[Computer-go] Are 4 'easy to avoid errors' common to all MC programs?

Darren Cook darren at dcook.org
Fri Jan 21 03:31:04 PST 2011


> 1    when behind (even slightly), Fuego appears to collapse and play
> rubbish, in my experience this is most easily perceived in the
> late/middle game.

Dynamic komi. (E.g. if it thinks it has only 30% chance of winning at
7.5pt komi, but if you reduce the komi to 5.5pts it thinks it has a 55%
chance of winning, then reduce the komi to 5.5pts: it will play an
intelligent looking endgame and lose by 1.5pts.)

This will make it weaker overall (because it won't try so hard to cause
trouble), but especially against humans I suspect it won't be
noticeable, and may even have the opposite effect.

I.e. program endgame is generally stronger than the humans of the same
rank; chances are a 1-dan human will make a few 1pt or 2pt errors during
the endgame.
Following on from the above example, as soon as they make one of these
the programs winning chance will jump from 55% to say 70% and so the
dynamic-komi program will try increasing komi from 5.5 to 7.5pt,
discover it is now winning by 52%, and go on to win the game. :-)

Nice theory anyway...

Darren

P.S. AFAIK Fuego does not have a dynamic komi option currently.



More information about the Computer-go mailing list