Author: Bob Durrett
Date: 18:18:38 01/08/04
Go up one level in this thread
On January 08, 2004 at 21:14:23, stuart taylor wrote: >On January 08, 2004 at 18:34:40, Stephen Ham wrote: > >>On January 08, 2004 at 17:16:17, Bob Durrett wrote: >> >>>HIARCS X plays as “sinferno” at ICC on an AMD Athlon XP 2400+ @ 2.5GHz computer. >>>It holds the "official" ICC Blitz Record at 3604 and has the fewest losses and >>>better overall records versus strong titled players than any other computer on >>>ICC. It has only lost five games to two titled players but has played a large >>>number of GMs, IMs and FMs. It is currently the top engine at ICC using >>>Standard time limits. >>> >>>Why is that? Why is it better than all the other chess engines no matter what >>>the hardware? Is SSDF missing something here? Is there a disconnect here? >>> >>>Bob D. >> >>Hi Bob, >> >>That's a great question, to which I'd like to learn the answer too. I don't know >>anything about the ICC; I've never been there. But a few nights ago I thought >>that it might be fun to have a 5' match (10-games) against Shredder 7.04. >> >>Although I'm bored stiff by 5' computer games, I saw value in this one since I'd >>completely overhauled Shredder's opening book to only play the lines that I play >>in Correspondence Chess. Ordinarily, at slow time controls, Shredder virtually >>ALWAYS wins. So I naturally feel smug about my opening book. ;-) Therefore I >>thought it would be fun to finally see what my book and Shredder would do to >>Hiarcs 9 at 5'. After all, Shredder 8 defeated the latest Fritz at Blitz in >>Graz. >> >>To my shock and horror, Hiarcs 9 won convincingly (I don't recall the score but >>it was probably +3). So I examined the games and saw that Hiarcs and it's goofy >>opening book actually played superior tactics than Shredder. So I repeated the >>experiment. This time the score was something like +2, favoring Hiarcs 9. Same >>story on game quality. So I prepared Shredder's opening book for some of the >>offbeat stuff that Hiarcs plays (Hiarcs repeats the same lines it has success >>with a great deal), and the next result over another 10-game match was 5-5. >> >>Conclusion: Hiarcs' 5' tactical skills seem superior to Shredder's. Also, while >>Hiarcs' opening book seems quirky (and often downright bad), Shredder seems to >>need more time to refute the stuff that Hiarcs plays. In my latest long >>time-control match and tournament, Shredder was dominant. >> >>If I get time, I'll try a long time-control match and switch opening books. I >>have a theory that Hiarcs 9 would perform better if it weren't handicapped by >>its opening book. Oddly enough, this quirky book just may be an asset at short >>time controls in that it often removes the other engine from its book, saving >>Hiarcs time on the clock. Clearly this isn't much of a factor at longer >>time-controls where quality of moves played prevails. >> >>Well, that's my 2-cents worth. >> >>All the best, >> >>Stephen > >Might it be possible that the next Hiarcs could be a breakthrough in computer >chess level, from what you can guess (based on your understanding)? I'm talking >about long timings too. >Or is it that Shredder etc. is the REAL monster when given time? (and Hiarcs >might just be optimized for blitz, but improves evaluation very slowly after >more time, by which time Shredder catches up and overtakes)? >S.Taylor HIARCS X has the highest rating at ICC at Standard Time controls. Bob D.
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.