| The idea of this test is to get a most reasonable overview,
if and how far the results of the following results and rankings differ
between a) Each programm gets its own PC and plays with Permanent Brain versus another program which also plays on an own PC with Permanent Brain. b) Like a), now both programs play without Permanent Brain c) Both programs play on one PC against each other, both programs without Permanent Brain. Variation a) can be found on all local and official tournaments (Paderborn, Leiden, Aufseß, Thüringen, WCh etc.), b) is performed very rarely, because if someone has 2 PCs it does not make much sense to not use the strongest settings, and c) is the most popular variation, because most of the computer chess enthusiasts only have one PC, or if they have 2 PCs, their speed difference is too large. And the protectionists of variation c) strongly hope that these results have universal validity and do not (or not significantly) differ from variation a), which would reduce testing time and resources to a minimum. In opposite to that, in my humble opinion, stand the real results of the local and occicial tournaments, which only rarely produce(d) similar results to c). However, there is some more or less justified criticism: - The number of games played in c) is much higher and therefore has a higher statistical relevance. - In the local and official tournaments only special versions (partly beta-versions or special engine settings) and/or special opening books are used. This may lead to different results than the normal, commercial program would reach. Contra-argument is that the number of games might be larger, the level usual is not comparable to a). In a) almost exclusively 40 moves in 120 minutes plus rest of the game in 60 minutes is used. In c) the preferred levels are game in 3, 5, 10, 15 (also called active or rapid chess), 30 minutes (also called active or rapid chess) and only seldom game in 60 minutes or even slower time controls. For these reasons, different hardware and different time controls, I think it is wrong to draw conclusions from results in c) to results in a) although this is common use. Up to now noone has done the work to compare the same programs under the same conditions in this setup. So again it is up to me to provide the results to prove me right or wrong. |