Re: computer-go: FPGA

[Joan Pons Semelis <joan@xxxxxxxxxxxxxxxx>]

> From: "Mark Boon" <tesuji@xxxxxxxxxxxxxxxxx>

[ lots of interesting thoughs snipped, I agree with all ]

> For the next step, the focus can then be on computer-generated patterns. One
> could think of making a program that uses a genetic algorithm to produce
> patterns. If a newly generated pattern improves the speed with which it
> finds the solution to some known problem it will be allowed to proliferate.
> For the time being, these patterns can still be human-generated, as long as
> the pattern and their function are properly separated. If pattern-learning
> technigues then later become more feasible they can be used instead. The
> power of this idea must be obvious, as the level of the computer can
> increase this way even without having increasingly more computer-power. Just
> let the program run at night, breeding patterns. No, correct that, let all
> the idle computers in the world run at night...
> 
> Mark

There're also idle computers during day-ligth ;-)

I think the bottle neck is still the evaluation function: it's very
difficult to get a reasonable selection after a look-ahead if it's
not good enough, and it's difficult to know what branches to prune
if the program doesn't know/understand (for some meaning of these
words) what's happening on the board.
My program (The Turtle, andd it's previous incarnations) has always
played better with deeper and wider look-ahead, but I tried some
quiescence, null-move and a variety of prunning methods that should
in theory make it player faster without loss of level and the results
where bad.
Aside from reading ladders (well, something more maybe), look-ahead
will not improve very much till a good evaluation function is there,
or, as in my case, it'll improve beacause the program is *so* weak.

Joan