Author: Matthew Hull
Date: 09:01:19 05/16/05
Go up one level in this thread
On May 16, 2005 at 11:27:29, Robert Hyatt wrote: >On May 15, 2005 at 15:53:11, Matthew Hull wrote: > >>On May 15, 2005 at 11:02:17, Robert Hyatt wrote: >> >>>On May 15, 2005 at 01:53:40, Joshua Haglund wrote: >>> >>>>On May 14, 2005 at 10:40:22, Robert Hyatt wrote: >>>> >>>>>On May 14, 2005 at 01:06:41, Joshua Haglund wrote: >>>>> >>>>>>On May 13, 2005 at 22:48:44, Robert Hyatt wrote: >>>>>> >>>>>>>On May 13, 2005 at 19:21:37, Joshua Haglund wrote: >>>>>>> >>>>>>>>On May 13, 2005 at 13:48:27, Pablo Ignacio Restrepo wrote: >>>>>>>> >>>>>>>>>Hello Robert >>>>>>>>>It was a mistake in redaction. Antiajedrez or Antichess have been my extreme >>>>>>>>>anticomputer style, across I have won many times many top engines in all time >>>>>>>>>game control. But usually Crafty have been for me a real problem for beat, iqual >>>>>>>>>than Rufian, Tiger Gambit 2.0 , Quark ..... >>>>>>>>>Usually I am playing into playchess. There I have won many top engines, but to >>>>>>>>>repeat..... Craftuy have ben very strong, but today in morning look ...... 2 >>>>>>>>>easy draw. Why? >>>>>>>>>Regard, >>>>>>>>>Pablo >>>>>>>> >>>>>>>> >>>>>>>>Hello Pablo, >>>>>>>> >>>>>>>>I've told you a couple times, if you remember, Dr. Hyatt removed the stone-wall >>>>>>>>detection code in version 19.15. It was removed because it appeared to Dr. Hyatt >>>>>>>>it wasn't needed anymore. >>>>>>>> >>>>>>>>As a result it should be easier for you using "anti-chess" playing Crafty. >>>>>>>> >>>>>>>>Joshua Haglund >>>>>>>>toneewa@yahoo.com >>>>>>> >>>>>>> >>>>>>>If I looked at the game correctly, this wasn't a stonewall type problem, it was >>>>>>>just e4 Nf3 and Bc4 for white if I am recalling the right game (I have looked at >>>>>>>several today so that could be wrong). >>>>>> >>>>>>[D]k1/2rbp1pp/1p3p1r/p2P1P2/P1P1B1P1/1P2R2P/5NK1 b - - 0 121 >>>>>>[D]n3/2rr1q1k/1p1p1p1p/p1p1pPpP/4P2N/P1PP2Q1/1P3PRK/6R1 b - - 0 38 >>>>>> >>>>>>These are to me "stone-wall" positions. I don't know how Crafty use to detect >>>>>>the "stone wall"... by pawn formation positions or by recognition of particular >>>>>>moves; maybe both. Pablo use to play Crafty and have not much luck before the >>>>>>removal of the "stone-wall" detection. >>>>>> >>>>>>Joshua Haglund >>>>>>toneewa@yahoo.com >>>>> >>>>> >>>>>The stonewall code looked for pawns at f4, e3, d4, c3. If three of those were >>>>>present, it assumed "stonewall"... This doesn't seem to match... >>>> >>>> >>>>What if the pawns are for black?? >>>>In these examples of Pablo's games; >>>> >>>>black pawns: a5,b6,c5,d6,e5,f6,g5,h6 >>>>definately stone wall with black. >>>> >>>>white pawns: a4,b3,c4,d5,f5,g4,h3 >>>>partial wall, but two strong pawn chains. >>>> >>>>The code was removed (19.15) so Crafty doesn't recognize anything like that now? >>>> >>>>Joshua Haglund >>>>toneewa@yahoo.com >>> >>> >>>Wouldn't matter. The "pattern" is a pawn at c3, d4, e3 and f4, or if you mirror >>>that for black, c6, d5, e6, f5. The "stonewall problem" is the pawn at e3/e6 >>>makes it very difficult for the bishop it is blocking. And the bishop tangles >>>other pieces up making a quick g/h file attack successful if the engine makes a >>>single mistake in untangling. >>> >>>If you have played e4/e5, then that problem doesn't happen, since d3/d6 lets the >>>bishop out. >> >> >>>To handle the case in the game(s) given, crafty depends on its >>>"blocked pawn" code to avoid the position getting so blocked and becoming >>>drawish. >> >> >>But is that logic only active if crafty knows it's playing human? > > >That's a good question. I'll go back a few versions and look. I don't think it >was "human only" however, but it might have been. > >Just looked at 19.6, and it was active for all opponents, not just humans. The >only way it was turned off was to turn off the asymmetric evaluation option, >which also turned the stonewall detection code off since it was asymmetric as >well. I'm probably thinking of the old internal GM and "blockers" lists, which used to take special care to prevent closed positions, IIRC.
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.