| Evolution is a relatively poor search algorithm that tends to get stuck at local maxima. Unlike science, it doesn't build maps and models for exploration and extrapolation. So far as I know there's no general theory of theories which quantifies this, so there's no way to make predictions about the cut-off point for evolutionary invention. But in a hand-wavy way, evolution's only feedback loop is first-order and binary - mutate and reproduce at a positive replacement rate, or not. The feedback loop in science is more complex. Instead of being driven by a random search, "mutations" are guided by a creative model. This creates momentum in the model space which isn't available to evolution - which in turn makes it possible to discover more complex and less immediately accessible solutions. It also makes it possible to build systems whose value is guaranteed, or at least strongly suspected, before resources are diverted to making them physical. The bottom line is evolution is only ever going to find a small subset of all possible biological configurations, and that space is going to exclude many features that are available to science-driven search. (Of course you can argue that scientific meta-search was a product of evolution anyway, so the distinction is academic.) |