From:Martin Mueller <[email protected]>
Replying to my question:
> How close do such estimates correlate with the estimate of strong players?
> When
>
> they differ significantly, is it possible to say whether the program or the
> players are more correct?
Sufficiently strong players are always correct :)
Fuego, and I suspect other MCTS programs as well, suffer from the fact that
random simulations often get the tactical details wrong. On the small board
that
is often hidden, because there are not many different fights, and the tree is
large enough to resolve many such tactics. On the big board it is often fatal -
you typically have five or ten settled local fights on the board by the late
middle game, but none of them may be 100% settled in the playouts, and a few of
them may be evaluated quite badly. If several get misevaluated in the same
direction, it adds up to a lot of bias. The effect is similar to thinking that
one of your groups is alive when in reality it is dead.
I've suspected as much; a systematic error in the status of a group (or groups)
can greatly bias the winrate signal; it looks to me like strong players exploit
this gap.
Strong players analyze local positions, decide which groups are alive or dead,
and leave them alone until something changes the analysis. If it is already
dead, no need to make it more dead. If it is alive already, no need to make it
more alive - but when an important liberty is removed, revisit the analysis and
act appropriately.
Programs already exist which can get the status right most of the time. Can
this
tactical analysis be incorporated into the playout engine? If so, it would
greatly improve the quality of the winrate estimation.
_______________________________________________
Computer-go mailing list
[email protected]
http://dvandva.org/cgi-bin/mailman/listinfo/computer-go
