On anti-MCTS bot strategy:

I don’t know of a strategy, but there sure are principles.
I can state one as a proverb:

"We you clarify, you are helping the bot."

E.g., If a connection works but is not obvious, if a semeai
can be won but is not obvious, etc. the bot has to discover
it for each visited tree path and, until it does, it is reading
wrong results and merging them with correct results. When a 40
point group should live 100% of the time and lives only 80%,
its weight is as if it was (around) 32 points. That is systematic
bias.

And for handicap games:

"First catch-up, then clarify."

Most superior humans that lost handicap games to bots did
the opposite: When they feel they have catched up enough, from
say -60 points to -10 they clarify thinking they will get the
other 10 points easily in yose. That surely works with their
students, but doesn’t work with MCTS. The bot is extremely
strong at counting and if you clarify the position it is very
hard to convert a losing game into a winning game. If you do
it properly, leave the bot in its blurred misevaluation status
until you have catched up, when you clarify the bot will resign.

Of course, this is easier said than done. Applying this requires
a lot of skill, but not applying it may be the reason why stronger
players lose their first games.

Jacques.


_______________________________________________
Computer-go mailing list
[email protected]
http://dvandva.org/cgi-bin/mailman/listinfo/computer-go

Reply via email to