Author: Ulrich Tuerke
Date: 00:35:59 04/30/99
Go up one level in this thread
On April 29, 1999 at 17:41:12, Roger D Davis wrote: >If the results of the SSDF are claimed to be OBJECTIVE, then what is wrong with >making all the games public? The fact that some of the games are not public is >bound to draw suspicion. And it should. What does the SSDF gain by keeping >certain games private? Nothing that I can think of. The SSDF is not a commercial >entity, right? Then why keep some of the games private? You should take into account that they have been testing for a lot of years now; in the beginning with chess computers only, which would not support automatic storage of game notations. Furthermore, the PGN standard wasn't (well) known those days. So, I simply guess that they only have a fraction of the games, nicely stored on disk in PGN format. May be, some SSDF member can comment on this ? > >Moreover, even if their motives are pure, rigor tends to slack over time. In my >experience, no one should be above quality control. Having an external check is >not an indictment, just a check. And it's in the best interest of the SSDF and >the broader chess community. If the integrity of the list is found to be sound, >then the reputation of the list is strengthened. If flaws are found, then they >can be corrected, and the reputation of the list is again strengthened. It's a >win-win situation. > >If enough of the games are public, it should be possible to compare the public >ratings to the private ratings in order to determine whether a statistical >difference exists. If no statistical difference exists, then the public-private >status of the games results in no systematic distortion in the rating process. >If it does, then we all know, and we can argue that the remainder of the games >be made public, or suggest other revisions. In principle, these arguments sound quite reasonable to me. If these game notations (of the remaining games) exist, it would be nice to have them too. Howvever, I am afraid that it will be extremely expensive to gather a statistical probe being sufficiently large to be compared with the SSDF results ? Regards, Uli > >Roger > > >On April 29, 1999 at 09:08:17, Ulrich Tuerke wrote: > >>On April 29, 1999 at 08:42:24, blass uri wrote: >> >>>Most of the games of the ssdf are not public and we have no way to check if they >>>really happened. >>> >> >>I cannot fully understand the motivation for this project. >>Obviously, you distrust the SSDF. If this is the case, then why you trust in the >>published games. These could be manipulated very well, though each of them being >>reproducable. E.g. they would just have to skip some games with unwanted >>results. >> >>Thus, I do not see much sense in your project. >> >>BTW - to avoid misunderstandings - it is not my opinion that the SSDF >>manipulates. I personally think, that they give their best to get reliable >>results. >> >>Regards, Uli >> >> >>>I agree to do part of the job and to find the results of part of the public ssdf >>>games but if we want to have all the results then we need more people to do it. >>> >>>I hoped that a program can do it but I understand that probably no program is >>>going to deal with it because the names of the programs in the pgn file of the >>>games are not exactly the same(for example sometimes Fritz5 is called Fritz 5 >>>and sometimes it is called to Fritz 5.0) . >>> >>>I hope that after we have all the results then it will be easy to compute rating >>>by a relevant program that get the names of the programs and the results. >>> >>>Uri
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.