Re: computer-go: FPGA

[Mike Gherrity <gherrity@xxxxxxxx>]

>> Dana Nau at U. Maryland showed there were games with this property.  The
>> deeper you search, the worse you play.  He called such games "pathological."

> I   don't  think this   result applies   in   practice  to  chess, go,
> tic-tac-toe or any  typical 2 player game we  are used to playing.

It doesn't apply to tic-tac-toe, and history suggests it doesn't apply to
chess either.  I think Go is the last of the perfect information games that
still has a chance against the brute-force, iterative deepening search
machine.  Since it is known that a pathological game would foil the search,
it is not unreasonable to suspect the last holdout is pathological.  Still,
I feel like I am routing for John Henry.

Some of the early work on AI algorithm development for chess was interesting.
That was when computers were too slow to do a brute-force, iterative deepening
search.  If Go succumbs to a brute-force, iterative deepening search, then
it would seem that only imperfect information games can be used for AI
algorithm development.  That will eliminate all "typical 2 player game(s) we
are used to playing."  Heavy sigh.

The problem is that people don't do a brute-force, iterative deepening
search.  How do they play so well?  It would be nice to figure out the
answer to this question using a game like Chess or Go.  However, if it
can't beat the brute-force, iterative deepening search machine, people
aren't interested.

So I find myself hoping Go is pathological.


	mike