His computer then repeatedly copies this wholly nonsensical message. However, at each iteration there is a certain probability of a mutation, of a random change in one of the letters. Selection is also simulated, because the computer is programmed to retain any mutations that move the sequence of letters even slightly toward a pre-selected goal, a particular, quite different sequence of twenty-eight letters. (Of course natural selection does not have some final ACGT sequence in mind, but—in preferentially replicating sequences that improve, even by a little, the fitness of the organism—it comes down to the same thing.) Dawkins’s arbitrarily chosen twenty-eight-letter sequence, toward which his selection was aiming, was
METHINKS IT IS LIKE A WEASEL.
(Hamlet, feigning madness, is teasing Polonius.)
In the first generation, one mutation in the random sequence occurs, changing the “K” (in DTJBKW …) to an “S.” Not much help yet. By the tenth generation, it reads
MDLDMNLS ITJISWHRZREZ MECS P,
and by the twentieth,
MELDINLS IT ISWPRKE Z WECSEL.
After thirty generations, we are at
METHINGS IT ISWLIKE B WECSEL,
and by forty-one generations, we’re there.
“There is a big difference,” Dawkins concludes, “between cumulative selection (in which each improvement, however slight, is used as a basis for future building), and single-step selection (in which each new ‘try’ is a fresh one). If evolutionary progress had had to rely on single-step selection, it would never have got anywhere.”7
Randomly varying the letters is an inefficient way to write a book, you might be thinking. But not if there are an enormous number of copies, each changing slightly generation upon generation, the new instructions constantly tested against the demands of the outside world. If human beings were devising the volumes of instruction contained in the DNA of the given species, we would, we might offhand imagine, just sit down and write the thing out, front to back, and tell the species what to do. But in practice we are wholly unable to do this, as is DNA. We stress again, the DNA hasn’t the foggiest notion a priori about which sequences are adaptive and which are not. The evolutionary process is not omnicompetent, far-seeing, crisis-avoiding, top-down. It is instead trial-and-error, short-term, crisis-mitigating, bottom-up. No DNA molecule is wise enough to know what the consequences will be if one segment of a message is changed into another. The only way to be sure is to try it out, keep what works, and run with it.
The more you know how to do, the more advanced you are—and, you might think, the better your chances for survival. But the DNA instructions for making a human being comprise some 4 billion nucleotide pairs, while those for a common one-celled amoeba contain 300 billion nucleotide pairs. There is little evidence that amoebae are almost a hundred times more “advanced” than humans, although the proponents of only one side of this question have been heard from to date. Again, some, maybe even most, of the genetic instructions must be redundancies, stutters, untranscribable nonsense. Again we glimpse deep imperfections at the heart of life.
Sometimes another organism inconspicuously slips through the defenses of the eukaryotic cell and steals into the heavily guarded inner sanctum, the nucleus. It attaches itself to the monarch, perhaps to the end of a time-tested and highly reliable DNA sequence. Now messages of a very different sort are dispatched out of the nucleus, messages that order the manufacture of a different nucleic acid, that of the infiltrator. The cell has been subverted.
Besides mutation, there are other ways (including infection and sex, to which we turn shortly) whereby new hereditary sequences arise. The net result is that a huge number of natural experiments are performed in every generation to test the laws, doctrine, and dogma encoded in the DNA. Each eukaryotic cell is such an experiment. Competition among DNA sequences is fierce; those whose commands work even slightly better become fashionable, and everyone has to have one.