Re: [gnugo-devel] 17 stone game

[SP LEE]

> I tried a 17 stone game with recent CVS, and won by 90 points.
>
> I did this mostly to find out which strategies I would try, and the
> non-surprising answer is: Create semeais! As soon as they are complicated
> enough that the tactical reading code cannot understand them, gnugo
> makes far too many huge mistakes.
>
> I think owl and semeai are _the_ areas for improvement after 3.4.
>
> Arend

The strategy of gnugo to improve is obviously to maintain a good escape 
route to avoid (too complicated) semeais. I have also watched many games 
gnugo played before, where gnugo could avoid semeai easily. Similarly 
narrowing opponent's escape routes is same important strategy.

SP Lee

_________________________________________________________________
The new MSN 8: smart spam protection and 2 months FREE*  
http://join.msn.com/?page=features/junkmail