Author: Dann Corbit
Date: 02:42:26 09/02/00
Go up one level in this thread
On September 02, 2000 at 05:09:45, Carlo Meier wrote: >There is the new "GS-2930-Test" for some time, in order to determine with very >strong chess programs (over 2500 ELO/SSDF) an exact ELO strength number. 14 >positions select by the program are to be calculated during 20 min. on the >analysis level, whereby "only the time counts, after that the solution course at >the display appear and also up to 20. minute is maintained". (a non-solve >position becomes 20 min. term in prison). The individual solve and terms in >prison are then added to a total time (TT) in minutes, from itself with the help >of the formula: GS = 2930-(TTx1,5) the ELO number results in. > >My questions: Is this test really reliable, or is worthwhile themselves the >immense expenditure of time (the monitor, since some programs display the >solution course, then again to reject, after some time again displays... etc. >must be finally observed continuously.)? >Who already checked the 14 positions on its computer or with its programs? The test is reliable to see which programs can get the same answers as the test problems say. Like all other test sets, that's what it's good for. I don't think there are enough positions in the test to make an accurate assessment of a program's tactical strength. But it may provide interesting data. Certainly, you would not want to try those silly extrapolations which provide an ELO number from the test. They are absurd.
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.