Re: poly: No expected value colonization equilibrium?

From: Robin Hanson <>
Date: Tue Dec 16 1997 - 10:40:24 PST

>... If probes fly in single file, far enough apart, an impact will
>only destroy the lead probe. In this model, attrition is linear.

A nice idea, though its not linear after you're down to one last probe.

>It may be possible for probes to tell which of the oases ahead of them have
>already been colonized, by observing signals sent out either accidentally
>or deliberately by the colonizers. ...
>By communicating among themselves, a fleet of probes may be able to monitor
>their own attrition. ...

I've been thinking about these possibilities, and I think we can model them
with a simple Q(P) function, where P(r,t) is the probability a random oasis
nearby is occupied, and Q is the probability a probe faces of trying to
colonize an already taken oasis. As long as there's at least some chance
of a failure to effectively communicate and coordinate, Q(P) > 0 for P > 0.

If you worry about having to share an oasis, I think its probably enough to
just send a signal that credibly shows that it comes from a stationary
This indicates a probe has stopped and is buzily growing there.
But when you feel confident you can destroy an incoming probe, you'd want
to keep quiet to induce your competitors to waste their probes. Whether you
announce upon landing depends on how long it takes you to set up defenses.

Probes may split into cooperating groups, telling others within the group
of attrition, landing intentions and landing outcomes, and maybe not telling
other groups.

Robin Hanson
RWJF Health Policy Scholar, Sch. of Public Health 510-643-1884
140 Warren Hall, UC Berkeley, CA 94720-7360 FAX: 510-643-8614
Received on Tue Dec 16 18:34:00 1997

This archive was generated by hypermail 2.1.8 : Tue Mar 07 2006 - 14:45:29 PST