Static Exchange Evaluation...

[hgm]

Is there a definite measurable performance gain in terms of CPU time for this? Probably the majority of SEE cases terminate very quickly too.

No, but it is much simpler. Why add code that is almost never used, and when it is, most likely gives a wrong result? Especially in Chinese Chess where you have Cannons that capture over other pieces doing a full SEE with X-rays is very complex.

[Don]

Is there a definite measurable performance gain in terms of CPU time for this? Probably the majority of SEE cases terminate very quickly too.

No, but it is much simpler. Why add code that is almost never used, and when it is, most likely gives a wrong result? Especially in Chinese Chess where you have Cannons that capture over other pieces doing a full SEE with X-rays is very complex.

It could be an enhancement if it's either faster, or if you use it as an uncertainty measurement. If you make a conservative assumption, you would prune less - if you make the liberal assumption (that the move is bad) you would prune more. I don't know how this idea works out - would you say you are being more conservative or more liberal - or is it hard to say?

Don