[nora4]

I don't think so. A/B testing is like gradient descent which is a greedy algorithm. You move in the direction that locally looks best. Evolution on the other hand allows for suboptimal species to persist for enough time to let them develop their advantage. (In the language of optimization evolution allows you to go past the local optimal and reach global optima by allowing you to move in non-optimal direction -- as long as the move is not catastrophic.)

[mygo]

Nope. A/B testing depends on what you choose to mutate. The problem is that we humans intentionally change the things. Nature randomly changes the things.

You’re not going to get to a global optimum driven by human choice of what to test (only local optima at best) unless the human setting up the tests is some sort of sage.

[btilly]

Do you Nature will eventually hit on a global optimum? If so, that belief is opposed by what biologists know about developmental and evolutionary pathways.

For a trivial and well-known example, there is now no developmental or evolutionary pathway that will lead to a vertebrate eye where the nerves run behind the retina. As a result every last vertebrate has a blind spot where all of the nerves dive through the retina. This makes our eyes less efficient because the nerves block out light.

We're therefore stuck with the first design of the vertebrate eye and can't change it. There is no pathway to the more sensible design of the mollusk eye.

This is but one of many examples. For another one, relative developmental timing of growth is fixed across vertebrates. For example the "hand" grows before the "arm". The result is that pterodactyls, whose wings are entirely hands, could fly from birth. But birds (whose wings are arms) and bats (a mix of the two) can't fly from birth. No matter how desirable an evolutionary trait that may be.

[mygo]

> Do you think Nature will eventually hit on a global optimum?

No. At least not in the way described. Optimal as considered by whom? Us? We look at something and say this would function better than that, but is something that is optimal perfect? Basic microeconomics shows that perfection is sub-optimal due to the law of diminishing marginal return. Perfection is wasteful and unsustainable. What do we know to be optimal? From what I know, “adequate enough to reproduce” may not be perfect, but it might be brushing up against optimal in its “good enough and no more” / lagom nature. Mollusks have mollusk eyes. Can we know that humans would be better off with mollusk eyes without rewriting our evolutionary lineage for us to be more like mollusks in other imperfect ways?

[Franciscouzo]

Now I wonder, has someone ran A/B tests but instead of using hill climbing for optimizing used simulated annealing?