[donh]

Reconcile this with the fact that places like MIT are copying the evolutionary process and using genetic algorithms as efficient problem solving strategies?

Sometimes effective, but really pretty hard to call efficient, problem solving strategies, given the right problem, entailing debugging and support challenges of substantial daunt-itude.

[Doctor Stochastic]

Evolutionary algorithms (computer or natural) can be more efficient that one might assume. The selection phase converges exponentially fast to the (currently active) fitness function. The drift (mutation) phase moves rather fast too. While the average drift motion increases proportional to Sqrt(Time), the extremes go like Time. There are drifts that move so fast that the have infinte variation and average; these cover regions that are far apart; bees searching for flowers for example. I don't know if any such drifts occur in genetics, though.

[Condorman]

Reconcile this with the fact that places like MIT are copying the evolutionary process and using genetic algorithms as efficient problem solving strategies?

Sometimes effective, but really pretty hard to call efficient, problem solving strategies, given the right problem, entailing debugging and support challenges of substantial daunt-itude.

True enough. Also worth noting that the massive trial-and-error process is not efficient in and of itself, but due to the speed of modern computers, the speed with which we can process the results makes it appear so.