selfish genes and kelly bets

techmologist

Does natural selection tend to result in genes that maximize their longterm growth rate (growth of the number of copies of themselves), or in genes that simply maximize the expected number of copies of themselves in the next generation? Since replication is a chance process, these are not generally the same thing.

Darwin123

Copies of themselves in the next generation. Genes have no way to predict the future.

Most species have gone extinct. Some species have gone extinct because of the long term consequences of their adaptation to short lived environmental conditions.

We may go extinct that way ourselves. However, history isn't done with us yet. So let us look at some other example.

A good example would be natural selection for size. Sudden catastrophes have tended to destroy the largest species of animals rather than the smallest. Having a huge adult size is an advantage under some short lived environmental conditions. While your biggest enemy is a member of the same genus as you are, being the biggest may be a deciding factor. When the meteor/magma flow/supernova comes, being large is not a great adaptation.

techmologist

Good points. The idea of longterm growth is not directly applicable anyway, since it assumes unchanging conditions and, in particular, resources that do not run out. However, if conditions are the same and resources are plentiful for many generations, I was wondering if natural selection might be able to weed out those genes that have a risky (high variance) strategy for getting into the next generation, even if the expected number of copies is higher than that for rival genes. Like you said, most species go extinct and their genes go with them. But some persist for a long time. I wonder if that is more than just good luck (or bad luck on the part of the species that went extinct, as when the conditions change abruptly).

enosis_

Perhaps we should define how many generations define longterm?

techmologist

I wouldn't know how to give a number, but here is what I have in mind. Long enough for a stable exponential growth to set in. Enough generations so that the long term growth rate becomes nearly equal to its expected value with high probability (law of large numbers).

Evo

Sorry, this violates our rules against overly speculative posts.