[extropy-chat] three coins in a fountain, part 2: the bayesian angle

Russell Wallace russell.wallace at gmail.com
Thu Oct 20 21:05:58 UTC 2005


On 10/20/05, spike <spike66 at comcast.net> wrote:
>
> Have you any suggestions for an overall strategy
> on such a test? Is there some systematic way for
> expressing your epsilon probabilities? You don't
> want to lose to the monkey, that would be
> embarrassing. Eliezer, how now?


This is analogous to one of those blind auctions, where you want to do
better than the other contestants, and have to guess what strategy they'll
use.

I'm not in the mood for crunching numbers right now since I've spent all day
crunching program code (so while I think the coin answer is 1/3, I'll let
someone else plug the numbers into Bayes' formula to check if I'm right)
but...

One element of strategy for this is to go a little higher than natural break
points. Don't bid $100, bid $101. Or if you think other people are wise to
that trick, bid $102... it depends on who you're competing against. (Also on
your utility function - is it more important to try and come first, or avoid
the risk of coming last?)

It also depends on how good you think the other guys are. If they're all
hotshot Bayesians with IQs of 160 and no social life so they'll be spending
all evening triple checking their answers, you'd better go for broke and
take your chances. If you're up against a class of monkeys, you can afford
to give yourself more margin.

I'd probably run this as an evolutionary simulation if I wanted some more
definite answers.

- Russell
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.extropy.org/pipermail/extropy-chat/attachments/20051020/7a50f86d/attachment.html>


More information about the extropy-chat mailing list