Author: Thom Perry
Date: 10:04:48 12/31/98
Go up one level in this thread
On December 31, 1998 at 11:44:53, Lanny DiBartolomeo wrote: >On December 31, 1998 at 07:48:33, Thom Perry wrote: > >>On December 30, 1998 at 16:40:15, Robert Hyatt wrote: >> >>>On December 30, 1998 at 04:18:11, Reynolds Takata wrote: >>> >>>>On December 30, 1998 at 03:45:52, blass uri wrote: >>>> >>>>> >>>>>On December 30, 1998 at 03:04:51, Reynolds Takata wrote: >>>>> >>>>>>What i was reffering to is pure chess strength, if you want to say pure chess >>>>>>strength of a program against kasparov specifically well that would be perfectly >>>>>>satisfactory with me :). >>>>> >>>>>It is not clear to me what is pure chess strength. >>>>>The strength of a program is different in different time control or in different >>>>>openings. >>>>> >>>>>Uri >>>> >>>>Blass do you want to try to answer this question or just keep trying to come up >>>>with a question yourself? >>> >>> >>>What he is saying is that you are searching for the 'holy grail' of computer >>>chess, and it probably doesn't exist in the form you desire. Programs are all >>>different. They play differently at different time controls, in different >>>openings, and in different types of games/positions. You can ask a dozen GM >>>players on ICC which program gives _them_ the most trouble in blitz, and get >>>a dozen different answers, some surprising. For bullet you will get probably >>>a different answer. And for those that play longer games you will get still >>>different answers. And if you look at the "tactical" IM/GM players you will >>>get a different answer than you will from the "quiet/positional" GM players. >>> >>>So *any* program could be the right answer to your question. Or the wrong >>>one... >> >>Extremely well said, Robert, and what you are saying is absolutely correct, I am >>sure. The reason for so many arguments on this board is the "My program is the >>'holy grail' of chess" mentality that prevails when someone dares to suggest >>that their pet program is not flawless. Notice the rash of messages questioning >>the testers whenever a new SSDF rating list is issued: "Gee, are you sure you >>tested my program correctly? Duh, it isn't number one on the list." > >I don't think he is asking which program is the strongest I believe he was >asking which program Garry Kasparov thought was the strongest based on different >informations that people heard. Regardless, I totally agreed with Robert's assessment regarding the current state of chess programs. His "Holy Grail" theory explains a lot of the problems on this board. Look at the current SSDF list and the difference between #1 and #5 on the list. According to statistical theory, Fritz 5.0's real rating could be as low as 2542, whereas Hiarc 6.0's real rating could be as high as 2549. Would anyone that really knows statistics want to really argue which program is actually higher rated as measured by the current SSDF test procedures?
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.