Re: Advice on evaluation

[John Clarke <johnc@xxxxxxxxx>]

David Fotland wrote:
>
> I'm thinking about evaluating differently based on who is moving next.
> For group strength today, I have a single value.  I'll have two values
> after this change:  one for white to move, and one for black to move.
> I can look at the difference between those values to see if a group is
> worth attacking or defending.

What you're describing is calculating the value of sente - humans
find that an important and useful concept - maybe your program will
too!

One thing puzzles me though:

You say it will save a ply of lookahead, but surely you'll have to try
each possible move to evaluate the group strength, so isn't the
lookahead
sneaking in the back door so to speak?

E.g. in your example how would each side know that the connection is
crucial?  The connection might be miai with an eye-making move inside
the surrounded group - that happens all the time in ordinary games.

Maybe you should run your existing code twice - once for black to play
first and once for white to play first.  Of course that's a slowdown
by 0.5 but it might be worth it, and it should be much easier to try
than the major hacking you mentioned.

Or maybe you have a superclever trick up your sleeve?

If you do try this let please us know how it pans out since I've always
thought that a double evaluation (black/white play first) is a good
idea.

Regards
John