Design and Nested Hierarchies

[Rumraket (Mikkel R.)]

@Giltil

Regarding this topic, @gpuccio has written a very interesting piece (see below).

I’ve read it before. It’s flatly wrong of course. There are no targets in evolution.

You may disagree with him, but if you read his piece carefully and without prejudice, you will have no choice but to acknowledge the rationality of his arguments.

I have read it carefully and without prejudice, and his argument flat out fails. It’s so simple to see why. The things that evolve, evolve due to a combination of chance (historical contingency in the types of mutations that happens to occur), and selection. Selection isn’t towards any particular thing(hence not towards any targets), it’s just towards fitness. What happens to have a positive effect on fitness isn’t known beforehand by any organism, or by the process of mutation.

There was never a search for some particular gene, or sequence, or mutation(so no targets). Mutations occurred, and they had the phenotypic effects that they did which in turn affected their propensity to stick around and rise in frequency in coming generations. Some particular sequence (whether that is a protein coding gene or not) is in this way a historically contingent outcome, not a target. Down the line, any particular long such series of accumulating mutations will look like something that was unlikely to have happened. But that doesn’t make it a target.

In fact, you yourself agrees that looking at historically contingent outcomes of accumulating mutations as extremely unlikely targets commits the Texas sharpshooter fallacy. Let me remind you:

@Giltil

Rumraket:
I raise the empirical counterxample of the complete genome sequence of me, and you, and anyone currently alive, compared to the genome sequence of our ancestor ten generations ago.
It is estimated that every human being is born with approximately 100 novel mutations. What is the probability that any of us gets the particular 100 mutations we do? Assuming only substitutions to keep it simple, it’s roughly one in (3.2x10^9 * 3)^100 = ~1.7x10^998.

Now compound the probabilities ten consecutive generations. (1.7x10^998)^10 = ~1.9x10^-9982
Roughly one in two times ten to the ninethousand ninehundred and eightysecond power.

You are committing here the Texas sharpshooter fallacy.

As was exactly my point.

So here you are, looking at some particular protein, or another genetic sequence, you calculate it would be unlikely for some accumulating series of mutations to result in that sequence(thus treating it as a target), and then you declare it couldn’t have happened. But as you correctly saw back then, that argument commits the Texas sharpshooter fallacy.