Null move in IvanHoe

[BB+]

I did the following experiment. Turn off null move in IvanHoe via the flags in material_value.c (I think the other place is when there are promoted pieces in evaluation.c), in InitFlags put FLAGS=0; right after the #if stuff. I played about 5000 games at 15s + 0.25s increment (1 cpu, 2.8Ghz Phenom). Result was a bit over 45 ELO worse, and the version with null move was about 1.5 ply deeper on average.

[Robert Houdart]

Thanks, interesting result.
It's a good complement to the "zugzwang" discussion we had recently in this topic.

Robert

[benstoker]

Thanks, interesting result.
It's a good complement to the "zugzwang" discussion we had recently in this topic.

Robert

Speaking of which ... what could be done with ivanhoe to improve its zugzwang detection? Is one solution simply to retard null move after reaching a condition of minimal pieces on the board?