Darwin's Dangerous Idea - Evolution and the Meaning of Life

208 BIOLOGY IS ENGINEERING The Computer That Learned to Play Checkers 209
been to use his wonderful machine to explore the mysteries of thought. 13
Turing published his prophetic essay, "Computing Machinery and Intelligence,"
in the philosophical journal Mind in 1950, surely one of the most
frequently cited articles ever to appear in that journal. At the time he wrote it,
there were no Artificial Intelligence programs—there were really only two
operating computers in the world—but within a few years, there were enough
machines up and running twenty-four hours a day so that Arthur Samuel, a
research scientist at IBM, could fill the otherwise idle late-night time on one
of the early giants with the activities of a program that is as good a candidate
as any for the retrospective title of AI-Adam. Samuel's program played
checkers, and it got better and better by playing against itself through the
small hours of the night, redesigning itself by throwing out earlier versions
that had not fared well in the nightly tournament and trying out new
mutations that were mindlessly generated. It eventually became a much better
checkers-player than Samuel himself, providing one of the first clear
counterexamples to the somewhat hysterical myth that "a computer can really
only do what its programmer tells it to do." From our perspective we can see
that this familiar but mistaken idea is nothing but an expression of Locke's
hunch that only Minds can Design, an exploitation of ex nihilo nihil fit that
Darwin had clearly discredited. The way Samuel's program transcended its
creator, moreover, was by a strikingly classical process of Darwinian
evolution.
Samuel's legendary program is thus not only the progenitor of the intel-
13. In fact, the bridge between computers and evolution goes back even farther, to
Charles Babbage, whose 1834 conception of the "Difference Engine" is generally credited
with inaugurating the prehistory of the computer. Babbage's notorious Ninth Bridgewater
Treatise (1838) exploited his theoretical model of a computing engine to offer a
mathematical proof that God had in effect programmed nature to generate the species!
"On Babbage's smart machine any sequence of numbers could be programmed to cut in,
however long another series had been running. By analogy, God at Creation had appointed
new sets of animals and plants to appear like clockwork throughout history—he
had created the laws which produced them, rather than creating them direct" ( Desmond
and Moore 1991, p. 213). Darwin knew Babbage and his Treatise, and even attended his
parties in London. Desmond and Moore (pp. 212-18) offer some tantalizing glimpses
into the traffic of ideas that may have crossed this bridge.
More than a century later, another London society of like-minded thinkers, the Ratio
Club, served as the hotbed for more recent ideas. Jonathan Miller drew my attention to
the Ratio Club, and urged me to research its history in the course of writing this book,
but I have not made much progress to date. I am tantalized, however, by the 1951
photograph of its membership that graces the front of A. M. Uttley's Information Transmission
in the Nervous System (1979 ): Alan Turing is seated on the lawn, along with the
neurobiologist Horace Barlow (a direct descendant of Darwin, by the way); standing
behind are Ross Ashby, Donald MacKay, and other major figures of the earliest days of
what has become cognitive science. It's a small world.
lectual species, AI, but also of its more recent offshoot, AL, Artificial Life.
Legendary though it is, few people today are familiar with its remarkable
details, many of which deserve to be more widely known. 14 Samuel's first
checkers program was written in 1952, for the IBM 701, but the learning
version wasn't finished until 1955, and ran on an IBM 704; a later version ran
on the IBM 7090. Samuel found some elegant ways of coding any state of a
checkers game into four thirty-six-bit "words" and any move into a simple
arithmetical operation on those words. (Compared with today's prodigiously
wasteful computer programs which run on for megabytes, Samuel's basic
program was microscopic in size—a "low-tech" genome indeed, with fewer
than six thousand lines of code—but, then, he had to write it in machine
code; this was before the days of computer programming languages.) Once
he'd solved the problem of representing the basic process of legal checkers
play, he had to face the truly hard part of the problem: getting the computer
program to evaluate the moves, so it could select the best move (or at least
one of the better moves) whenever possible.
What would a good evaluation function look like? Some trivial games, like
tic-tac-toe, have feasible algorithmic solutions. There is a guaranteed win or
draw for one player, and this best strategy can be computed in realistic
amounts of time. Checkers is not such a game. Samuel (p. 72) pointed out
that the space of possible checkers games has on the order of 10 40 choicepoints,
"which, at 3 choices per millimicrosecond, would still take 10 21
centuries to consider." Although today's computers are millions of times
faster than the lumbering giants of Samuel's day, they still couldn't make a
dent on the problem using the brute-force approach of exhaustive search. The
search space is Vast, so the method of search must be "heuristic"—the
branching tree of all possible moves has to be ruthlessly pruned by semiintelligent,
myopic demons, leading to a risky, chance-ridden exploration of a
tiny subportion of the whole space.
Heuristic search is one of the foundational ideas of Artificial Intelligence.
One might even define the task of the field of AI as the creation and inves-
14. Samuel's 1959 paper was reprinted in the first anthology of Artificial Intelligence,
Feigenbaum and Feldman's classic, Computers and Thought ( 1964 ). Although I had read
that paper in Feigenbaum and Feldman when it first came out, I had, like most readers,
passed over most of the details and savored the punch line: a 1962 match between the
"adult" program and Robert Nealey, a checkers champion. Nealey was gracious in defeat:
"In the matter of the end game, I have not had such competition from any human being
since 1954, when I lost my last game." It took a superb lecture by my colleague George
Smith in an introductory course in computer science that we cotaught at Tufts to rekindle
my interest in the details of Samuel's article, in which I find something new and
valuable every time I reread it.