Nice to see something different for a change. My program RomiChess has zero endgame knowledge and I am not even close to the point where it can be added. There is so much to add first, like 'material based' draw detection and a few dozen other things. Still it would be interesting to compare how well such an engine would do in your test. You can download RomiChess from the WBEC 'engine detail' pages. Or you can email me at mjsherwin@netzero_dot_net (o.n) and I will send the latest beta to you. If you are interested.
Mike
If you are on a sidewalk and the covid goes beep beep
Just step aside or you might have a bit of heat
Covid covid runs through the town all day
Can the people ever change their ways
Sherwin the covid's after you
Sherwin if it catches you you're through
Hi Walter,
Thanks for your interesting work. I have a few questions about your test. Are all 100 positions solved by at least one program?
If so, what is the smallest set of programs (including which programs) that solves all 100?
If not, how many positions are solved by at least one program, and what is the smallest set of programs (including which programs) that solves them?
Which four programs, when their results are combined, solve the maximum number of positions and which programs are they?
Or if answering would be too much of a pain, perhaps you have a download link to an xls file from which answers to such questions could be extracted?
jdart wrote:I took a look at this position recently (#39):
[D] k2N4/1qpK1p2/1p6/1P4p1/1P4P1/8/8/8 w - -
I think this is less a test of endgame knowledge than a null move test - if your program allows null move with reduced material, it will fail this.
--Jon
well, I think Walter is right in adding such a position because zugzwang IS endgame knowledge. E.g. Quark has a very shallow verification search which doesn't cost too much and uses it only in endgames. So it took some time (found at ply 24) to find the solution. Anyway it CAN solve that position. I always have Dieter Buerrsners reasoning for implementing verfication search in mind: "I simply do not like that there are positions that the engine can't solve even when it gets endless time"...
And when I remember correctly several half & full points were lost on different tournaments just because of such nullmove problems. (I have Rybka & Fritz to name some here in mind)
Quarks verification search is maybe 2%... so you can't count it in elo. Well here and there half a point is maybe also countless in elos - but you sleep better...
