Author: Robert Hyatt
Date: 13:54:46 11/08/02
Go up one level in this thread
On November 08, 2002 at 16:06:44, Ricardo Gibert wrote: >On November 08, 2002 at 15:30:20, James Swafford wrote: > >>On November 08, 2002 at 15:15:43, Ricardo Gibert wrote: >> >>>On November 08, 2002 at 14:41:34, James Swafford wrote: >>> >>>> >>>>Sorry if this is old news... I don't get much time to >>>>read the posts these days. >>>> >>>>Crafty and Sjeng played a 5 3 match the other day which >>>>ended in favor of Crafty. Have they played since? >>>>I'd like to know if GCP's claim that Sjeng is stronger >>>>is true. >>>> >>>>My guess is the engines are pretty close, but I'm not yet >>>>convinced Sjeng is stronger. >>>> >>>>-- >>>>James >>> >>>Last thing I heard sounded like a comedy of errors. GCP inadvertantly used the >>>"Tal" personality, which was not intended for comp-comp play. Sjeng fell behind >>>in the first series of games as a consequence. GCP corrected this and Sjeng held >>>its own thereafter. However, RH discovered he had inadvertantly used an >>>experimental version not ready for ICC play! Consequently, all those games >>>appeared to be meaningless. >> >>Right. I'd like to see GCP and Bob get "the best they got" and >>play a no excuses match. Winner gets bragging rights. I don't >>feel like anything has been settled yet, just two people both >>claiming their engine is stronger than the other's. Should be >>easy to settle. >> >>GCP has stated (at least) twice that I can see for myself when >>it's released, but I'd rather see a public match. They are much >>more interesting (and less time consuming on my part :) ). >> >>-- >>James >> >> >> >>> >>>I just now looked at Sjeng's history on ICC and got the following: >>> >>> Recent games of Sjeng: >>> >>> Opponent Type ECO End Date >>> 18: + 2828 W 2676 crafty [ br 5 3] C42 Mat Nov 07 02 16:14 >>> 17: + 2818 B 2686 crafty [ br 5 3] B26 Mat Nov 07 02 15:59 >>> 16: + 2807 W 2697 crafty [ br 5 3] C42 Mat Nov 07 02 15:32 >>> 15: + 2795 B 2709 crafty [ br 5 3] B61 Mat Nov 07 02 15:19 >>> 14: a 2782 B 2722 crafty [ br 5 3] A00 Sho Nov 07 02 15:17 >>> 13: + 2782 W 2722 crafty [ br 5 3] B01 Mat Nov 07 02 14:53 >>> 12: = 2767 B 2737 crafty [ br 5 3] D66 Rep Nov 07 02 14:40 >>> 11: + 2769 W 2735 crafty [ br 5 3] C42 Mat Nov 07 02 14:21 >>> 10: = 2753 B 2751 crafty [ br 5 3] D00 Rep Nov 07 02 13:57 >>> 9: + 2753 W 2751 crafty [ br 5 3] C42 Mat Nov 07 02 13:38 >>> 8: - 2735 W 2792 TombRaider [ br 3 3] B22 Res Nov 07 02 12:30 >>> 7: + 2750 B 2759 crafty [ br 5 3] B84 Mat Nov 07 02 06:46 >>> 6: = 2732 W 2777 crafty [ br 5 3] D45 Rep Nov 07 02 06:23 >>> 5: = 2730 B 2779 crafty [ br 5 3] B84 Agr Nov 07 02 06:08 >>> 4: = 2728 W 2781 crafty [ br 5 3] D46 Agr Nov 07 02 05:54 >>> 3: = 2725 B 2784 crafty [ br 5 3] D45 Agr Nov 07 02 05:40 >>> 2: - 2722 W 2787 crafty [ br 5 3] D46 Mat Nov 07 02 05:25 >>> 1: = 2736 B 2773 crafty [ br 5 3] B84 Agr Nov 07 02 05:12 >>> 0: + 2734 W 2775 crafty [ br 5 3] C96 Mat Nov 07 02 04:54 >>> 99: = 2714 B 2795 crafty [ br 5 3] B81 Rep Nov 07 02 04:44 >>> >>>This shows Crafty being handed its head in the most of the most recent games >>>near the top. Considering the errors that were made in conducting this match, I >>>would not draw any conclusions from this at all...except that running a match on >>>ICC appears to be quite challenging ;-) > >I looked at Crafty's and it appears to be a little bit more informative: > >Recent games of crafty: > > Opponent Type ECO End Date > 12: + 2557 W 2665 EA6PZ [ sr 15 3] B23 Res Nov 07 02 21:45 > 11: = 2534 B 2527 HangerOn [ sr 45 5] C44 Agr Nov 07 02 20:33 > 10: - 2670 B 2910 EA6PZ [ br 5 3] D11 Mat Nov 07 02 20:07 > 9: - 2677 W 2922 BountyHunter [ br 5 3] D29 Mat Nov 07 02 19:45 > 8: + 2684 W 2485 IgorIvanov [ br 5 3] D11 Res Nov 07 02 19:10 > 7: a 2676 B 1917 Hyatt [ bu 5 5] A43 Agr Nov 07 02 18:25 > 6: - 2676 B 2828 Sjeng [ br 5 3] C42 Mat Nov 07 02 16:14 > 5: - 2686 W 2818 Sjeng [ br 5 3] B26 Mat Nov 07 02 15:59 > 4: - 2697 B 2807 Sjeng [ br 5 3] C42 Mat Nov 07 02 15:32 > 3: - 2709 W 2795 Sjeng [ br 5 3] B61 Mat Nov 07 02 15:19 > 2: a 2722 B 1917 Hyatt [ bu 5 5] A00 Sho Nov 07 02 15:18 > 1: a 2722 B 1917 Hyatt [ bu 5 5] A00 Sho Nov 07 02 15:18 > 0: a 2722 W 2782 Sjeng [ br 5 3] A00 Sho Nov 07 02 15:17 > 99: a 2722 B 1917 Hyatt [ bu 5 5] C20 Agr Nov 07 02 15:17 > 98: - 2722 B 2782 Sjeng [ br 5 3] B01 Mat Nov 07 02 14:53 > 97: = 2737 W 2767 Sjeng [ br 5 3] D66 Rep Nov 07 02 14:40 > 96: - 2735 B 2769 Sjeng [ br 5 3] C42 Mat Nov 07 02 14:21 > 95: = 2751 W 2753 Sjeng [ br 5 3] D00 Rep Nov 07 02 13:57 > 94: - 2751 B 2753 Sjeng [ br 5 3] C42 Mat Nov 07 02 13:38 > 93: - 2534 W 2658 Yace [ sr 60 30] C10 Mat Nov 07 02 11:04 > >If you'll note, RH appeared to make some type of adjustment to crafty which he >tested in games #99 to #2. After that, Crafty lost 4 straight. The first few games were played on the quad 400. I left for class and Crafty reverted back to the quad 700 for most of the remainder of that first set of games that GCP reported on here. That night I worked myself on the quad 400 trying to integrate a lot of the "personality stuff" Mike Byrne sent to me. I got some of it done, but had some real quirks that I didn't feel like fixing late that night. The next day, GCP asked if I could switch to the quad-400, and I did. Without remembering my "in progress" debugging. Crafty played horribly and lost games against everybody. I finally realized what had happened when I got out of class and got home, and asked my sysadmin to "pull the cable" as I couldn't shut crafty down remotely. That's where things ended. It doesn't take much to see that something was _seriously_ wrong with the last games in Crafty's history against everybody... The problem related to badly screwed up "piece values"... > >I suspect the adjustment contained yet another error in setting Crafty up, since >I don't think Sjeng is really that much better (if at all) than Crafty. Then >rather than make yet another excuse, which would have come off sounding lame, RH >simply threw in the towel and discontinued the match. The more honorable course >of action. I wouldn't be surprised if RH never managed to use a reasonable >version of Crafty in this match. This seems the most plausible explanation to >me. I don't think Sjeng is that much better. > >If you're a Sun Tzu fan, you might speculate that RH "craftily" played the >entire match with an inferior version of Crafty on purpose to project apparent >inferiority when superiority is the actual case ;-)
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.