Author: James Robertson
Date: 16:01:32 06/19/99
Go up one level in this thread
On June 19, 1999 at 18:24:40, blass uri wrote: > >On June 19, 1999 at 17:33:49, James Robertson wrote: > >>On June 19, 1999 at 16:32:41, blass uri wrote: >> >>>Junior 4.5 out of 7 is better than cilkchess 5 out of 7 because Junior played >>>against better oponnents. >>> >>>Junior won cilkchess and played against >>>Shredder,Fritz,Darkthought,Nimzo,Hiarcs,Eugen(all of these program have at least >>>4.5 out of 7 except hiarcs and Eugen with 4 out of 7) >>> >>>Cilkchess was lucky to play against weak programs like patzer and sos in the >>>first 2 rounds and did not play against strong opponents except Junior,dark >>>thought and Hiarcs7. >>> >>> >>>The result is that the humans are going to have an easier job tommorow because >>>they are not going to play the best 4 computers. >>> >>>Uri >> >>Of all the people to be saying that we should make decisions without evidence of >>games, I would have expected you to be the last. :( >> >>We have 7 games of Cilkchess total and 10 of Deep Junior (including Gelfand >>games), and you have already decided that Deep Junior is better? You do not have >>much evidence for that. Perhaps Cilkchess really deserves to be champion and it >>was just bad luck it did not become champion. > >I know that we have not enough evidence but it is clear that deep Junior's >results are better than cilkchess results. We don't know that..... If by fluke all of Ferret's games looked like his debacle against Hiarcs (or even just 2 or 3 of them) we would be in error concluding it was very weak. 7 games is a minute sample. >If we repeat the same competition again and again and if we assume that the >results are going to be always the same then Junior's rating is going to be >better than cilkchess rating. That is if we *assume* that the results will always be the same. We cannot do that, as the sample is too small. > >If we should decide only by results then Junior deserve a better place then >cilkchess. 7 games is nothing as far as results. You yourself have mentioned before that you have seen two identical program play each other and one lose 8-2. James > >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.