Don Dailey wrote:

Is there a problem with viewing each playout
> as 2 games,  either 2 wins, 2 losses or a win
> and a loss?

Yes. This is equivalent to the draw = 1/2*win
statement made previously, but the 1/2 constant
is just a magic constant. The problem is bicriterion
an draw = lambda*win can be justified for all lambda
in [0,1] as the admissible region.

In first round, against a 300 Elo points stronger
program, draw = win (i.e. lambda = 1) is acceptable
for the weak program, but only draw = loss (lambda = 0)
is acceptable for the stronger.

In the final round of a tournament where you only
need one point to win (or move forward to the next
round) draw = win is perfect. If you need two points
the draw = loss is optimal even against a stronger
opponent.

One day this may become an issue in 9x9 (not in a
foreseeable future in 19x19). I guess when this
happens playing 9x9 won't make much sense anymore.

Jacques.






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

Reply via email to