Author: Robert Hyatt
Date: 05:49:02 11/10/97
Go up one level in this thread
On November 10, 1997 at 05:26:34, Amir Ban wrote: >On November 09, 1997 at 16:48:08, Chris Whittington wrote: > >> >>On November 09, 1997 at 16:38:38, Thorsten Czub wrote: >> >>>>I have tried to explain thoses things to Mr Van den Herik before >>>>the beginning of the Tournament. But there was nothing to do, >>>>his decision was made long before. To justify accelerated pairing, >>>>he said that there was a real gap between the strongest programs and >>>>the weakest ones. After the tournament, many games prove it is not >>>>true (between the parenthesis, the final rankings of the programs): >>>>Stobor (24 ) won against Fritz 5 (16), Chess Tiger (27) made a draw >>>>against Dark >>>> Thought (6) , Chess Guru (14) won against Shredder (3), etc... >>>>During the tournament, I have heard Mr Marsland himself say to Bruce >>>>Moreland that the accelerated pairing was not necessary... >>>> Errare humanum est, perseverare diabolicum >>> > >We did not need the accelerated pairings. The problem is that it brings >the climax too early and creates garbage-time in the late rounds, when >you have difficulty in finding equal matches especially at the top and >bottom. In Paris it looked like it was happening, but it was fortunate >that Shredder came from behind in the second half and his games in the >9th and 10th rounds against the leaders were to my mind the climax of >the tournament. > > >>>CSTal was in the B group of programs that are not that strong ! >>>We did not play bad as a program from the low-level-group. >> >> >>It is very difficult to put programs into groups. the french programs >>were mostly in the lower group. van den Herik told me he didn't know >>much about the French programs, so that was why. >> >>I don't really care which group van den H puts the programs, why is it a >>problem ? >> > >It's not a real problem. A high rank gives you a slight advantage in >pairings, but even from the 34th position you have no problem in winning >the tournament. > > >>More problematical is the desire of certain programmers to ban 'weak' >>programs. Only if you are a top-ish program should you be able to enter, >>they want restrictions to 24 programs; and make sarky comments about the >>weakest program only allowed for political reasons usw. usw. >> >>Need one point out that Junior was a 'weak' program until recently; > >?? When was that ? > > >>would Junior never allow to be entered ? >> > >There's really no problem in demonstrating your strength if you have it, >even if you are a new program. You have ICC, Aegon, Paderborn, local >championships. If everything else fails you can play a match at home >against some respected program and send the score and pgn's to the ICCA. >It's dubious, but it's something. > > >>Or that, often, programs are rated or rated badly by the cognoscenti for >>personal and/or political reasons ........... ? >> > >This is true. I thought some of the discussion in rgcc about the >possible outcome was particularly uninformed. Some favorites were >suggested for no better reason than that they scored well on >test-suites, which IMO shows a real misunderstanding of computer chess. > >I thought the ICCA did well to ignore all such amateur opinions, and >they produced a reasonable ranking, with some minor exceptions (I was >surprised to find CSTal ranked so low. Kallisto was probably overrated. >My own ranking (8th) I thought to be 2-3 places too low, but the general >public would have ranked me lower so it was a relief in a sense). > >The ICCA basis for the ranking was quite obvious. They gave much weight >(correctly IMO) to past ICCA events, disregraded the CPU issue, and >factored in things like SSDF and Aegon, the last one much too high. > >Amir if you run thru the math, 11 rounds actually works out nicely, because you can have perfectly *random* seeding, and still produce good results from top to bottom. If you have 34 players and only 6 rounds, which is barely enough to produce a clear 1st if there are no draws, seeding becomes critical as the tie-break would favor the top seeds, assuming the seeds are ordered correctly. Because a high-seed will play better opponents than a lower seed, purely based on proper pairing rules being followed. With 4 extra rounds, they could have listed the programs alphabetically and paired from that with no worse results than perfect seeding. We used to have real serious discussions at the ACMevents, because with 4 rounds (and later on 5 rounds) we *barely* had enough rounds without draws, and any draws by the top finishers would need a tie-break. IE we won the 1986 WCCC on tie-break after 5 rounds, we tied for first at the ACM 1982 event but ended up second on tie-break. The top-seed in 1982 won, we were the 2nd or 3rd seed. In 1986 we were probably seeded 1 or 2 with Hitech the other top-seed, but we beat them in round 5 so that wasn't an issue there.
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.