Author: Robert Hyatt
Date: 10:37:48 05/23/99
Go up one level in this thread
On May 23, 1999 at 12:30:21, José de Jesús García Ruvalcaba wrote: >On May 22, 1999 at 21:12:31, Robert Hyatt wrote: > >>On May 22, 1999 at 16:38:07, James Robertson wrote: >> >>>On May 22, 1999 at 15:53:48, Mark Young wrote: >>> >>>>On May 22, 1999 at 15:40:58, Marc Plum wrote: >>>> >>> >>>[snip] >>> >>>> >>>>Well, I would not put to much stock in this one game. Rebel 10 played I think >>>>the best player when playing against computer programs. Also again it seems >>>>Rebel 10's bug?? or huge tactical hole in its program that I have talked about >>>>before came up and bit Rebel 10 hard again. >>> >>>Bug? Huge tactical hole? On my machine (P233), my program searches more than 10 >>>minutes and shows no signs of seing the danger for white _after_ 26. f4. Crafty >>>16.3 still does not see the danger and it has been searching for more than 11 >>>minutes. >>> >> >>You are using the wrong crafty. 16.6 is far better at king safety and already >>feels that this is big trouble for white, even before f4. IE at move 23, at >>depth=14 (1.5 minutes on my quad) the best move (exd4) had already failed low >>for white... >> >>The problem is that _crafty_ was picking these threats up at depth=13/14, >>which was way beyond Rebel's search depth most likely... >> >>After f4, this is crafty's analysis during the game: >> >> (4) 12 24.80 -0.26 26. ... Qd3 27. Rxb7 Qxd4+ 28. Kh1 >> Rc8 29. Qa1 Qd2 30. Rd7 Re8 31. Re7 >> Rc8 32. b3 Qc3 >> 12 1:05 -0.07 26. ... h5 27. Rxb7 Ra5 28. Rb5 Rxb5 >> 29. cxb5 Kh7 30. b6 Qd3 31. Kh1 Qxd4 >> 32. Bf3 <HT> >> 12-> 1:05 -0.07 26. ... h5 27. Rxb7 Ra5 28. Rb5 Rxb5 >> 29. cxb5 Kh7 30. b6 Qd3 31. Kh1 Qxd4 >> 32. Bf3 <HT> >> 13 1:38 ++ 26. ... h5!! >> 13 8:21 3.43 26. ... h5 27. Rxb4 Re8 28. Rb5 Re3 >> 29. Qa1 Re4 30. Qa8+ Re8 31. Qxe8+ >> Qxe8 32. Kf2 Bg4 33. Rxb7 Qe2+ 34. >> Kg1 Qe3+ 35. Kg2 Qxd4 >> 13-> 8:45 3.43 26. ... h5 27. Rxb4 Re8 28. Rb5 Re3 >> 29. Qa1 Re4 30. Qa8+ Re8 31. Qxe8+ >> Qxe8 32. Kf2 Bg4 33. Rxb7 Qe2+ 34. >> Kg1 Qe3+ 35. Kg2 Qxd4 >> 14 9:42 3.43 26. ... h5 27. Rxb4 Re8 28. Rb5 Re3 >> 29. Qa1 Re4 30. Qa8+ Re8 31. Qxe8+ >> Qxe8 32. Kf2 Bg4 33. Rxb7 Qe2+ 34. >> Kg1 Qe1+ 35. Kg2 Qd2+ 36. Kh1 Qd1+ >> 37. Kg2 Qxd4 >> >>Any crafty should find h5 reasonably quickly... as this is just a tactical >>rip of white, totally... Took mine 1 minute to realize that h5 was the >>right move, and 1:38 to realize that it was outright winning... >> >> >>>>As in my computer vs computer games >>>>Rebel 10 gets a advantage >>> >>>You honestly believe that Rebel had an advantage before playing 26. f4? That >>>move was just the icing on the cake. >>> >>>James >> >>Right... white was lost (according to crafty) when it played Rb5, as my >>scores started failing low/high (depending on who was on move) getting very >>good for black, but at extreme depths (but depths it can actually reach as this >>analysis was being done real-time during the game in channel 211). Rohde said >>white was lost on Ra5, which was move 21. However, f5 showed that something >>is definitely missing in king safety. And there are a _bunch_ of players around >>that will exploit that until it gets cleared up. As the saying goes, been >>there, done that. >> > > Were these analysis done by crafty 16.6 or by crafty 16.7? 16.7, but the only difference is evaluation. Search hasn't changed at all, and what it was seeing was tactical issues, not evaluation issues...
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.