In a message dated 9/25/98 10:00:54 AM, hanson@econ.berkeley.edu wrote:
>This thread started when Eliezer referred to punctuated equilibrium
>as support for his claim that future intelligence trajectories will
>mainly consist of big fast jumps between periods of little change.
Punctuated equilibrium is a poor model for that. Change can be fast but increases in ability are quite slow. You may have a blizzard of new fish species in the African lakes, but they're all still just fish and not fundamentally different from fish everywhere.
>Even if "the combined effects of the most important genes explains most
>of the variation between species", it is not clear to me that this implies
>most biological innovation happens at speciation.
The idea is that if you look at variation within a species you see relatively little, even when looking at geographically isolated groups; if you look at variation between species you see a lot. So as long as a species remains a single species not much changes; but during the speciation process a lot happens.
>And even for genes
>which do "follow the exponential model", the crucial question is "over
>how many orders of magnitude"?
Wouldn't you say that if the exponential model fits for the important changes during speciation, it's not really relevant whether it fits minor changes? After all, had the minor changes not happened you'd still have two distinct species; but had the ones modeled by the exponential model not happened, the species would be indistinguishable.