Author: Dave Gomboc
Date: 14:38:42 02/17/00
Go up one level in this thread
On February 17, 2000 at 17:17:08, blass uri wrote: >On February 17, 2000 at 15:01:57, Dave Gomboc wrote: > >>On February 17, 2000 at 12:39:36, Jeremiah Penery wrote: >> >>>On February 16, 2000 at 19:31:12, Dave Gomboc wrote: >>> >>>>On February 16, 2000 at 16:37:30, Jesus de la Villa wrote: >>>> >>>>> >>>>>After several hours neather(sp?)my software say something interesting, nor I. :( >>>>>Would you supply your main line, please. >>>>> >>>>>Thanks. >>>> >>>>I don't have one! NCO ends with an eval of +-, that's it! >>>> >>>>Here's Crafty 17.6's latest: >>>> >>>>depth=16 1/36 +0.00 14. Nxg5 Rxf1+ 15. Qxf1 h6 16. Bc4+ Kh8 17. Nf7+ Kh7 18. >>>>Qd3+ g6 19. Qe3 Qf8 20. Rf1 Qg7 21. e6 Nc6 22. Ng5+ Kh8 23. Nf7+ Kh7 >>>>Nodes: 2669305457 NPS: 70807 >>>>Time: 10:28:18.30 >>>> >>>>I am going to stop it thinking now; I have to get some real work done with my >>>>computer again. :-) If someone else has some decent hardware and another >>>>program, I'm interested to see how it does. >>> >>>On ply 15 with my modified Crafty, I was getting a fail-high with score >.02 >>>Since it was taking hours (it was running on 8 hours just to resolve the >>>fail-high), I was forced to stop it, so I could play Planescape. :) I did >>>follow the line for a few moves, and the score kept going up as I went further >>>along. I think it's just a case of computer mis-evaluation - white is slightly >>>down on material (at first), but black's pieces in the corner are useless to >>>him. The computer doesn't realize this. Not to mention that all white's >pieces are aiming at the nearly-naked black king. >> >>Crafty's variation wasn't bad at all. In the main line above, 21. Nxh6 wins the >>house, and Black's only real alternative is 20...Bf5, which still loses. But I >>was disappointed that it wasn't seeing that it was winning: the point of the >>variation is that White can castle (to reach the diagram position where I forced >>Black to play ...fxg5.) If a program can't see that the piece is untakeable, >>the program can't see that White can castle safely. >> >>Dave > >chessmaster6000(ss=10) can see in a less than 3 minutes that white is better >(+0.03) >The main line also begins with Ng5 Rxf1+ >The evaluation in the next iteration is +0.64 > >Uri +0.03 looks like a draw score artifact (i.e. draw score, but slightly adjusted, maybe to favor long draws over short draws). The +0.64 might be enough to allow CM to castle into the position. Dave
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.