Re: Ethics (was FAI (aka 'Reality Hacking'))

From: Eliezer S. Yudkowsky (
Date: Mon Jan 31 2005 - 12:48:14 MST

Phil Goetz wrote:
> B) See Stephen Gould's book /Full House/. The
> entire book is a debunking of the idea that
> evolution causes systems to become more complex
> over time. Gould claims, with much data to back
> him up, that evolution is a random walk, and that
> there is no tendency for species to increase in
> complexity over time. The distribution of species
> complexity may grow wider over time, but that is
> only because there is a limiting "left-wall"
> minimal level of complexity below which a species
> cannot go. This makes species complexity a
> one-dimensional random walk with a left boundary.

Stephen Gould's reputation is trash among serious evolutionary biologists,
not always because Gould gets it wrong, but because Gould claims, as his
own brilliant revolutionary ideas, warped and dumbed-down versions of quite
ordinary tenets of modern evolutionary biology. (Tooby and Cosmides 1997; In this case, the point
made by Gould in his 1996 book _Full House_ is stolen from George
Williams's classic _Adaptation and Natural Selection_ (1966). Nor was it
Williams's original idea, and I can't look up whose it was because I don't
have my books right now. But in 1966, there actually *was* a widespread
Panglossian optimism to inveigh against. When Gould wrote his book in
1996, such ideas were no longer current among serious evolutionary
biologists; Williams had dispelled them long since.

It's true that there's no steady increase in species complexity, but the
reason given above is not correct. Rather, natural selection tends to
increase genetic complexity, as genomes accrete additional adaptive
functions, *until* natural selection hits the right boundary, which is the
limited selection pressure available to maintain genetic information
against the noise of mutation. See Williams 1966 for more details on this.

If Gould had written his books in the 1940s, they might have done some
good. For Gould to write such books in the 1990s and present himself as
the hero of past revolutions is simply intellectual dishonesty, as Tooby
and Cosmides and others observe.

Eliezer S. Yudkowsky                
Research Fellow, Singularity Institute for Artificial Intelligence

This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:50 MDT