Norris writes:
> >In my continuing bid to become the group's curmudgeon, let me say that
> >natural selection can quite easily select for extinction.
>
> I might argue a semantic point here. While you make a valid argument that
> past natural selection can lead to evolutionary dead ends and extinction, I
> don't think it is accurate to say natural selection is selecting "for"
> extinction itself. Rather, natural selection "for" certain traits (other
than
> extinction-proneness) may ultimately lead to extinction. A minor point
> perhaps, but an important distinction.
No, what you write is not a small point. When I wrote that "natural selection
can quite easily select for extinction," I was simply being sloppy in my
language. What I meant, and what I should have written was "natural selection
can
quite easily select for [the short-term advantageous conditions that
ultimately lead to the] extinction [of the lineage].
Interpreting literally what I wrote leads to a condition that I normally rail
against myself. Selection never selects "for" anything. Selection operates
only as a culling mechanism, removing the least appropriate, least competitive
phenotypes of the demic excess that currently fills the competitive arena.
If selection were only culling the black balls from an urn filled with red
and black balls, what would be the harm in saying that it was selecting "for"
the red balls? If the genetic representation of the individuals' code were that
independent, there wouldn't be any, but no such situation can exist in a
complexly interwoven informational system, especially one where the twin
phenomena
of polygeny and pleiotropy dominate.
I've previously written about this misuse of language as being one of the
fundamental philosophical errors that plagues evolutionary biology, so I'm more
than a little embarrassed that I wrote that line myself, but it's not what I
meant, and hopefully that's clear from the context of my other comments.
One paper that is on-line which contains my criticisms of such language is at:
http://aics-research.com/research/notes.html#IIIC
This paper is on the simulation of evolution for purposes of evolving machine
intelligence and was published in an engineering journal in 1994. Although
the idiom of the paper is primarily engineering, engineers designing extremely
complex systems face precisely the same problems that nature does in optimizing
its designs, and thus the subjects of accurately determining what is being
evolved, optimized and selected very rapidly converge.
If you get these qualities wrong, you're offered every opportunity to quite
completely misunderstand the evolutionary process, which is arguably a more
serious consequence for engineers than biologists.
Wirt Atmar
