Chapter 7: Junk DNA and a Demolition Tool
A Reality Check
There's a biological study that bears on this. Researchers took a virus that infects bacteria, removed a section of one of its proteins that the virus needs in order to infect, and put a random chain of 139 amino acids in its place. Then they let mutation and selection go to work, round after round, keeping whichever versions infected best. In word-ladder terms, it's like starting from a string of gibberish and asking whether one-letter changes, each kept only if the result reads a little better, can climb all the way to a working sentence.
The early climbing came fairly easily. Then it stalled: above a certain level, the landscape turned rugged, full of small local peaks where every single-letter change made things worse. The researchers estimated that reaching the level of the original, working protein by this route would require something on the order of 1070 variants — a number with no realistic meaning in a universe that has existed for about 1017 seconds. Even the slowest result in the word-ladder game above, at 7,422 letters tried, is nothing next to a number like that. The researchers' own conclusion was that random substitution alone could get only part of the way, and that something else, such as the shuffling of larger chunks of existing genetic material, would have to be involved as well. If a chain of point mutations struggles this badly just to climb to the level of one working protein, the walk from one distinct, functioning protein to another is unlikely to be any easier.
I don't want to overstate what the word-ladder numbers show. They're specific to English, to this particular word list, and to how many five-letter combinations happen to form real words in this language — none of which necessarily has anything to do with how proteins are distributed in whatever the biological equivalent of "word space" actually is. We simply don't know enough about the relevant variables in biology to say. It is not a settled fact that small, single-point changes can reliably carry something from one functional arrangement to another.
What this modest game does show is that the task isn't automatically easy — that real obstacles exist even in a far smaller, far friendlier space than biology's. To assume this is simply what happened, without knowing whether the necessary bridges even exist, is to assume the process is practically possible. At this point, that's an assumption. It isn't a proven fact.
Indeed, as best we can tell, biology seems to work, to some extent, the same way as the word-ladder game — and the evidence, if anything, points toward biology having it harder, not easier. We've already seen research showing that even a small handful of substitutions — changes that are each perfectly tolerated on their own — can destroy a protein's function entirely when combined. Individually harmless doesn't mean collectively harmless. And the distance from random sequence to working protein, as the virus study shows, seems too large to cross one letter at a time.
Last revised 2026-11-19