Author: Uri Blass
Date: 11:28:55 09/26/00
Go up one level in this thread
On September 26, 2000 at 13:21:17, Steffen Jakob wrote: >On September 26, 2000 at 12:29:04, Bruce Moreland wrote: > >>On September 26, 2000 at 11:25:03, Steffen Jakob wrote: >> >>>[D]8/6k1/p5P1/1p5K/1P6/8/8/8 b >>> >>>Here Hossa continued with 5... Kf6 which is indeed losing. All moves seem to be >>>losing here. But when I forced 5... b5 6. bxa5 Hossa gave a score of mate in 54! >>> >>>How is this such a position handled in tournament chess? Is this a draw (I think >>>the rules have changed here several times?!)? If it's a draw then I will change >>>my tablebase probe code to return a draw score if the mate distance is > 50. >> >>You mean 5. ... a5, which goes into a QP vs Q ending. As was pointed out in >>another post, your example is probably moot here since the pawn gets to move and >>resets the counter, but even if it doesn't, you can ignore this case. >> >>We currently have DTM tables, and you you assume that DTM > 50 is a draw you'll >>get worse results than if you leave it alone. You'll play into positions where >>your opponent can convert in under 50 moves, and against humans you'll avoid >>endings that they will mess up. > >Yes, but it would be nice to have tables where a draw score is stored if the >conversion is not possible. I assume that those tables would be harder to >create. > >Greetings, >Steffen. No These tables are more easy to create because you need less time to calculate them. The reason that programs use the tables of distance to mate is only the fact that they are too lazy to write tablebase generator and the tablebase generator that is available is of distance to mate. Practically the difference in result is less than 1 elo(I do not know about one game that was drawn by the 50 move rule when tablebases said long mate in the ssdf games) and programmers who care about the rating of their programs prefer not to work about it(there are more important improvements for this purpose). 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.