Author: Mike S.
Date: 02:52:54 02/29/04
Go up one level in this thread
On February 29, 2004 at 04:17:05, Jouni Uski wrote: >I think we can safely remove Ruffian from PRO engines, because there is now 4 >(!) versions after free 1.0.5, but it's unclear if ANY is better. This is not >professional product! Haven't the latest updates been released just a few days ago? Isn't it a bit early for "final conclusions?" :-) For the SSDF ratings, Ruffian 2.0.0 has so far been matched against 4 different programs only, 3 of which were running on K6-2/450 or even P200 only (F5.32). The Elo system underestimates the chances of the weaker player to draw (as statistics experts have explained). This means, a program will always be underestimated when matched against opposition which is much weaker (in average). Compare the opponents 1.0.0 had to play against. Also, SSDF is the only rating list I'm currently aware of, where 2.x versions are clearly behind 1.x versions (actually it isn't so clearly when you consider the statistical margins of error). Another one is St.Pohl's FRC ranking list - but that isn't classical chess - where 2.0.0 is (only) 10 points behind 1.0.1. http://www.beepworld.de/members53/frc-list/ewig.htm Some other ratings: CSS ratings: 2.0.0 56 Elo points better than 1.0.1 http://www.computerschach.de/rangliste/ewige%20Rangliste.htm BfF ratings: 2.0.0 14 Elo points better than 1.0.5 http://www.beepworld.de/members39/computerschach2/bff-liste.htm Blitz ratings (by N.Mielke): 2.1.0 84 Elo points better than 1.0.x http://www.schachseiten.de/rangliste.htm These are from short to medium time controls. I think 40/2h isn't representative for average practise among computerchess users anyway. I'm sure the typical time per position engines are used, games or analysis, is much less than 3 minutes average. I don't claim nobody would use 40/2h with chess programs, but I think it's not the majority of cc fans & users. (For corresspondence players, 3 min. are not typical either :-)) The only real problem we have here IMO, is a psychological one: Too great expectations. It is very unlikely that a new chess engine who has just joined the commercial market, can immediatly compete with the very top for a "podest rank" on the same level (IIRC Tiger being the only glorious exception so far). Any top-10 engine is a masterpiece (if not any top-20 etc.), so I see absolutely no reason to panic... Regards, M.Scheidl
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.