comparing between search or evaluation

[Uri Blass]

I say that A has better search than B if the following happen:

1)You start from A
2)You modify the evaluation of A to give the same mathematical function as the evaluation of B and get a program that you call A-Beval.

Note that you can use some monotonic function like evaluation of B is twice the evaluation of A and I define it to be the same evaluation.

3)if A-Beval scale better than B then I say that A has better search than B.

Note that the speed of the implementation of the evaluation of B in the data structure of A is not important and you can start with time handicap that you get result near 50% and see that A-Beval score more than 50% if you increase the time control.

4)if in the same conditions A-Beval scales better than A then I say that B has better evaluation than A.

Note that I do not get conclusion if the opposite of 3 and 4 happen because it is possible that every program is going to scale worse by modification because the search is not tuned for the evaluation.

Now the question if is people are interested in trying to compare between search and evaluation of chess programs.

I guess that it is possible to show by this definition that stockfish has better search than other chess programs with free source code.

What is your opinion?

[Cardoso]

That would be a very interesting experiment I would like to be made.
But it requires time, effort and stubbornness.
Unfortunately I only program checkers, and also have little time.
This experiment can broaden our horizons on the relative merits between search and evaluation.
Mabe the stockfish comunity can help, I certainly wish they could do so.