Anyone interested in helping with this Experiment?

[Norm Pollock]

I am looking for someone to help me run an experiment. The requirement is that you would have to run 3 variations of the arasan test suite (200 positions each) on a variety of engines of your choice, and report the results. The purpose is to see if there is any variation between moving as Black or as White on logically equivalent data suites. Although everyone assumes that color plays no part in engine analysis, I do not know of any experiment that verifies that. Unfortunately I have no experience with running test suites and collecting its data, so I need someone to assist me.

I chose to use 3 variations of the excellent arasan test suite by Jon Dart. This test suite has 200 test positions. The first version is the official version. It has 150 positions where White moves, and 50 positions were Black moves. The second variation has 200 positions where Black moves. The third variation has 200 positions where White moves.

To create the Black/White versions I used "epdFlip" from my "40H-EPD" package. It is freely available from the links at the bottom of this page. All 3 versions can be downloaded here:

Original version :
https://www.mediafire.com/file/l5wfo225 ... 4.epd/file

Black to move version (opcodes removed) :
https://www.mediafire.com/file/ad0s3t96 ... e.epd/file

White to move version (opcodes removed) :
https://www.mediafire.com/file/dpfa1c88 ... e.epd/file

I do not expect any differences in engine performance due to the color to move, but I think it should be checked.

[Norm Pollock]

I first had to put the suggested solutions (best moves) into the 2nd and 3rd files above since I had left them out. The corresponding links above have been updated to include the best moves.

I noticed that record #50 was an "am' (avoid move) record so I had to check that manually. I used Arena 3.5.1 (Engines Automatic Analysis) on an AMD Ryzen5 pc using 6 cores. I used 5 sec per move. I used the Feb 2, 2025 developmental version of Stockfish.

The results were as expected with no significant difference between White and Black. The scores were Black 174/200 and White 171/200.