Author: Mike S.
Date: 02:13:56 12/10/03
Go up one level in this thread
On December 09, 2003 at 14:30:28, Tord Romstad wrote: >>(...) No, because Kb7! gives the fastest mate. Every variant ends in 44. (6.) c7-c8#. >>Many engines prefer c4 at first, which ends in mate too - but later, so it's >>clearly second best only. >OK, but I still think this is not a very good test position. Many engines >(including mine) are "lazy" and start pruning more aggressively for the >winning side when a clear win has already been found. The idea is that it >is more important to make sure that there isn't a very deep refutation >of the apparently winning move than to look for an even more crushing win. This is ok! But it's for example not ok to stop analysing (in analysis mode) after the first variant forcing mate has been found. Some engine do this and stop the calculation after an mate announcement for c4, although they may be able to find Kb7 within a few seconds more. I see no reason for engines to stop analysis and not search for a faster mate, in analysis mode (I don't know what your engine does). I cannot say if pruning which misses 1.Kb7 with the fastest mate, could be seen as too aggressive. Nevertheless I think, the miss of a - relatively simple - # in 6 justifies nowadays, that engines gets no testing score then. (1.c4 as an solution would no make sense as a test position, because it's much too easys and the most obvious move.) >In general, I don't like tactical test positions with more than one clear >win. I like the #16 position a lot, because it combines several issues: Build a cage around the enemy king (I guess no engine uses this motif), pawn race, don't stop the search after a mate has been found, and an attractive surprising solution move IMO, from a computer game. - In general you are right though, especially when it's not very clear which one of the possible wins really is better that the other anyway. Sometimes, such "2nd solutions" have been discovered during engine testing. I had such positions in earlier tests, but not in the Quicktest. Regards, Mike 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.