Other than ELO..what else should CC engines be striving for?

gabeeg · Post by **gabeeg** » Wed Oct 26, 2011 4:22 pm

I am becoming less and less interested in the ELO ratings of a Chess engines. These days I am more interested in an engine that can play a more "Human" game, one that has enhanced positional play, will make sacrifices for a better position and not just for the immediate win, quick and accurate evaluation, etc. Of course these enhancements can and do improve ELO also. My question for the forum is: What else do you guys think a chess engine can do outside of striving for the top ELO to make it an interesting, helpful and enjoyable program? Is ELO strength the biggest selling point of an engine for you? Is ELO strength the most important aspect of an engine for your game analysis these days?

IA · Post by IA » Wed Oct 26, 2011 5:52 pm

Hi gabeeg

Undoubtedly human playing style (Hiarcs, Shredder, Deep Junior ....) and versatility of a chess engine (The King, Hiarcs ....), in short, an engine of human service as a learning tool, Hiarcs complies well with the characteristics of modern engines on all platforms ( PC, Palm, Mac, Pocket Pc, iPad.....)

Regards.

gabeeg · Post by **gabeeg** » Wed Oct 26, 2011 9:22 pm

What actually is "Human Style"...I mean I totally agree, I too want a more human playing style....and I know it when I see it when playing against engines...but what is it? When playing Humans the style is incredibly varied...is that it variation?

IA · Post by IA » Wed Oct 26, 2011 9:41 pm

Hi gabeeg

Programs with a humanized game play behave quite surprising in many cases, much happens to me when I play against Palm Hiarcs where I think I'm facing a human due to unforeseen reactions of the machine, the style of game programs so you can switch to play Solid, active .... As an example I remember one of the tournaments in 2008 the program Mercosur Pocket Fritz 3 ( Hiarcs 12 ) taught continuous improvement and showed a striking positional understanding that really impressed the human Master, besides these new self-learning programs are not to repeat the items because the engines learn from a bitter defeat ...

Regard.

RadioSmall · Post by **RadioSmall** » Wed Oct 26, 2011 11:20 pm

gabeeg wrote:What actually is "Human Style"...I mean I totally agree, I too want a more human playing style....and I know it when I see it when playing against engines...but what is it? When playing Humans the style is incredibly varied...is that it variation?

Although the Human style is incredibly varied , in a chess program "Human Style" means a good positional understanding of the game. Note that this Positional Understanding does not necessarily help in Comp vs Comp games , and hence the Strongest Computer program on say the SSDF list is not also the strongest against humans . For a style that is really "Human" , one needs to look at the Programs Chessmaster , M Chess , and Gandalf.Humans generally do not outcalculate one another when playing chess , but win more on positional aspects ( pawn structure , endgame knowledge , etc) . For example the way a strong human loses to Fritz is different than the way he or she loses to a positional based program. Anybody who has played against a program that does not have a human style can recognize certain patterns immediately : a very dry mechanical style , moves that seem to come out of the blue , and extreme aggression in certain positions....