Author: Peter Kappler
Date: 15:41:32 06/26/05
Go up one level in this thread
On June 26, 2005 at 18:16:05, Uri Blass wrote: >On June 26, 2005 at 15:40:32, Peter Kappler wrote: > >>On June 26, 2005 at 14:49:03, Sandro Necchi wrote: >> >>>On June 26, 2005 at 12:40:38, Peter Kappler wrote: >>> >>>> >>>>>> And I do think you need to give more than one example to support your claim. >>>>>>If it was truly a 150-point bug, it should be easy to find several positions. >>>>> >>>>>By mistake part of the program was deleted. This bug could not be seen at blitz >>>>>level and was effecting the game only at high depths. >>>>>It was related to pieces exchange handling info, this is why it was not able to >>>>>see deeper on tactical combination if involving pieces exchange and was giving >>>>>away pawns without compensation in some positions. >>>>> >>>>>> >>>>>>It is strange that Stefan makes no mention of such a serious bug in his >>>>>>post-match report. >>>>> >>>>>Well, he was not pround to discover that, but you can ask him. >>>>> >>>>>I found this problem after the 2004 WCCC and he did not believed me as all he >>>>>did to the program was improving something. Yes, but by mistake he removed part >>>>>of the program. >>>>>So, the same version got 2nd at the 2004 WCCC and I really don't know how that >>>>>was possible...I guess a lot of luck. >>>>> >>>> >>>>Or maybe the bug wasn't quite as harmful as you think? What you describe above >>>>sounds a like a problem with a recapture extension, not something that would >>>>reduce playing strength by 150 ELO. >>> >>>Well, it the program performs about 150 points elo less in matches I do not care >>>if it should not do this. The score are facts, what a bug can do or not it could >>>be a matter of opinions. >>> >>>I told you what Stefan told me when he found the problem. Previously I could >>>only see that it was playing a lot weaker in some positions, but could not >>>understand why. >>> >>>Sandro >>> >> >>I find it hard to believe that the best chess programmer in the world showed up >>at the 2004 WCCC with a bug so severe that it weakened his program by 150 ELO >>and he didn't even know about it. Are you telling me that he entered this >>version of Shredder in the WCCC without doing any testing? > >I do not think that SMK is the best chess programmer in the world. >The best chess programmer in the world is probably the programmer of fruit >fabian. > >I expect the next version of fruit to be stronger than shredder9. > >It will probably be no stronger than Shredder10 but the only reason is that >fruit is open source so SMK can learn from it when Fabien cannot learn from the >source of shredder. > >Uri Uri, I didn't intend for the "best programmer" comment to become a point of debate. :) I think we can both agree that SMK is one of the all-time greats in the field, and it's hard to believe that his testing methodology is so poor that he entered the 2004 WCCC without realizing he had a bug that had weakened his program by 150 ELO. You don't get to #1 on the SSDF list with that kind of sloppiness. -Peter
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.