> With those odds populations cant have everyone taking risk. Such populations sooner or later go extinct.

Not trying to be a jerk, but there is a logical fallacy here. If you've ever read the Selfish Gene the central idea is that there is a common misunderstanding that animals/societies evolve for "the good of the population". A population is better understood as a collection of individuals with each being a collection of genes, and really it is each gene that is trying to replicate itself. Applying it to your example, a risk-taking gene with .998% chance of failure would probably not replicate itself successfully, unless the .002% of individuals that succeed were quite prolific at procreation. The good of the population does not really come into it.

Selfish Gene story is detached from how systems survive environmental change and unpredictability. Thats where Explore Exploit Ratio comes in. And how many Explorers a population can sustain has an upper limit. Which is always low. The Limit is not dependent on whether people are explorers or exploiters but on resources available in the environement around them which is constantly changing.