Re: [gnugo-devel] gnugo 3.4 problems

[bump]

> As long as we're redoing ko handling, we might want to add another
> parameter of how many local ko threats one side has; I think between
> that and how many times the ko must be won, we can handle the majority
> of normal cases.

Two objections to taking into account the number of internal
ko threats.

How would you quantify this?

It's unclear (to me) that it's relevant to alpha/beta search
since one seldom thinks ``I'll play this way to maximize the
number of internal ko threats.''

And I think my suggestion was just a precise statement of the
number of times the ko must be one.

Dan