Re: poly: Solar system development

Robin Hanson, <hanson@econ.berkeley.edu>, writes:
> I don't follow this analysis at all. How can you say these *replicators*
> "can't any longer reproduce"? They are replicating! If there is some
> distribution of inter-oasis and within-oasis strategies, there is a
> selection effect where some of these strategies tend to appear more at
> the frontier. A challenge is to analyze this selection effect to
> figure out which colonization strategies will be selected for at the
> frontier.

I am talking about the case where it turns out that the fastest way to
spread is to spend little time in each oasis and quickly do a single,
massive launch outward. As you have described, this may not turn out to
be the best strategy, but supposing that it is, you get an interesting
result.

In that case, once a probe has reached an oasis, quickly reproduced itself,
and sent its children onward, its further behavior has no effect (by
assumption) on the success of its children's expansion.

In effect the replicators "can't reproduce" from the point of view of the
frontier, because they can't reach the frontier any more. Any aspects
of their "genome" which are only relevant to their post-launch,
post-reproduction behavior are not selected for.

As the frontier sweeps through a new region of space, the first
replicators to arrive will be those which have achieved maximum expansion
speed. Their behavior after launch has occured has had no effect in
the past on the survival and success of their outward-moving offspring.

At the frontier, the behavior of replicators will be determined solely
by the strategies which lead to maximum speed of expansion. If it makes
no difference (from the point of view of speed of expansion) what the
replicators do after launching, then frontier replicators will face no
selection pressure on post-launch behavior.

We should therefore expect that behavior to gradually evolve to be random.

This is paradoxical because it would seem that post-launch behavior would
be very relevant to the overall success of the replicator. One which
launches quickly and then holds its territory would become more numerous
than one which launches and then dies. However at the outward-moving
frontier, the first has created no more children than the second, hence
frontier replicators will be no more likely to behave the first way than
the second way.

Hal
Received on Mon Dec 15 19:52:10 1997