[ExI] Group Selection Advances

Aware aware at awareresearch.com
Fri May 1 18:19:41 UTC 2009


The arguments about "group" or "multilevel" selection remind me of
similar silliness about the evolvability of modularity in genetic
programming.

* Yes, it's mathematically demonstrable that the alleles of
individuals within a group will not evolve in favor of fitness acting
at the level of the group.
* Yes, given static objectives, there is no increased fitness to
offset the added cost of functional modularity.

But, characteristically, I'd like to point out it's a matter of context.

* Groups do tend to split and then evolve independently, such that
their distinct genes do tend to eventually compete at the group level.
* Environments do tend to change, conferring increased fitness on
modular reuse of functional building blocks.

And overall, evolution isn't about genes or particular modules, but
about exploiting increasingly synergistic (extropic, multi-level)
solutions maximizing freedom with least action (at least within the
system.)

- Jef



More information about the extropy-chat mailing list