY STATISTICAL 

ANALYSIS 

G.R. Hunter conducted the first 

rudimentary statistical analysis of the 

Indus script in the early 1930s. In the 

absence of computers, Hunter handenumerated 

frequently occurring 

clusters of signs, segmenting Indus 

texts into short “words” of two or 

more signs. This enabled him to infer 

important syntactic characteristics 

of the script such as the tendency of 

certain symbols and words to occur 

at specific positions within texts. For 

example, Hunter was among the first 

to note that the “jar” sign , which is 

the most frequently occurring sign 

in the texts, acts as a “word ender,” 

and that the “fish” signs frequently 

occur in pairs (such as and ), 

occupying the same relative position 

within texts. 

In the 1960s, the fact that sign clusters 

have particular positions within 

Indus texts was confirmed independently 

with the help of computers by a 

Finnish team led by Asko Parpola and 

a Soviet team led by Yuri Knorozov 

(who played a key role in deciphering 

the Mayan script). More recent work 

has demonstrated that the frequency 

of certain two-, three-, and four-sign 

combinations is much higher than 

would be expected by chance, and 

that a majority of the texts longer 

than five signs can be segmented into 

these smaller, frequently occurring 

sign combinations (N. Yadav et al., 

“Segmentation of Indus Texts,” Int’l 

(a) (b) 

Figure 1. Indus script. (a) Three examples of square stamp seals, each with an 

Indus text at the top (image credit: J.M. Kenoyer/Harappa.com). Texts were usually 

written from right to left (inferred, for example, from writing on pottery where a 

sign is overwritten by another on its left) but this direction was reversed in seals 

(that is, left to right as in these images) to form correctly oriented impressions. (b) 

A small subset of the 400 or so signs in the Indus script (selected from Mahadevan’s 

concordance). 

J. Dravidian Linguistics, vol. 37, no. 

1, 2008, pp. 53-72). Such regularities 

point to the existence of distinctive 

syntactic rules underlying the Indus 

texts. 

MARKOV AND 

NfiGRAM MODELS 

The presence of statistically significant 

clusters of symbols with 

positional preferences suggests that 

there is sequential order in the Indus 

script. One way to capture such 

sequential order is to learn a Markov 

model for the script from available 

texts. 

The simplest (first-order) model 

estimates the transition probabilities 

P(s i |s j ) that sign i follows sign j. The 

obvious way of estimating P(s i |s j ) is 

to count the number of times sign i 

follows sign j, an approach equivalent 

to maximum likelihood estimation. 

However, given that there are approximately 

400 signs and only a few 

thousand texts, a large number of 

sign pairs will have a frequency of 0 

even though their actual probability 

may not necessarily be 0. This is a 

common problem in statistical language 

modeling and can be addressed 

using smoothing techniques. 

A prominent smoothing technique, 

the modified Kneser-Ney algorithm, 

was used to learn a first-order Markov 

model of the Indus script (R.P.N. Rao 

et al., “A Markov Model of the Indus 

Script,” Proc. National Academy of 

Sciences, vol. 106, no. 33, 2009, pp. 

13685-13690). The data for training 

the model came from Iravatham 

Mahadevan’s The Indus Script: Texts, 

Concordance and Tables (Archaeological 

Survey of India, 1977). Once trained, 

the Markov model can be used to 

generate new samples of Indus texts. 

This can reveal interesting subunits of 

grammatical structure and recurring 

patterns, as Figure 2a shows. 

There exist a large number of damaged 

Indus seals, tablets, and other 

artifacts that contain texts with one 

or more missing or illegible signs. A 

Markov model of the Indus texts can 

APRIL 2010 

A 


aG 

F 

77

A 


aG 

F 

78 

COMPUTER 

Log likelihood 

AI REDUX 

0 

–10 

–20 

–30 

–40 

(a) (b) 

Indus A B C D 

(c) 

Figure 2. Markov model of the Indus script. (a) (Top) A new Indus text generated by 

the Markov model. (Below) Two closest matching texts in the training corpus. (b) 

(Left) Text from a damaged seal containing one or more missing signs (indicated 

by the shaded box). (Right) Three possible restorations predicted by the Markov 

model. The črst and third texts actually exist in the corpus. (c) Log likelihood under 

the Markov model for four texts (A through D) found in foreign lands compared to 

average log likelihood for a random set of 50 Indus region texts not included in the 

training data (error bar denotes +/- 1 standard error of mean). The 50 Indus region 

texts had the same average length as the foreign texts. 

be used to predict these missing or 

illegible signs. The first-order Markov 

model was found to be surprisingly 

good at predicting signs deliberately 

obliterated for testing purposes, performing 

at a 75 percent accuracy 

level in a fivefold cross-validation 

study. Figure 2b shows an example 

of restoration of an actual damaged 

Indus inscription from Mahadevan’s 

concordance, as suggested by the 

first-order Markov model. 

Several seals with Indus signs have 

been discovered outside the Indus 

region, as far away as Mesopotamia 

and the Persian Gulf. One can compute 

the likelihood of these “foreign” 

texts with respect to a Markov model 

trained only on texts from the Indus 

region. As Figure 2c shows, the likelihood 

values for many of these foreign 

texts are several orders of magnitude 

lower than those for Indus region 

texts, indicating their low probability 

of belonging to the same language. 

Indeed, an examination of these 

foreign texts reveals that although 

they contain commonly used Indus 

signs, the sequential order of the signs 

differs dramatically from that in texts 

originating in the Indus region—for 

example, the sequence in the foreign 

text C in Figure 2c never occurs 

on an Indus seal. This suggests that 

A 

B 

C 

D 

Indus traders in foreign lands may 

have used the script to represent different 

content, such as foreign names 

or goods, or an altogether different 

language. 

More recent work examined the 

utility of higher-order N-gram models. 

An N-gram model is essentially an 

(N 1)th-order Markov chain where 

the transition probability depends on 

the previous N 1 symbols instead 

of just the previous symbol. The 

results suggest that a bigram model 

(N = 2) captures a significant portion 

of the syntax, with trigrams and 

quadrigrams making more modest 

contributions (N. Yadav et al., “Statistical 

Analysis of the Indus Script 

Using N-Grams,” PLoS One, to appear 

in 2010). 

THE LANGUAGE QUESTION 

AND ENTROPIC ANALYSIS 

The brevity of existing Indus 

inscriptions and other attributes, 

such as the low frequency of many 

Indus signs, has prompted some 

to propose that the Indus script is 

not a script at all but instead is a 

collection of religious or political 

symbols. Adherents of the “nonscript” 

thesis have likened the Indus 

script to nonlinguistic systems such 

as traffic signs, medieval heraldry, 

markings on pottery in the Vinc a 

culture of southeastern Europe, 

and carvings of deities on boundary 

stones in Mesopotamia. 

Interestingly, this is not the first 

time that a script of a major ancient 

civilization has been deemed to be 

nonlinguistic. The Mayan script was 

long considered not to be a writing 

system at all until Knorozov and 

others finally worked out the rich 

phonetic underpinnings of the script 

in the 1950s and 1960s, revealing 

it to be a fully functional writing 

system. 

Several key features of the Indus 

script suggest that it represents 

language: 

the texts are usually linear, like 

the vast majority of linguistic 

scripts and unlike nonlinguistic 

systems such as heraldry or traffic 

signs; 

symbols are modified by the 

addition of specific sets of marks 

over, around, or inside a symbol, 

much like later Indian scripts 

that use such marks to modify 

the sound of a root consonant or 

vowel symbol; 

the script possesses rich syntactic 

structure, with particular 

signs or clusters of signs preferring 

particular positions 

within texts, similar to linguistic 

sequences; 

the script obeys the Zipf- 

Mandelbrot law, a power-law 

distribution on ranked data, 

which is often considered a necessary 

(though not sufficient) 

condition for language; and 

texts found in Mesopotamia and 

the Persian Gulf use the same 

signs as texts found in the Indus 

region but alter their ordering, 

suggesting that the script was 

versatile enough to represent 

different subject matter or a different 

language. 

Such attributes are hard to reconcile 

with the thesis that the script 

A 


aG 

F

A 


aG 

F 

merely represents religious or political 

symbols. 

Further evidence for the Indus 

script’s linguistic nature comes from 

quantitative studies comparing the 

entropy of the Indus texts with that 

of various languages. In some nonlinguistic 

systems, such as the Vinc a 

system, the signs do not seem to 

follow any order and appear to be 

juxtaposed randomly. Other nonlinguistic 

systems, such as deities 

carved on Mesopotamian boundary 

stones, exhibit a rigid order reflecting, 

for example, the hierarchical order of 

the deities. 

In languages, on the other hand, 

sequences of words and characters 

exhibit a degree of order intermediate 

between random and rigid. This 

intermediate degree of randomness 

arises from the grammatical rules 

and morphological structure of languages. 

The degree of randomness in 

a sequence can be measured quantitatively 

using entropy. 

The smoothed first-order Markov 

model can be used to compute conditional 

entropy, which measures the 

average flexibility allowed in choosing 

the next sign given a preceding 

sign. The conditional entropy of Indus 

texts has been shown to fall within the 

range of natural languages (R.P.N. Rao 

et al., “Entropic Evidence for Linguistic 

Structure in the Indus Script,” Science, 

vol. 324, no. 5931, 2009, p. 1165). 

A potential shortcoming of the 

conditional entropy result is that it 

only captures pairwise dependencies. 

Figure 3 shows new results 

(presented here for the first time) 

on higher-order entropies for blocks 

of up to six symbols. These block 

entropies were calculated using 

the state-of-the-art NSB estimator 

(I. Nemenman, F. Shafee, and 

W. Bialek, “Entropy and Inference, 

Revisited,” Advances in Neural Information 

Processing Systems 14, MIT 

Press, 2002, pp. 471-478), which has 

been shown to provide good estimates 

of entropy for undersampled 

discrete data. 

Normalized block entropy 

6 

5 

4 

3 

2 

1 

0 

1 

Indus 

Max Ent 

DNA 

Protein 

Tamil 

Eng chars 

Eng words 

Sansk 

Tagalog 

Sumer 

Fortran 

Min Ent 

The new results in Figure 3 extend 

the conditional entropy result to 

sequences of length up to six: The 

block entropies of the Indus texts 

remain close to those of a wide range 

of natural languages and far from 

the entropies for randomly and rigidly 

ordered sequences (Max Ent and 

Min Ent, respectively). Also shown in 

the plot for comparison are the entropies 

for a computer program written 

in Fortran and two sample biological 

sequences (DNA and proteins). The 

Fortran program and the biological 

sequences have noticeably lower and 

higher block entropies, respectively, 

than the Indus script and natural 

languages. 

Entropic similarity to natural languages 

by itself is not sufficient to 

prove that the Indus script is linguistic. 

However, given that it exhibits 

other key features of linguistic scripts 

as enumerated above, this similarity 

increases the probability in a 

Bayesian sense that the Indus script 

represents language. 

2 3 

4 5 6 

Sequence length (block size) 

Figure 3. Entropy of the Indus script compared to natural languages and other 

sequences. Symbols were signs for the Indus script; bases for DNA; amino acids for 

proteins; characters for English; words for English, Tagalog, and Fortran; symbols in 

abugida (alphasyllabic) scripts for Tamil and Sanskrit; and symbols in the cuneiform 

script for Sumerian. To compare sequences over dićerent alphabet sizes L, the 

logarithm in the entropy calculation was taken to base L: 417 for Indus, 4 for DNA, 

and so on. The resulting normalized block entropy is plotted as a function of block 

size. Error bars denote one standard deviation above/below mean entropy and are 

negligibly small except for block size 6. 

PROSPECTS FOR 

DECIPHERMENT 

Can the Indus script be deciphered 

without a bilingual artifact such as 

the Rosetta Stone? History suggests 

it could be: The Linear B script used 

in ancient Greece was deciphered in 

the 1950s without a bilingual artifact. 

The decipherment relied on several 

factors such as being able to identify 

common roots and suffixes, hypothesizing 

that the script was syllabic, and 

guessing the pronunciation of some 

symbols, which revealed the script 

to be a form of Greek. In the case of 

the Indus script, the short length of 

the available texts makes such an 

approach difficult. It may be possible, 

however, to obtain results by focusing 

on particular types of Indus texts and 

the contexts in which they are found. 

Most of the Indus texts found are 

on stamp seals, which were typically 

used in Bronze Age cultures for regulating 

trade. Seals were pressed onto 

clay tags affixed to packaged goods. 

The tags often listed the contents, 

APRIL 2010 

A 


aG 

F 

79

A 


aG 

F 

80 

origin or destination, type or amount 

of goods being traded, name and title 

of the owner, or some combination of 

these. Numerous such clay tags have 

been found at various sites in the Indus 

civilization, bearing seal impressions 

on one side and impressions of woven 

cloth, reed matting, or other packing 

material on the other. 

If the Indus script was used for 

trade, as the evidence suggests, then 

we would expect to find signs representing 

numerical quantities and 

units of measure. Progress in this 

direction has recently been reported 

by Bryan Wells, who estimated the 

volumes of two pots, one bearing 

the inscription and the other the 

inscription . By showing that the 

estimated volume of the second pot 

was in fact twice that of the first, 

Wells was able to conclude that 

strokes such as and represent 

numbers, and the sign probably 

represents a unit of volume. 

Other efforts by Parpola and 

Mahadevan have assumed that at 

least some of the texts probably 

represent names. Phonetic values 

for specific signs can then be sug- 

gested by assuming an underlying 

language—for example, proto-Dravidian—and 

using the rebus principle 

to guess the pronunciation of pictorial 

signs such as “fish,” “jar,” and 

“arrow.” In English, for example, 

the rebus principle could be used to 

represent an abstract word such as 

“belief” with the picture of a bee followed 

by a picture of a leaf. Ancient 

scripts often used the rebus principle 

to represent language. 

Probabilistic models could 

help in this decipherment process 

in several ways. Recently proposed 

algorithms for probabilistic 

grammar induction could allow 

construction of a partial grammar 

for the Indus texts, facilitating the 

identification of root words, suffixes, 

prefixes, and other modifiers. 

This may facilitate the use of deciphering 

techniques similar to those 

applied to Linear B. Reconstructing 

a grammar would also allow 

comparison with the grammars of 

other languages, helping narrow 

down the set of candidate language 

families to consider when using the 

rebus principle. 

Silver Bullet Security Podcast 

COMPUTER 

AI REDUX 

In-depth interviews with security gurus. Hosted by Gary McGraw. 

www.computer.org/security/podcasts 

Sponsored by 

A site-by-site analysis of the Indus 

texts using probabilistic models could 

indicate whether different languages 

or dialects were spoken in different 

regions of the Indus civilization. Similarly, 

training probabilistic models 

on texts found on specific types of 

artifacts, such as seals versus tablets, 

could ascertain whether the content 

of the texts varies according to artifact 

type. 

In summary, the study of the 

Indus script has emerged as an 

exciting area of interdisciplinary 

research, offering a unique opportunity 

for probabilistic models to shed 

new light on one of the world’s oldest 

civilizations. 

Rajesh P.N. Rao is an associate professor 

in the Department of Computer 

Science & Engineering at the University 

of Washington. Contact him at 

rao@cs.washington.edu. 

________________ 

Editor: Naren Ramakrishnan, Dept. of 

Computer Science, Virginia Tech, Blacksburg, 

VA; _________ 

naren@cs.vt.edu 

_____________________________________________ 

A 


aG 

F

A 


aG 

F 

SOFTWARE TECHNOLOGIES 

When TV Dies, Will 

It Go to the Cloud? 

Karin Breitman and Markus Endler, PUC-Rio, Brazil 

Rafael Pereira and Marcello Azambuja, Globo.com 

Coupled with the expected growth in bandwidth through the 

next decade, cloud computing will change the face of TV. 

The increasing popularity 

of videos on the 

Internet, allied to recent 

advances in network 

technology, are drastically changing 

television as we know it. In the past 

few decades we experienced a very 

clear role distinction: TV channels 

and independent companies produced 

video footage to be distributed 

by broadcasters and consumed by 

passive general audiences. Roles and 

responsibilities were clear down the 

line, and everyone was happy. 

A few years ago this “family 

around the TV set” scenario began to 

crack. Television faced several daunting 

challenges as cable multiplied the 

number of viewing choices and hardware 

prices let middle-class families 

own as many TV sets as there were 

members in the family. 

SEA CHANGE 

The Internet brought the potential 

to completely reinvent TV. First, 

it let users see what they wanted, 

when they wanted, while suppressing 

the need for additional hardware. 

Digital video recorders, notably the 

TiVo, popularized the concept of letting 

users choose more convenient 

times to watch their favorite programs. 

However, with the increase 

in bandwidth availability for the 

last mile—via cable and asymmetric 

digital subscriber line (ADSL) technol- 

0018-9162/10/$26.00 © 2010 IEEE 

ogy—it makes more sense to stream 

the content directly from the Internet 

than record it for later use. 

Second, and more importantly, the 

Net removes the barrier that separates 

producers, distributors, and consumers. 

On the Internet, anyone can 

produce and distribute high-quality 

content. As a result, there is much 

more content available, making the 

competition for audiences tougher and 

changing viewing habits irreversibly. 

Third, the Internet allows mixing 

and matching of multisource content. 

It has become commonplace for networks 

to mix their own footage with 

user-generated content to provide a 

more holistic experience. Video production, 

distribution, and selection are 

no longer the privilege of a few. Users 

now have a front seat, forcing media 

giants to explore new business models. 

From a technical viewpoint, huge 

challenges remain, however, including 

the ability to process, index, 

store, and distribute nearly limitless 

amounts of data. This is why cloud 

computing will play a major role in 

redefining TV in the next few years. 

RAMPING UP 

Electronic devices that can be used 

to consume content on the Internet 

are proliferating at an increasing rate: 

PCs, tablet and laptop computers, 

PDAs, mobile phones, handheld video 

game consoles, and eBook readers, 


not to mention router-type devices 

that let regular TV sets display Internet 

content. 

It seems our lives are driven by 

media displays, as Steve Jobs demonstrated 

with his recent unveiling 

of the iPad. It is easy to imagine a 

middle-class adult in the work force 

soon having access to two or more 

such displays on a regular basis. 

Displays can be organized in 

five strata, according to their evolution 

(B. Mazloff, Le 5éme écran 

– Les medias urbains dans la ville 

2.0 – editions Fyp - 2009), as Figure 

1 shows. Movie projection screens 

were the first to appear, in open 

public spaces. Then came the TV 

and PC—no longer public, but still 

collective. Mobile devices followed, 

for personal use in both private and 

public environments. More recently, 

public electronic displays on everything 

from billboards to bus stops are 

changing the urban landscape. 

Fifty years ago, video was produced 

in a single format since its 

consumption was restricted to some 

specific displays. Today, on the other 

hand, several different versions are 

required to enable access on PCs, 

iPhones, PDAs, game consoles, 

and other devices. This is a critical 

problem for broadcasters, as video 

processing is as computationally 

expensive as it is data intensive, consuming 

time and resources. 

APRIL 2010 

A 


aG 

F 

81

A 


aG 

F 

82 

SOFTWARE TECHNOLOGIES 

Shared 

public 

spaces 

open 

access 

Besides the proliferation of different 

devices, it is not uncommon for 

the market to prefer particular platforms. 

A decade ago, RealNetworks’s 

RealVideo and RealMedia took the 

lead, followed by MS Windows Media 

and Adobe’s Flash, with the future 

looking toward HTML 5. Each of these 

technologies defined a set of associated 

formats, codecs, and transport 

protocols that everyone must follow. 

On top of that, the last mile is pushing 

the demand for high-quality content 

and high-definition videos that require 

many more computational resources. 

SPLIT AND MERGE 

Split and Merge is a cloud-based 

platform for distributed video 

encoding that can overcome these 

drawbacks. The basic idea is to use the 

cloud’s elasticity to engage resources 

dynamically, then distribute and parallelize 

the video-encoding processes. 

The Split and Merge platform, shown 

in Figure 2, was designed to reduce 

video-encoding times to fixed thresholds, 

independently of the input size 

of the video file, using only dynamic 

resource provisioning in the cloud. 

Split and Merge fragments every 

video received, processes the 

fragments in a distributed cloud 

environment, and merges partial 

results. As in most map-reduce implementations, 

it makes efficient use of 

available computing resources. It also 

allows for the customization of tech- 

COMPUTER 

Shared 

private 

space 

open 

access 

Shared 

private 

space 

restricted 

access 

Individual 

mobile 

very 

restricted 

access 

Shared 

public 

spaces 

open 

access 

Figure 1. Evolution of media display types. The Internet is enabling the rapid 

proliferation of electronic displays. 

niques used in the split, distribute, 

process, and merge steps as needed. 

This ensures flexibility, adaptation, 

extensibility, and the accommodation 

of different applications. In the 

case of video processing, allowing 

a choice among codecs, containers, 

audio streams, and different splitting 

techniques is paramount. 

Considering the growth of Internet 

video use along with the demand 

for higher-quality HD-level content, 

dealing efficiently with transcoding 

processes is a strategic priority. The 

cloud solution offers a competitive 

advantage by providing the unprecedented 

possibility of processing 

videos several hours long in the same 

time frame as those that run only a 

few minutes. 

DISTRIBUTION CHALLENGES 

Unlike TV signals, which are transmitted 

independently of the number 

of viewers, Internet media consumption 

is based on connectivity: Each 

user who wants to watch a video 

must open a separate connection. 

With unicast, the most popular distribution 

model, there is a separate 

connection for each user to the server 

responsible for content distribution. 

This limitation becomes increasingly 

evident when a variable volume of 

information must be processed to 

meet spikes in demand during, for 

example, popular sports competitions 

or public emergencies. 

No matter how resourceful the 

entity distributing the contents is, 

there will be peak times with excessive 

demand and other times of 

idleness. Given that service providers 

can build applications with an elastic 

infrastructure and adjust effectively 

to demand variations, the cloud has 

all the characteristics needed to deal 

with this type of content. 

In cases where there is a very 

large or seasonal demand, the use 

of public clouds for information processing 

and storage is emerging as an 

attractive alternative. The concept of 

hardware as a service (HaaS) relieves 

the necessity of making large infrastructure 

investments, while allowing 

on-the-fly resizing and adaptation to 

current needs. 

With a public cloud, users can 

quickly gauge the resources required 

to perform a particular task and pay 

only for those effectively used. A 

successful example of this is the registration 

system for the Big Brother 

Brasil reality TV show. 

The application process is open 

to any resident in the country, with 

contestants required to send a video 

of themselves in the format of their 

choice. Videos are encoded in the 

MPEG-2 standard to facilitate visualization 

by the jurors and spare them 

the hassle of dealing with a plethora 

of different codecs. The system is 

able to receive a very large number of 

videos during the three-month application 

process. 

The system in particular leverages 

the power of cloud computing to deal 

with uncertain storage and processing 

requirements, allocate resources 

needed during the application and 

selection processes, and scale up to 

rare but extreme high-peak situations—for 

example, during the last 

weekend, when 60 percent of the 

total submissions with about 200,000 

videos are expected. 

Applications with seasonal but 

very large demand are not rare. 

The Internet transmission of sports 

events such as FIFA’s World Cup and 

A 


aG 

F

A 


aG 

F 

the Olympics, and of breaking news 

such as Michael Jackson’s death, 

require a huge infrastructure for a 

very short time, making them killer 

cloud applications. 

CONSUMER CHALLENGES 

The new paradigm of video content 

consumption quickly became 

clear to major broadcasters, whose 

revenues had been falling consistently 

since 2005 and who had 

been forced to rethink their content 

distribution models. For example, 

NBC Universal partnered with News 

Corp. to launch Hulu.com in 2007, 

with Disney joining two years later. 

The popular website, which offers 

commercial-supported streaming 

video of TV shows and movies, aims 

to recover part of the lost revenue 

resulting from the decrease of traditional 

TV audiences. 

During the recent high-definition 

format war between HD-DVD (supported 

by Microsoft and Toshiba) and 

Blu-ray (supported by Sony) technologies, 

videogames played an important 

role. Sony pushed Blu-ray adoption, 

using the technology as the underlying 

basis for its PlayStation 3, which 

doubles as a Blu-ray disc player. 

Although Microsoft’s Xbox 360 console 

supported HD-DVD, a disc player 

sold as expansion hardware and was 

not required to play videogames. 

Shortly after Blu-ray emerged victorious 

in 2008 (http://news.bbc.co.uk/1/ 

________________ 

hi/business/7252172.stm), it become 

clear that Microsoft’s strategy was not 

focused on the disc formats but rather 

on the rising market for online video. 

In July 2008, less than six months 

after the end of the HD format wars, 

Microsoft announced its partnership 

with Netflix to stream movies and 

TV episodes using the XBox 360 console 

(www.microsoft.com/presspass/ 

________________________ 

press/2008/jul08/07-14instantstreampr. 

____ mspx). This constituted a shift from 

the paradigm of media discs, enabling 

console owners to rent movies with the 

press of a button. Apple offered a similar 

initiative with Apple TV, enabling 

chunk1 chunk2 

subscribers to download movies 

through its iTunes store. 

While the Internet offered new revenue 

possibilities, it also augmented 

the complexity of production and 

distribution business models. Hulu 

(www.fastcompany.com/magazine/ 

_____________________ 

140/the-unlikely-mogul.html), for 

example, preferred to follow a model 

similar to that used by TV, with 

commercial spots at intervals in the 

content. Even this simple model can 

be further refined, with Web technology 

easily identifying and addressing 

specific group campaigns such as 

gender, age, and location. 

Cable TV networks are pursuing 

a similar approach to avoid losing 

ground to Internet-based TV providers. 

Many have launched their own 

video sites, which offer added services 

to justify their value. For example, 

Comcast recently launched Xfinity 

(blogs.wsj.com/digits/2009/12/15/ 

______________________ 

comcast-opens-fancast-xfinity-tv), a 

service that lets users watch TV programming 

on the Internet anywhere. 

The popularity of online videos 

has also raised the interesting possibility 

of the Internet serving as a large 

and collaborative content producer, 

with user-generated content complementing 

TV programming. 

Tempo 

chunkN 

chunk1 chunk2 chunkN 

Full encoded video 

Figure 2. Split and Merge architecture. The system fragments every video received, 

processes the fragments in a distributed cloud environment, and merges partial 

results. 

Video production and distribution 

are no longer 

restricted to large broadcasters. 

Users now have the power 

to make, distribute, mix, and match 

their own content. Cloud computing 

will play a decisive role in this 

extremely demanding scenario. The 

evolution of connectivity, coupled 

with the expected growth in bandwidth 

over the next decade, will 

indelibly change the face of TV. 

Karin Breitman is an assistant 

professor in the Departamento de 

Informática, PUC-Rio. Contact her at 

______________ 

karin@inf.puc-rio.br. 

Markus Endler is an associate 

professor in the Departamento de 

Informática, PUC-Rio. Contact him at 

______________ 

endler@inf.puc-rio.br. 

Rafael Pereira is a senior media engineer 

at Globo.com. Contact him at 

____________________ 

rafael.pereira@corp.globo.com. 

Marcello Azambuja is a digital media 

manager at Globo.com. Contact him at 

azambuja@ieee.org. 

_____________ 

Editor: Mike Hinchey, Lero—The Irish 

Software Engineering Research Centre; 

mike.hinchey@lero.ie 

____________ 

APRIL 2010 

A 


aG 

F 

83

A 


aG 

F 

84 

INVISIBLE COMPUTING 

Questioning Invisibility 

The invisibility of today’s 

technology is often a 

blessing. My laptop 

can be used as a word 

processor, TV, and stereo without 

COMPUTER 

Leah Buechley, MIT Media Lab 

Should invisibility be the guiding design goal for ubiquitous 

computing? 

A good tool is an invisible tool. By invisible, I mean that the tool does not intrude on your consciousness; you focus on the task, 

not the tool. 

– Mark Weiser 

changing shape, and I’m grateful 

that I don’t have to think about transistors, 

compilers, and operating 

systems while I use it. However, is 

invisibility always a good thing? 

Figure 1. Crashes and errors make the invisible technologies underlying our lives 

alarmingly visible. 

Should it be the guiding design goal 

for ubiquitous computing? 

WHEN INVISIBILITY FAILS 

Benjamin Mako Hill has been 

examining instances where the invisibility 

of technology breaks down. 

His website, revealingerrors.com, 

documents occasions when a crash 

or error makes one of the invisible 

technologies underlying our lives 

alarmingly visible, forcing us to confront 

its technological innards. Figure 

1 shows a few examples. 

The crashed ATM with its exposed 

Windows desktop and the GPS system 

that locates itself in the ocean are 

self-explanatory errors, but the third 

example requires some elucidation. 

The text at the top of the webpage 

should read “Gay eases into 100 final 

at Olympic trials,” Gay being the 

surname of athlete Tyson Gay. This 

strange headline reveals the fact that 

the website posting the news feed, 

a conservative network called One 

News Now, uses a program to adjust 

the language of AP news stories. Here 

we see that part of their program 

replaces instances of the word “gay” 

with the word “homosexual.” 

Hill argues that such breakdowns 


A 


aG 

F

A 


aG 

F 

Figure 2. The LilyPad Arduino toolkit consists of a set of sewable electronic modules that users can stitch together with electrically 

conductive thread to make interactive textile-based computers. 

aren’t necessarily bad. Though 

inconvenient and frustrating, they 

can reveal the inelegance, fragility, 

and questionable practices behind 

technological systems. I can use that 

information to make more informed 

decisions; I might very well change 

banks if I knew more about my favorite 

ATM’s operating system. 

Does this mean we should start 

designing error-prone systems? Of 

course not. But it does raise a question 

about the narrow design goal of 

invisibility. Systems that vanish into 

the background aren’t inherently 

good ones. The examples in Figure 1 

are troubling demonstrations of some 

of the things invisibility can hide. 

LEVERAGING VISIBILITY 

FOR EDUCATION 

The errors documented on Hill’s 

website are funny, fascinating, and 

enlightening; consequently, they 

provide wonderful opportunities for 

teaching and learning. 

Imagine the curiosity and delight 

that an encounter with one of these 

broken systems could inspire in a 

young person. There are two facets 

to this appeal. First, the errors are 

engaging—they surprise us and force 

us to notice technology we might 

have otherwise ignored. Second, they 

introduce legibility to technology— 

they reveal interesting information 

about how it works. 

In the research group I direct at 

the MIT Media Lab (hlt.media.mit. 

__________ 

___ edu), we design educational technology 

that is deliberately engaging and 

legible. We exploit these properties 

to get students excited about technology. 

Our particular approach focuses 

on using new and unusual physical 

materials to create compelling and 

comprehensible systems. 

For example, during the past several 

years we have been working 

in the emerging field of electronic 

textiles, or “e-textiles,” integrating 

electronics, computers, and textiles. 

We began by tackling a fundamental 

engineering challenge: How do you 

attach computers to fabric? Once we 

had developed robust solutions to this 

problem, we focused on making the 

domain accessible to students. 

The result of our efforts was the 

LilyPad Arduino (L. Buechley et al., 

“The LilyPad Arduino: Using Computational 

Textiles to Investigate 

Engagement, Aesthetics, and Diversity 

in Computer Science Education,” Proc. 

26th Ann. SIGCHI Conf. Human Factors 

in Computing Systems, ACM Press, 

2008, pp. 423-432). As the leftmost 

image in Figure 2 shows, this toolkit 

consists of a set of sewable electronic 

modules—similar in spirit to Lego 

Mindstorm modules—that users can 

stitch together with electrically conductive 

thread to make interactive 

textile-based computers. 

In a series of courses and workshops, 

we found that this very visible 

new technology could uniquely capture 

the imagination of teenagers 

and teach them the fundamentals of 

computer science and electrical engineering. 

Interestingly, young women 

were especially excited about this 

blend of textiles and computing. The 

other two images in Figure 2 show a 

young woman from one of our workshops 

first constructing and then 

modeling her course project, an interactive 

light-up jacket with an ambient 

thermometer. 

We’re now expanding this 

approach to another unusual material, 

paper—exploring how we can 

blend paper and paint with computers 

and electronics. We’ve designed a 

paper computing kit that consists of 

magnetic sensors, actuators, microcontrollers, 

and wireless devices; 

ferrous paper, to which the modules 

stick; and a jar of conductive paint, 

which is used to connect the modules 

across the paper. Students can 

use the kit to paint beautiful, functional 

electronic devices onto paper. 

Figure 3 shows images of a prototype 

kit and a construction example: 

an electronic pop-up book built by 

Jie Qi, an undergraduate researcher 

in our lab. 

We’ve just begun to hold workshops 

to explore how we might employ this 

kit to teach embedded computing (J. Qi 

APRIL 2010 

A 


aG 

F 

85

A 


aG 

F 

86 

INVISIBLE COMPUTING 

Figure 3. Paper computing kit. Left: Prototype kit. Right: Electronic popables. 

and L. Buechley, “Electronic Popables: 

Exploring Paper-Based Computing 

through an Interactive Pop-up Book,” 

Proc. 4th Int’l Conf. Tangible, Embedded, 

and Embodied Interaction, ACM 

Press, pp. 121-128). 

Technologies like these—ones that, 

at their best, captivate, empower, 

and educate people—stand in direct 

opposition to ubicomp’s tenet of 

invisibility; the kits, and often the 

devices constructed with them as 

well, are intentionally visible objects 

that reveal the inner workings of technology 

and explicitly require users’ 

time and attention. 

Other researchers have also noted 

this conflict between invisibility and 

education (M. Eisenberg et al., “Invisibility 

Considered Harmful: Revisiting 

Traditional Principles of Ubiquitous 

Computing in the Context of Education,” 

Proc. 4th IEEE Int’l Workshop 

Wireless, Mobile, and Ubiquitous Technology 

in Education, IEEE CS Press, pp. 

103-110; Y. Rogers, “Moving On from 

Weiser’s Vision of Calm Computing: 

Engaging UbiComp Experiences,” 

Proc. 8th Int’l Conf. Ubiquitous Computing, 

LNCS 4206, Springer, 2006, pp. 

404-421). 

INVISIBILITY, VISIBILITY, 

AND DESIGN 

In our lives there are objects we 

ignore and objects we treasure. Some 

vanish into the background and stay 

there and others please, amuse, 

COMPUTER 

and comfort us. There are the forgotten 

paper clips, printers, and 

towels, and the beloved dining room 

tables, sports uniforms, and jewelry. 

Why should ubicomp—or any 

other computing discipline, for that 

matter—consign itself to the ignored, 

invisible realm? 

Our research group strives to build 

systems that fit into people’s aesthetic 

and emotional lives as much as their 

technological ones. As Figures 2 and 3 

suggest, exploring the aesthetic possibilities 

of technology provides us with 

new and unusual ways to excite and 

engage students. We also investigate 

aesthetic and emotional dimensions 

when we develop systems that are 

less explicitly educational. 

A recent project along these lines 

is a unique piece of wallpaper that 

we constructed with our paper computing 

kit. The Living Wall is a flat 

surface built out of ferrous paint, 

conductive paint, traditional paint, 

and electronics to which our paper 

computing modules can be attached. 

With its wide range of attachable 

components, the Living Wall can provide 

lighting, sense information about 

its environment, communicate with 

other devices like computers and 

phones, and act as a large-scale input 

device. The News Brief “Researchers 

Build a Wall That Acts Like a Remote 

Control” on p. 18 has more information 

on the system and an illustration 

of its use. 

The Living Wall is a flexible ubicomp 

system. However, it also 

functions as a decorative element 

in a home. Like the other examples 

mentioned here, it wasn’t designed 

to be invisible; it elicits and rewards 

attention. 

Invisibility is a narrow design 

goal. It’s not necessarily a bad 

one, but it doesn’t capture the 

full range of technological or creative 

possibilities. If we as computer scientists 

and engineers only strive to 

build invisible systems, we’ll neglect 

to build important technology that is 

educational, engaging, and beautiful. 

We should expand our focus and our 

rhetoric. 

Leah Buechley is an assistant professor 

and AT&T Career Development 

Professor of Media Arts and Sciences 

at MIT Media Lab, where she also 

directs the High-Low Tech research 

group. Contact her at leah@media. 

mit.edu. 

Editor: Albrecht Schmidt, Institute for 

Computer Science and Business Information 

Systems, University of Duisburg-Essen, 

Germany; albrecht.schmidt@gmail.com 

Selected CS articles and columns 

are available for free at 


Join the 

IEEE 

Computer 

Society 

www.computer.org 

A 


aG 

F

A 


aG 

F 

EDUCATION 

Cutting Across 

the Disciplines 

Jim Vallino, 

Rochester Institute of Technology 

Engineering and computing educators must design curricula 

that require students to work outside their own domain. 

Multidisciplinary, 

interdisciplinary, 

transdisciplinary— 

these terms are 

all the buzz in engineering and 

computing curriculum design. 

Throughout their career, most 

graduates will work on project 

teams with people from multiple 

fields of expertise. Engineering and 

computing educators must accordingly 

design curricula that require 

students to work outside of their 

own domain. This opportunity can 

range from small-group collaboration 

in a single course to large-scale 

efforts that span every program at 

an institution. 

There is often resistance to incorporating 

multidisciplinary work into 

curricula. Academic institutions, 

like most other organizations, have 

administrative and management 

entities that can unintentionally construct 

impediments to cutting across 

entrenched boundaries. At colleges 

and universities, fingers usually point 

at the department as the main culprit. 

Typical impediments are joint sched- 

0018-9162/10/$26.00 © 2010 IEEE 

uling of cross-listed courses, faculty 

teaching-load accounting, and support 

for shared facilities. 

One reaction to this is a push to 

eliminate administrative entities. 

However, this creates a new set of 

problems when program responsibility 

becomes ill-defined. At the core, 

what is required is a commitment 

to improve students’ education by 

promoting multidisciplinary teams 

and projects. The motivation may 

be provided top-down, especially 

for institute-wide programs, but 

in most cases it will be a limited 

effort led from the bottom up by a 

small number of individual faculty 

members. 

ACCREDITATION 

PROVIDES A MOTIVATION 

Whether you believe that multidisciplinary 

work will be the savior 

of engineering and computing education 

or is just another fad that 

will blend into the curriculum landscape, 

educators will see it in one 

form or another in the foreseeable 

future. 


Programs accredited by the Engineering 

Accreditation Commission of 

ABET (www.abet.org) must demonstrate 

that students have 

“(c) an ability to design a system, 

component, or process to meet 

desired needs within realistic 

constraints such as economic, 

environmental, social, political, 

ethical, health and safety, manufacturability, 

and sustainability”; 

“(d) an ability to function on 

multidisciplinary teams”; and 

“(h) the broad education necessary 

to understand the impact of 

engineering solutions in a global, 

economic, environmental, and 

societal context.” 

All three of these program outcomes 

look beyond a traditional 

single-discipline curriculum. Because 

the program is responsible for demonstrating 

these outcomes, the 

experiences can’t be haphazardly 

inserted in the curriculum or based 

on elective coursework that students 

may or may not actually take. 

APRIL 2010 

A 


aG 

F 

87

A 


aG 

F 

88 

COMPUTER 

EDUCATION 

The Computing Accreditation 

Commission specifies similar criteria 

for the accreditation of computing 

programs. 

The broad field of computing cuts 

across disciplines in unique ways. 

Computing typically has a domain 

presence as a necessary tool rather 

than an area of detailed study— 

students must learn the fundamentals 

of computing to successfully 

apply it in all their work. Perhaps 

more important, they also must 

learn when a project has a computing 

component that requires deeper 

expertise. This becomes an opportunity 

for educators to blend programs 

in computing, engineering, and other 

disciplines across an institution. 

MULTIDISCIPLINARY 

IN THE SMALL 

There are conflicting definitions 

for the terms multidisciplinary, interdisciplinary, 

and transdisciplinary, 

but educators shouldn’t get stuck on 

what to call it. Any approach that 

gets students interacting with material 

and people outside their own 

technical discipline, or in ways that 

are atypical for their discipline, is 

beneficial. 

Those who want to experiment 

with a multidisciplinary approach 

should start small within one of their 

courses by assigning exercises to 

students that require them to think 

from unfamiliar perspectives. In one 

course, I ask students to capture their 

thoughts on aspects of the course 

material in Haiku poems. Here’s an 

example of one called “Basic Principles” 

from a software engineering 

design course: 

Coupling, Cohesion 

Antagonistic forces 

Tradeoffs to be made. 

In that same course, I have students 

ask themselves about the 

structure shown in M.C. Escher’s 

Waterfall (www.mcescher.com/ 

_____________________ 

Gallery/recogn-bmp/LW439.jpg). For 

each of those questions about this 

illusionary structure, I ask them 

to pose an analog question when 

studying the design of a software 

system. 

This exercise isn’t what most 

readers may have in mind when 

they think multidisciplinary, but on 

a small scale that a faculty member 

can do in any course, it requires students 

to think using paradigms from 

other disciplines. 

The next step would be to have students 

enrolled in courses in different 

The students as well as the instructor must commit 

to the multidisciplinary approach. 

disciplines collaborate on projects. 

Individual faculty can coordinate this 

with little administrative support. As 

in all such instances, the students as 

well as the instructor must commit 

to the multidisciplinary approach: A 

project that serves as an extracurricular 

or small-credit activity for some 

students but counts as significant 

coursework for others won’t be very 

effective. 


BY DESIGN 

If educators intend the multidisciplinary 

experience to satisfy 

accreditation requirements, it’s preferable 

to carefully design it into a 

program. 

All ABET-accredited software engineering 

programs require students to 

be able to work in one or more major 

application domains. However, what 

constitutes an application domain 

varies widely across programs (D. 

Bagert, “The Role of Application 

Domain Tracks in Software Engineering 

Programs,” Proc. 2006 Ann. 

Conf. Am. Soc. for Eng. Education, 

ACEE, 2006; http://soa.asee.org/paper/ 

_______________________ 

conference/paper-view.cfm?id=2832). 

Our software engineering program 

at the Rochester Institute of 

Technology (RIT) uses application 

domains as a vehicle to expose students 

to uses of computing outside 

their comfort zones. Each domain 

requires three courses, at least two 

of which must be taken from colleges 

other than our College of Computing 

and Information Sciences. This 

prevents the application domain 

from being simply a set of computing 

electives. 

A cross-curriculum multidisciplinary 

approach involves designing 

a course, or course cluster, with 

multi-disciplinary interaction as a 

basic course principle. Educators 

must make the course material 

interesting and accessible to students 

from all the disciplines expected to 

participate. 

This can be challenging. With a 

computer engineering colleague, I 

created a cluster of three courses 

in real-time and embedded systems 

at RIT (J. Vallino and R.S. 

Czerniskowski, “Interdisciplinary 

Teaming as an Effective Method to 

Teach Real-Time and Embedded Systems 

Courses,” Proc. 2008 Workshop 

Embedded Systems Education, 2008, 

pp. 25-32). Even with closely related 

disciplines like software engineering 

and computer engineering, it 

was more difficult than we originally 

thought to keep the course interest 

balanced between the two. 

Educators must carefully select 

course projects to ensure that all 

disciplines contribute in different 

but equivalent ways. Some registration 

controls may be needed to 

help maintain balance across all 

disciplines. Ideally, these courses 

are team taught, but unless an 

institution’s teaching-load accounting 

easily accommodates that 

arrangement, it may have to settle 

for more traditional single-instructor 

teaching. 

A 


aG 

F

A 


aG 

F 

Finally, what seems like it should 

be easy could require regular attention. 

Each term, we have to schedule 

multiple course listings at the same 

time, in the same room, and with a 

common final exam time. 

We’re currently applying lessons 

learned from the successful multidisciplinary 

real-time and embedded 

systems cluster to an applied cryptography 

cluster incorporating computer 

science, computer engineering, and 

software engineering. 

LARGEfiSCALE 


EXPERIENCES 

More ambitious forays into multidisciplinary 

work require significantly 

more coordination and administrative 

support. RIT has implemented many 

such programs for seniors in its College 

of Engineering (P.H. Stiebitz, E.C. 

Hensel, and J.R. Mozrall, “Multidisciplinary 

Engineering Senior Design 

at RIT,” Proc. 2004 Ann. Conf. Am. 

Soc. for Eng. Education, ACEE, 2004; 

http://soa.asee.org/paper/conference/ 

_________________ 

paper-view.cfm?id=20471). 

Extended multidisciplinary 

efforts start with designing appropriate 

content and uniform assessment 

mechanisms for courses that encompass 

a wide range of projects. It’s 

usually necessary to designate a 

project coordinator to solicit sufficient 

cross-disciplinary content 

from internal and external sponsors. 

Faculty coaches or advisors must be 

assigned to every project along with 

a student team. To achieve faculty 

buy-in, it may be necessary to give 

teaching credit to those working 

with project teams. 

Accreditation requirements initially 

motivate many college-level 

multidisciplinary programs. Once 

the programs have overcome startup 

problems and are steadily operating, 

however, the benefits beyond accreditation 

become evident. 

Ultimately, educators may want to 

embrace a multidisciplinary culture 

that spans the entire institution—for 

example, including business, fine 

arts, and liberal arts as well as engineering 

and computing. This might 

be tied to initiatives requiring undergraduate 

students to have research or 

international study experience. 

Cutting across disciplines with 

diverse curriculum models adds new 

challenges to making multidisciplinary 

work a reality. There should 

be a way to define projects from any 

corner of the campus. Each project 

needs at least one faculty member as 

an advocate to specify the necessary 

team skills. 

Every program must offer students 

an opportunity for required or elective 

multidisciplinary project work. 

Students could search a project registration 

database for interesting 

projects, or the system could suggest 

projects based on students’ selfdefined 

skill sets. 

Some material, such as discussion 

of problem-solving techniques 

and how to be a good team player, 

should be common to all projects or 

integrated into prerequisite courses, 

perhaps hosted by multiple entities 

using different paradigms. Students 

could be required to, for example, 

take a problem-solving course appropriate 

for their home discipline and 

one from another area. 

To satisfy accreditation requirements, 

a faculty member in a 

student’s home discipline would 

need to review the student’s work 

to ensure that he or she has gained 

discipline-appropriate experience on 

the project. 

The pressure on engineering 

and computing curricula 

is to broaden programs 

(National Academy of Engineering, 

The Engineer of 2020: Vision of Engineering 

in the New Century, National 

Academies Press, 2004). Many educators 

view the current curriculum 

model for most programs as being 

too focused. I’ve been in meetings 

where a view was expressed that 

highly structured engineering programs 

are a primary impediment 

to creative and innovative thought. 

I certainly don’t subscribe to that 

view, but given the direction that 

engineering and computing education 

is headed, I have some concerns. 

Teams work best when they bring 

together people with individual depth 

in multiple disciplines who are open 

to cutting across the disciplines. 

Will a collection of broadly trained 

students be as effective as one of students 

who have depth in their own 

individual disciplines? 

One benefit of multidisciplinary 

projects is the presence of “low-hanging 

fruit” between the disciplines. 

As institutions broaden programs 

and students attain breadth across 

disciplines, will educators abandon 

individual disciplines themselves 

in pursuit of easy-to-pick fruit? Will 

students be able to tackle the hard 

problems in the individual disciplines 

or even in the multidisciplinary 

domain? 

When redesigning curricula to 

incorporate multidisciplinary activities, 

educators must be careful. To 

students’ potential employers, it could 

be a short curricular distance from 

“Your students need more breadth” 

to “Your students don’t know very 

much.” 

Jim Vallino is a professor in the 

Department of Software Engineering 

in the B. Thomas Golisano College of 

Computing and Information Sciences 

at Rochester Institute of Technology. 

Contact him at j.vallino@se.rit.edu. 

_____________ 

Editor: Ann E.K. Sobel, Department of 

Computer Science and Software Engineering, 

Miami University; sobelae@muohio.edu 

____________ 


are available for free at ____ 

http:// 

ComputingNow.computer.org. 

APRIL 2010 

A 


aG 

F 

89

A 


aG 

F 

90 

COMPUTER 

SECURITY 

Integrating Legal 

and Policy Factors in 

Cyberpreparedness 

James Bret Michael, Naval Postgraduate School 

John F. Sarkesain, Aerospace Corp. 

Thomas C. Wingfield, George C. Marshall European 

Center for Security Studies 

Georgios Dementis, Hellenic Navy 

Gonçalo Nuno Baptista de Sousa, Portuguese Navy 

Cyberwarfare countermeasures must consider more than 

technological capabilities. 

Attacks in cyberspace 

are commonplace. The 

effects of such attacks 

can range from minor 

nuisances, such as defacing webpages 

or temporarily denying service 

to noncritical systems, to major disturbances 

that interrupt international 

commerce or threaten to destabilize 

a nation-state. 

Anyone can wage an attack in 

cyberspace: individual citizens, 

criminal syndicates, terrorist organizations, 

even entire nations. Such 

attacks can be extremely sophisticated 

and involve many actors. The 

cyberattacks on Estonia in 2007 by 

so-called “patriotic hackers,” criminal 

elements that leased out botnets, 

and alleged state-sponsored information 

warriors combined some 

of the characteristics of a military 

campaign with those of a covert 

operation (www.economist.com/ 

________________________ 

world/international/displaystory. 

cfm?story_id=E1_JNNRSVS). 

__________________ 

CYBERPREPAREDNESS 

Regardless of who perpetrates 

a cyberattack, defenders of the 

attacked systems must be prepared 

to respond, even if only to mitigate 

the attack’s effects. Cyberpreparedness 

can be said to have three 

dimensions (E. Tikk and T. Wingfield, 

“Frameworks for International Cyber 

Security: The Cube, the Pyramid, and 

the Screen,” presentation, Int’l Cyber 

Conflict Legal and Policy Conf., 2009): 

technical feasibility—the 

“possible”; 

legal—the “permissible”; and 

policy—the “preferable.” 

From a technical vantage, a 

defender could use a computerbased 

tool such as NetSPA to assess 

a computer network’s vulnerability 

to attack and develop appropriate 

countermeasures (K. Ingols et al., 

“Modeling Modern Network Attacks 

and Countermeasures Using Attack 

Graphs,” Proc. Ann. Comp. Security 

Applications Conf., IEEE, 2009, pp. 117- 

126). However, a defender also needs 

a distributed command, control, and 

battle management (C2/BM) system 

to maintain situational awareness of 

and respond to cyberattacks in nearreal 

time (N. Howes, M. Mezzino, 

and J. Sarkesain, “On Cyber Warfare 

Command and Control Systems,” 

Proc. 9th Ann. Int’l Command and Control 

Research and Technology Symp., 

2004; www.dodccrp.org/events/9th_ 

____________________ 

ICCRTS/CD/papers/118.pdf). 

_________________ 

LAW AND POLICY 

Turning now to the “permissible” 

and “preferable,” the customary 

guiding principles of jus in bello, 

“customary legal standards for the 

conduct of war”—discrimination, 

necessity, proportionality, and chivalry—also 

apply to cyberwarfare, 

as does the jus ad bellum, “law governing 

the transition from peace to 

war” (J.B. Michael, “On the Response 

Policy of Software Decoys: Conducting 

Software-Based Deception in 

the Cyber Battlespace,” Proc. 26th 

Ann. Int’l Computer Software and 

Apps. Conf., IEEE, 2002, pp. 957-962). 

Cyberattacks can have the equivalent 

effects of attacks waged with kinetic 

weapons, rising to the level of a “use 

of force” under international law (J.B. 

Michael, T. Wingfield, and D. Wijesekera, 

“Measured Responses to Cyber 

Attacks Using Schmitt Analysis: A 

Case Study of Attack Scenarios for 

a Software-Intensive System,” Proc. 


A 


aG 

F

A 


aG 

F 

26th Ann. Int’l Computer Software 

and Applications Conf., IEEE, 2003, 

pp. 622-627). 

Thus, regardless of the technical 

measures used, law and policy must 

be in place to guide defenders as 

they craft responses to cyberattacks. 

Although a response might be legal, 

policy dictates the latitude in applying 

the law, taking into consideration 

such factors as the impact on society, 

diplomatic relations, and even 

privacy concerns, such as a given 

response’s level of intrusiveness. The 

currency of the topic of cyberpreparedness 

is evidenced by the recent 

front-page news account of the legal 

and policy discussions that ensued 

prior to the cyberattack sponsored 

by the US Department of Defense to 

dismantle an online forum used by 

terrorists to plan kinetic attacks on 

US troops (E. Nakashima, “For Cyberwarriors 

Murky Terrain; Dismantling 

of Saudi-CIA Web Site Illustrates need 

for Clearer Cyberwar Policies,” The 

Washington Post, 19 March 2010, pp. 

A1 and A16). Policy issues included 

the impact on US-Saudi relations and 

whether the response should be conducted 

as an intelligence or military 

operation. 

In addition, techniques for conducting 

principled analyses must be 

ready to help minimize legal uncertainty 

about cyberincidents and allow 

the most complete range of effective 

responses. Whether the response is 

conducted via kinetic or cyber means, 

it must be lawful (T. Wingfield, The 

Law of Information Confiict: National 

Security Law in Cyberspace, Aegis 

Research Corp., 2000). 

LEVERAGING THE 

LEGAL CELL 

We propose merging the three 

dimensions of cyberpreparedness 

by adding a new type of virtual cell 

to cybersystems: the legal cell. 

Virtual cells within such systems 

serve as dynamic virtual 

communities in which cyberopera- 

tors can interact in support of making 

informed decisions. Cyberoperators 

encompass subject-matter experts 

in cyberoperations along with the 

related communities of the diplomatic 

corps, intelligence analysts, the 

military, politicians, legal authorities, 

law-enforcement agents, economists, 

and those responsible for creating 

and maintaining the national critical 

infrastructure. Cyberoperators can 

create, join, leave, and even dissolve 

cell instantiations. 

The legal cell provides cyberoperators 

with the means to obtain 

advice from experts in information 

operations (IO) law. Within cyber C2/ 

BM systems, cells subscribe to infor- 

We propose merging 

the three dimensions 

of cyberpreparedness 

by adding a new type 

of virtual cell to cybersystems: 

the legal cell. 

mation and the system publishes it 

continuously in a digestible format. 

This differs from the information-pull 

approach incorporated into most of 

today’s C2/BM systems when used in 

conjunction with kinetic warfare. 

Through the legal cell, IO experts 

can maintain a current picture of the 

cyberspace situation and provide 

advice as needed. This is an important 

capability, given the rapid tempo 

of cyberbattles and the high degree 

of space compression, which asserts 

that the physical distance from a 

target does not matter as it would 

with kinetic weapons and that a 

flattened command structure is necessary 

for conducting cyberwarfare. 

On instantiation of a legal cell, the 

system populates it with presumptions 

and algorithms. In this context, 

presumptions follow black-and-white 

if-then rules automatically enacted by 

mobile agents, whereas algorithms 

involve more complex data processing 

than with production rules but, 

like presumptions, are processed 

without human intervention and 

whose results trigger a course of 

action. 

These three elements—the law, 

presumptions, and algorithms—are 

what Tikk and Wingfield refer to as 

the “pyramid,” representing decision 

making along a continuum of 

operational speed and the inherent 

limitations on what types of problems 

computers solve in support of decision 

making. 

ATTACK RESOLUTION 

Given the unpredictability and 

quickness of cyberattacks, a rapid 

response is critical, necessitating 

use of at least some autonomic 

systems. According to a report by 

the National Institute of Standards 

and Technology (NIST) outlining a 

smart grid cybersecurity strategy 

and requirements (http://csrc.nist. 

gov/publications/drafts/nistir-7628/ 

draft-nistir-7628_2nd-public-draft. 

__ 

pdf). 

Ultimately, the cyberoperation 

leader will ascertain the effect of 

the attack’s automated response, 

particularly in the military because 

a commander cannot delegate 

responsibility for the actions of their 

command. While a commander may 

delegate the authority to execute 

certain actions on to a subordinate 

within the command, the commander 

still retains full and complete responsibility 

for the outcome of those 

delegated actions, for good or bad. 

Some defensive systems that involve 

using kinetic weapons require extensive 

automation, such as those used 

for ballistic missile defense, in which 

the duty cycle for core functions of 

the so-called “kill chain”—detect, 

track, assign weapons, engage, and 

assess outcome—cannot be performed 

by humans swiftly enough to 

achieve the desired levels of operational 

effectiveness. 

APRIL 2010 

A 


aG 

F 

91

A 


aG 

F 

92 

COMPUTER 

SECURITY 

An example of a presumption in 

this context is: “If a tracked object 

enters a restricted airspace, then 

assign a weapon to that object and 

engage.” An instance of engaging 

the human operator would occur if, 

for example, the US were to have a 

chance to intercept missiles over 

another country. In this case, intercepting 

the missiles would be lawful 

but the potential consequences— 

such as residual debris from the 

missiles’ chemical, nuclear, or biological 

payloads landing on the other 

country’s territory—would need to be 

weighed against the risk of the missiles 

reaching their targets. 

Other aspects of preparedness 

could be incorporated into cyber 

C2/BM systems, such as a brokering 

capability to facilitate controlled dissemination 

of information among 

cyberoperators from different 

administrative domains who are collaborating 

in an ad hoc but formal 

coalition, or persistent operations to 

defend assets through their combined 

cyberwarfare resources. 

For example, a legal advisor might 

need to subscribe to certain data 

feeds to pull together the information 

required to make a legal determination, 

but he or she might not have a 

need to know about where the data 

originated or related information. 

That is, the brokering capability provides 

a means for enabling the sharing 

of information by protecting the anonymity 

of the information sources 

and their methods of collection, only 

sharing what cyberoperators playing 

particular roles are entitled to access. 

Such an information-sharing capability 

would have been useful to the 

informal assembly of private- and 

public-sector organizations that came 

to the aid of Estonia in 2007. 

Systems have become increasingly 

interconnected through trends such 

as the adoption of cloud computing 

and the creation of systems of systems. 

These compositions provide 

one or more value-added capabilities 

that none of the individual systems 

can deliver on their own. Preparedness 

for cyberattacks will entail 

cooperation between the many stakeholders 

in these systems, even with 

sound mechanisms for containment 

of behavior between systems. 

Efforts at achieving cyberpreparedness, 

such as that by NIST, are 

a step in the right direction, as the 

NIST report on the smart grid shows: 

Some of the responses [to cyber 

attacks] must be autonomic—timely 

response is a critical requirement for 

grid reliability. However, for a quick 

response to treat the symptom locally 

and effectively, the scope and extent 

of the impact of the failure needs to be 

quickly determined. Not all responses 

are autonomic however. New research 

is needed to measure and identify 

the scope of a cyber attack and the 

dynamic cyber threat response options 

available in a way that can serve as a 

decision support tool for the human 

operators. 

We believe that our 

research is on target 

regarding support for 

cyber command-and-control understanding, 

planning, and decision 

activities. Preparation must include 

provision of integrated full-spectrum 

capabilities in cyber C2/BM systems 

along with specialized dynamic virtual 

cells like the proposed legal cell 

enabling collaboration among cyberoperators 

proactively. 

James Bret Michael is a professor in 

the Department of Computer Science 

and a member of the Center for Cyber 

Warfare at the Naval Postgraduate 

School. Contact him at bmichael@ 

nps.edu. 

_____ 

John F. Sarkesain is a senior engineer 

with the Aerospace Corporation. Contact 

him at john.f.sarkesain@aero.org. 

Thomas C. Wing eld is a professor 

at the George C. Marshall European 

Center for Security Studies. Contact 

him at ____________________ 

thomas.c.wing eld@marshallcenter.org. 

_______ 

Georgios Dementis is a lieutenant 

in the Hellenic Navy. Contact him at 

ydementis@gmail.com. 

_______________ 

Gonçalo Nuno Baptista de Sousa is 

a lieutenant in the Portuguese Navy. 

Contact him at _______________ 

goncalobsousa@gmail. 

___ com. 

Editor: Jeffrey Voas, National Institute 

of Standards and Technology; 

j.voas@ieee.org 

_________ 

________ 

_________________ 

Disclaimer 

The views and conclusions 

contained herein are those 

of the authors and should not 

be interpreted as necessarily 

representing the official policies 

or endorsements of the US 

Government. 

A 


aG 

F

A 


aG 

F 

BOOKSHELF 

Topology for Computing, Afra 

J. Zomorodian. This emerging 

field utilizes theory from 

topology and the power of computing 

to solve problems in diverse fields. 

Recent applications include computer 

graphics, CAD, and structural 

biology, all of which involve understanding 

the intrinsic shape of some 

real or abstract space. The author 

strives to present basic concepts 

from topology and Morse theory so 

that nonspecialists can grasp and 

participate in current research in 

computational topology. 

This book’s self-contained presentation 

of mathematical concepts 

from a computer scientist’s viewpoint 

combines point set and algebraic 

topologies, group theory, differential 

manifolds, and Morse theory. The 

author also presents some recent 

advances in the area, including topological 

persistence and hierarchical 

Morse complexes. Throughout, the 

focus is on computational challenges 

and presenting algorithms and data 

structures when appropriate. 

Cambridge University Press; ____ www. 

cambridge.org; ISBN-978-0-521- 

13609-9, 258 pp. 

Symbolic Dynamics and Geometry 

Using D* in Graphics and Game 

Programming, Brian Guenter and 

Sung-Hee Lee. The authors explain 

how to use the D* symbolic-differentiation 

system for applications 

in computer games and engineering 

simulation. They describe how 

to create procedural 3D geometric 

models, link them together to form 

multibody physical systems, and 

simulate and display their physical 

behavior in real time. 

The symbolic-differentiation capabilities 

of D* can be used in a wide 

variety of technical applications, 

including computer graphics, engineering, 

and mechanical simulation. 

Many applications of D* are covered, 

but two—real time Lagrangianphysics 

simulation and procedural 3D 

0018-9162/10/$26.00 © 2010 IEEE 

geometric modeling—are developed 

in great detail. 

A K Peters, Ltd; www.akpeters. 

___ com; ISBN-978-1-56881-280-9, 

250 pp. 

Leading the Virtual Workforce: 

How Great Leaders Transform 

Organizations in the 21st Century, 

Karen Sobel Lojeski. Just 20 years 

ago, going to work meant driving 

to a physical location to interact 

face-to-face with a boss and coworkers. 

Today, this might only require 

stepping across the hall to a home 

office or joining a videoconference. 

According to the author, the implications 

of these changes are staggering 

and require a whole new leadership 

model. She asserts that the virtual 

workforce has exploded, with mobile 

workers set to become 73 percent of 

the US workforce by 2011. 

According to Lojeski, physical separation, 

technology mediation, and 

disconnected relationships characterize 

this new virtual distance. These 

dynamics lead to a psychological 

separation that negatively affects 

productivity, innovation, and trust. 

When virtual distance is relatively 

high, innovation falls by more than 

90 percent and project performance 

suffers. The author concludes that 

creating context, cultivating community, 

and co-activating new leaders 

builds greater trust, higher levels 

of satisfaction, and better behavior 

overall. 

Wiley; www.wiley.com; 978-0-470- 

42280-9; 512 pp. 

Touch of Class: Learning to Program 

Well with Objects and 

Contracts, Bertrand Meyer. The 

author provides an introductory 

programming text that can make 

learning to program fun and rewarding. 

Instead of traditional low-level 

examples, Meyer builds his presentation 

on a rich object-oriented 

software system supporting graphics 

and multimedia, which students can 


use to produce applications from day 

one. They can then explore “from the 

outside in” as they learn new programming 

techniques. 

Coverage is notable in both its 

depth and breadth. In addition to 

core programming concepts such 

as control structures, algorithms, 

and fundamental data structures, it 

encompasses recursion, reference 

and pointer manipulation, and topological 

sort as examples of algorithm 

and API design, event-driven programming, 

and high-level function 

closures. 

Springer; www.springer.com; 978- 

3-540-92144-8; 876 pp. 

Inside Larry and Sergey’s Brain, 

Richard L. Brandt. This book 

focuses on what drives these two 

men and where they will take Google. 

Looking at this brainchild of two brilliant 

minds, the author examines 

Google’s business decisions in light 

of its founders’ ambitions and beliefs. 

Larry is the main strategist, while 

Sergey is the primary technologist 

and idealist. Through interviews with 

current and former employees, competitors, 

partners, and senior Google 

management, the author demystifies 

the secret society that is Google while 

clarifying several misconceptions. 

Viking Press; http://us.penguingroup. 

___ com/; 978-1-59184-276-7; 256 pp. 

Send book announcements to 

___ 

newbooks@computer.org. 

APRIL 2010 93 

A 


aG 

F

A 


aG 

F 

94 

COMPUTER 

THE PROFESSION 

Currently, it has taken the form of an 

eBook reader with an easy-to-read 

display. However, even when we leverage 

an innovative distribution system 

using wireless technology, presenting 

static books falls short of employing 

the computer to full advantage. 

With larger displays, fiexible and 

lightweight units, and full color we 

may see more dramatic changes. 

Will eInk replace printed paper in 

all domains or be the savior of traditional 

newspapers? If the technology 

matures the consequences will be 

dramatic for many industries. But this 

particular if entails speculation. 

Instead of focusing on possible 

advances in new technologies, we 

observe that change may also come 

based on ingredients that have been 

present for some time. Email and TCP/ 

IP provide good examples. It took 

many years from when the technology 

and standards emerged to when 

they saturated society and made a 

major impact. 

POS 

The technology that provides 

our case study here is the simple 

point-of-sale (POS) terminal. This 

technology was not developed to 

change the world, but simply to 

make monetary transactions easier. 

By using a credit or cash card, the 

terminal lets us perform an alldigital 

transaction with no cash 

involved. The advantages of this 

are so substantial that in many 

countries a large percentage of all 

monetary transactions are digital, 

based on POS terminals or using 

credit cards over the Internet. Data 

is captured automatically and can be 

transferred at almost no cost. 

At the same time, this information 

can be used directly for inventory 

Continued from page 96 

control, accounting, and statistics. 

In my home country, Norway, people 

can avoid cash altogether, because 

all shops, hotels, restaurants, gas stations, 

cabs, and many buses allow 

digital payment. Most banks offer free 

service, and cards are used also for 

very small amounts. Even the pizza 

boy brings along a wireless POS terminal. 

Looking at value, 98 percent of 

all monetary transactions in Norway 

are digital. In Stockholm, people can 

even pay their bus fare by cell phone. 

Small countries have an advantage 

and can often offer high dissemination 

numbers for new technologies. 

National standards are easier to 

implement in a small country than in 

a large one. Fewer installations make 

it easier to update equipment, while 

a homogeneous population makes it 

Small countries have an advantage and can often offer 

high dissemination numbers for new technologies. 

simpler for people to accept new services. 

However, larger countries are 

following suit in the digital economy. 

A few years ago travelers needed 

cash when in cities such as Rome and 

Madrid. Today nearly all businesses 

accept a card. 

CASHfiFREE SOCIETY 

When only a small fraction of 

transactions are in cash an interesting 

question arises: Can we remove 

cash altogether? Money, in the form 

of coins, has been in use for nearly 

three thousand years. Paper bills 

have been used in Europe for nearly 

four hundred years, and much longer 

than that in China. It seems dramatic 

to convert to a true digital economy. 

There are, however, many advantages 

to going digital. Great savings 

can be achieved for businesses if we 

avoid storing, counting, handling, and 

transporting cash. 

The advantages for society 

as a whole may be even greater, 

especially if cash is removed altogether. 

As we learn from the news 

every day, cash encourages crime. 

The public may experience that a 

wallet is stolen from a pocket, or 

more severely, become the victim 

of armed robbery. Institutions such 

as banks and security transports 

must spend signiflcant resources to 

thwart criminals and sometimes still 

fail to deter them. Death and both 

physical and psychological injuries 

can be the consequences of holding 

all this “bait” in the form of cash. 

It would be naive to expect crime 

to disappear in a digital economy. 

Credit card fraud is already a daily 

occurrence. However, today it is stimulated 

by the possibility of retrieving 

the stolen money in cash from ATMs. 

Without cash it will be much easier to 

follow the transactions. 

We can, however, realistically 

hope to make being a criminal less 

convenient. A monetary economy is 

much more effective than bartering. 

People understood this thousands 

of years ago. It was easier then to 

sell a horse for money that could be 

spent freely than to trade it for a cow 

and some chickens on the side. The 

seller might not need chickens, or the 

potential horse buyer might not have 

a cow for sale. 

What we could expect by going 

digital is to enjoy the advantages of 

the monetary system without being 

threatened by criminals. The pusher 

down the street, for example, would 

have problems if customers no longer 

had anonymous cash handy for their 

transactions. They would be forced to 

accept gold, PCs, TVs, and refrigerators 

instead, thereby confronting all 

the hassles of a barter economy. 

DIGITAL CRIME 

Criminals can also go digital, 

although this requires computing 

competence. Today, with improved 

information security systems, successful 

cybercrime implies a fairly 

high level of competence. Such skill 

can also be sold in the open economy 

A 


aG 

F

A 


aG 

F 

at lucrative prices, and digital criminals 

can be thwarted by digital police. 

Some tasks, such as detecting whitewashing, 

might even be performed 

automatically. 

Do we see big brother here? Cash 

is anonymous while digital money 

can be tracked. For ordinary lawabiding 

citizens this should not 

be considered a threat as long as 

good laws to protect public privacy 

remain in place. However, 

tax authorities now have a weapon 

against the black market economy. 

For example, it won’t be so easy to 

pay the carpenter, off-the-record, 

without cash. Some might view this 

as a disadvantage, but we will gain 

instead a more fair tax system and 

increased state income. 

Willing or not, we are moving 

slowly but smoothly into a cashfree 

society. Today the $100 bill is 

the highest denomination in circulation, 

with $500, $1,000, $5,000, 

and $10,000 bills no longer used. 

As inflation takes a percentage of 

the real value of the $100 bill every 

year, we will soon only have cash 

as small change. On the fiip side, it 

is interesting to note that the US still 

has a one-cent coin. In Norway, the 

lowest valued coin is now flfty øre or 

approximately nine cents, up from 

the one øre that was removed from 

circulation in 1972. 

The process toward a cash-free 

society will be accelerated when the 

real costs of using cash are placed on 

the individual cash-paying customer, 

not shared among all customers. The 

growth of true Internet banks, where 

all transactions are digital, poses a 

severe threat to banks that still have 

high cash handling costs. It might 

also become more awkward to use 

cash. Many countries now require a 

national bank notiflcation whenever 

large sums are withdrawn or deposited 

as cash. We could go further, 

perhaps placing a risk tax on the 

use of cash to compensate for public 

injury and other expenses that follow 

from robberies. 

RISK 

Computing professionals must 

be concerned about the risk of getting 

into a situation where people 

no longer have cash as a backup. 

Truthfully, though, we have already 

reached this stage. Look into your 

wallet and count the bills. Is that 

amount sufflcient for food, rent, gas, 

and all other expenses? For how long? 

While there have been incidents of 

credit card and banking system hacking, 

computer security is steadily 

improving. As with cars, safety only 

became an issue late in the technology’s 

development. First we had to 

solve the “important” problems: getting 

the systems to run, making their 

services affordable, adding core functionality, 

and offering a commercially 

aesthetic design. 

Most likely, one of the small countries 

will flrst take the leap toward 

an all-digital economy. As I have 

noted, it is easier to get the necessary 

infrastructure in place. This 

may be done to issue cash cards 

to persons who do not qualify for 

commercial banking services, or to 

implement improved ID systems and 

other safeguards. It is also easier to 

remove cash from a currency in use 

on a national level only, unlike the 

dollar and euro, which already have 

widespread acceptance. 

Given that there are clear 

advantages to being the 

flrst country to go all digi- 

IEEE 

THE #1 ARTIFICIAL 

INTELLIGENCE 

MAGAZINE! 

tal, we can expect this development 

to happen sooner rather than later. 

Crimes will be averted, and national 

industry and service providers will 

have a home market for technologies 

that might be in widespread use in 

just a few years. 

This transition to a digital economy 

can be performed in measured 

steps, starting with removing the 

high-denomination bills and placing 

cash-handling costs on the 

shoulders of cash-paying customers. 

This process could persist until the 

economy is fully digital, or banks 

might see a beneflt in keeping smalldenomination 

bills and coins in circulation. 

As computing professionals, 

we are privileged to take part in these 

dramatic historic developments. 

Ultimately, we should consider it an 

improvement if bandits replace guns 

with keyboards—there will at least 

be fewer injuries. 

Kai A. Olsen is a professor at Molde 

College and the University of Bergen, 

and is an adjunct professor at the University 

of Pittsburgh. Contact him at 

_______________ 

Kai.Olsen@himolde.no. 

Editor: Neville Holmes, School of Computing 

and Information Systems, University of 

Tasmania; neville.holmes@utas.edu.au 

________________ 


are available for free at ____ 

http:// 

_________________ 

ComputingNow.computer.org. 

IEEE Intelligent Systems delivers 

the latest peer-reviewed research on 

all aspects of artificial intelligence, 

focusing on practical, fielded applications. 

Contributors include leading experts in 

 

 

 

Visit us on the Web at 

www.computer.org/intelligent 

APRIL 2010 

A 


aG 

F 

95

A 


aG 

F 

96 

COMPUTER 

THE PROFESSION 

Computing a 

Better World 

Kai A. Olsen, 

Molde College and University of Bergen, Norway 

While government leaders struggle to implement significant 

change, computing professionals increasingly achieve this goal 

as a side-effect. 

As computing professionals 

we have been 

involved in many projects. 

Most have had a 

limited scope, duration, and budget. 

These milestones span decades, from 

developing a system for sending text 

messages between universities and 

research centers in 1969, to manufacturing 

a chip for a calculator in 

1971, to setting standards for electronic 

communication in 1974, to 

designing a computer system that 

offers an improved user experience 

in 1981, to simplifying access to scientiflc 

papers in 1990, to developing 

new technologies for ranking Web 

pages in 1996, and to creating an 

online student directory with photos 

in 2003. 

I suspect few people involved in 

these projects had any idea they were 

changing the world. Yet, in retrospect, 

we see this is what really happened. 

The Arpanet project gave us the flrst 

e-mail system in 1969, Intel developed 

the microprocessor in 1971, and an 

international effort gave us the TCP/ 

IP standard for data communication 

in 1974. Then, in 1981, researchers 

at Xerox Parc designed the flrst 

complete graphical workstation; Tim 

Berners Lee deflned the basis for the 

World Wide Web at CERN in 1990, 

Larry Page and Sergey Brin created 

the foundations for Google in 1996, 

and Mark Zuckerberg created Facebook 

in 2003. 

CHANGING LIVES 

Combined, these computing 

advances have changed the lives of 

millions throughout the world. In the 

industrial countries we work, shop, 

play, and socialize using a computer. 

Whole business areas are experiencing 

this revolution. Music, movies, 

and books are now separated from 

their physical representation, with 

a consequently huge impact on the 

industry. Photography has gone 

digital. Analog products, such as 

fllm, cameras, and their connected 

services—are now disappearing. 

Newspapers that have existed for 

more than 200 years are in danger 

of closing down as the competition 

from the net becomes overwhelming. 

Meanwhile, TV is losing viewers to the 

Internet and sites like YouTube. 

Physical libraries, with collections 

in the form of clay tablets, papyrus, 

or books, have been an integral 

part of educational institutions for 

thousands of years. Today digital 

collections are taking over and the 

physical library has an uncertain 

future in many domains. Companies 

that hardly existed flve to ten years 

ago—such as Google, Facebook, and 

Twitter—are today known by everybody. 

New services and applications 

are offered every day to make our 

lives more effective and, perhaps, 

more interesting. It is tempting to 

compare these trends with politics. 

While many state leaders struggle to 

implement signiflcant change, computing 

professionals increasingly 

achieve this as a side-effect. 

CATALYST 

Clearly, the computing industry 

is today’s main catalyst for change. 

No other industry has a comparable 

impact and none is as dynamic. As 

professionals in this area we are 

truly shaping the world into something 

better—or so we hope. Some 

argue that computing has become 

a commodity, like electricity, something 

you still need but that has lost 

its competitive advantage. However, 

the Googles, YouTubes, Wikipedias, 

and Twitters show that the pace has, 

if anything, accelerated. So what can 

we expect in the next decennium? 

eInk offers one new technology 

that flows from computing. In the 

past 10 years it has moved from the 

labs to commercial product status. 

Continued on page 94 


A 


aG 

F

A 


aG 

F 

_____________________ 

____________________________ 

A 


aG 

F

A 


aG 

F 

Running in 

Circles 

Looking for 

a Great 

Computer 

Job or Hire? 

The IEEE Computer Society Career Center is the best niche 

employment source for computer science and engineering jobs, with 

hundreds of jobs viewed by thousands of the finest scientists each 

month - in Computer magazine and/or online! 

_____________________________________________________ 

http://careers.computer.org 

The IEEE Computer Society Career Center is part of the Physics Today Career Network, a niche job board 

network for the physical sciences and engineering disciplines. Jobs and resumes are shared with four partner 

job boards - Physics Today Jobs and the American Association of Physics Teachers (AAPT), American Physical 

Society (APS), and AVS: Science and Technology of Materials, Interfaces, and Processing Career Centers. 

> Software Engineer 

> Member of Technical Staff 

> Computer Scientist 

> Dean/Professor/Instructor 

> Postdoctoral Researcher 

> Design Engineer 

> Consultant 

A 


aG 

F

computing lives - FTP Directory Listing

Create successful ePaper yourself

Delete template?

Save as template?