[robotresearcher]

> evolution is a massive, groping search

It's not a search. There is no goal, no searcher, and nothing to find. The word evolution was chosen deliberately in its meaning of a dynamical system moving through state space. Not searching: just evolving.

[samizdatum]

In a banal sense, it's correct to say that there's no searcher, no goal to evolution. As far as we know, "natural selection" isn't "real" in the sense that electromagnetism is "real"; natural selection is just a staggeringly convenient shorthand that refers to the multitude of ways organisms can die. But if we go further, and think of evolution as a mere random walk through state space, we've suddenly lost all explanatory power. The independent emergence of echolocation in whales and bats, for instance, becomes a brute mystery.

Evolution only makes sense if we think of it as as some kind of massively parallel Monte Carlo that's weighted to explore regions of the state space with high fitness. So it's perfectly valid to talk, as biologist do, of selection pressures and directional selection, because- and here we've come full circle- there is a goal to evolution: to exist.

[JoeAltmaier]

And, to exist with a minimum of fuss. Why do people get cancer, or get feeble when old? Well, if the organism can reproduce and support their offspring well enough, given the eco-niche and available calories etc, then they are Fit. So, just enough avoiding cancer to work out in the end.

Kind of like the space program, where everything is made by the lowest bidder.

[haddr]

While Evolution clearly doesn't have any "goal" I wonder if we can say that it directs to the goal that is constantly changing (that is, a goal only known once some organism is there)?

[robotresearcher]

No, you can't. There is no look ahead at all. No gradients are available to measure. Without looking ahead, the notion of a goal is meaningless.

[antisthenes]

It's a random walk with probabilities changing from 1 generation to the next based on surviving geno/phenotypes.