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

160 PRIMING DARWIN'S PUMP Molecular Evolution 161
fragments of self-replicating nucleotide strings that will eventually come to
"specify" them in the macros they compose. Eigen shows how the vicious
circle can turn friendly if it is expanded into a "hypercycle" with more than
two elements (Eigen and Schuster 1977). This is a difficult technical concept,
but the underlying idea is clear enough: imagine a circumstance in which
fragments of type A can enhance the prospects of hunks of B, which in turn
promote the well-being of bits of C, which, completing the loop, permit the
replication of more fragments of A, and so forth, in a mutually reinforcing
community of elements, until the point is reached where the whole process
can take off, creating environments that normally serve to replicate longer
and longer strings of genetic material. (Maynard Smith 1979 is a great help in
understanding the idea of a hypercycle; see also Eigen 1983.)
But even if this is possible in principle, how could it get started? If all
possible proteins and all possible nucleotide "texts" were truly equiproba-ble,
then it would be hard to see how the process could ever get going. Somehow,
the bland, mixed-up confetti of ingredients has to get some structure imposed
on it, concentrating a few "likely-to-succeed" candidates and thereby making
them still more likely to succeed. Remember the coin-tossing tournament in
chapter 2? Somebody has to win, but the winner wins in virtue of no virtue,
but simply in virtue of historical accident. The winner is not bigger or
stronger or better than the other contestants, but is still the winner. It now
seems that something similar happens in prebiotic molecular evolution, with
a Darwinian twist: winners get to make extra copies of themselves for the
next round, so that, without any selection "for cause" (as they say when
dismissing potential jurors), dynasties of sheer replicative prowess begin to
emerge. If we start with a purely random assortment of "contestants" drawn
from the pool of self-replicating fragments, even if they are not initially
distinguishable in terms of their replicative prowess, those that happen to win
in the early rounds will occupy more of the slots in the subsequent rounds,
flooding the space with trails of highly similar (short) texts, but still leaving
vast hypervolumes of the space utterly empty and inaccessible for good. The
initial threads of proto-life can emerge before there is any difference in skill,
becoming the actuality from which the Tree of Life can then grow, thanks to
tournaments of skill. As Eigen's colleague Bernd-Olaf Küppers (1990, p.
150) puts it, "The theory predicts that biological structures exist, but not what
biological structures exist." 3 This is
3. Kiippers (1990, pp. 137-46) borrows an example from Eigen (1976) to illustrate the
underlying idea: a game of "non-Darwinian selection" you can play on a checkerboard
with differently colored marbles. Start by randomly placing the marbles on all the squares,
creating the initial confetti effect. Now throw two (eight-sided!) dice to determine a
square (column 5, row 7, for instance) on which to act. Remove the marble on that
enough to build plenty of bias into the probability space from the outset.
So some of the possible macros, inevitably, are more probable—more
likely to be stumbled upon in the Vast space of possibilities—than others.
Which ones? The "fitter" ones? Not in any nontrivial sense, but just in the
tautological sense of being identical to (or nearly identical to) previous
"winners," who in turn tended to be almost identical to still earlier "winners."
(In the million-dimension Library of Mendel, sequences that differ at a single
locus are shelved "next to" each other in some dimension; the distance of any
one volume from another is technically known as the Hamming distance.
This process spreads "winners" out gradually—taking leaps of small
Hamming distances—from any initial starting point in any and all directions
in the Library.) This is the most rudimentary possible case of "the rich get
richer," and since the success of the string has an explanation with no
reference beyond the string itself and its resemblance as a string to its parent
string, this is a purely syntactic definition of fitness, as opposed to a semantic
definition of fitness (Kiippers 1990, p. 141). That is, you don't have to
consider what the string means in order to determine its fitness. We saw in
chapter 6 that mere typographical change could never explain the Design that
needs explaining, any more than you could explain the difference in quality
between two books by comparing their relative frequencies of alphabetic
characters, but before we can have the meaningful self-replicating codes that
make this possible, we have to have self-replicating codes that don't mean a
thing; their only "function" is to replicate themselves. As Eigen (1992, p. 15)
puts it, "The structural stability of the molecule has no bearing upon the
semantic information which it carries, and which is not expressed until the
product of translation appears."
This is the birth of the ultimate QWERTY phenomenon, but, like the
cultural case that gives it its name, it was not entirely without point even
from the outset. Perfect equiprobability could have dissolved into a monopoly
by a purely random process, as we have just seen, but perfect
equiprobability is hard to come by in nature at any point, and at the very
beginnings of this process of text generation, a bias was present. Of the four
bases—A, C, G, and T—G and C are the most structurally stable: "Calculation
of the necessary binding energies, along with experiments on binding
square. Throw the dice again; go to the square they name and check the color of the
marble on this square and put a marble of that color on the just-vacated square ( "reproduction"
of that marble). Repeat the process, over and over. Eventually, it has the effect
of unrandomizing the initial distribution of colors, so that one color ends up "winning"
but for no reason at all—just historical luck. He calls this "non-Darwinian selection"
because it is selection in the absence of a biasing cause; selection without adaptation
would be the more familiar term. It is non-Darwinian only in the sense that Darwin didn't
see the importance of allowing for it, not in the sense that Darwin ( or Darwinism ) cannot
accommodate it. Manifestly it can.