Author: chandler yergin
Date: 12:01:30 04/22/05
Go up one level in this thread
On April 22, 2005 at 14:46:46, Terry McCracken wrote: >On April 22, 2005 at 14:36:40, chandler yergin wrote: > >>On April 22, 2005 at 12:32:15, Terry McCracken wrote: >> >>>On April 22, 2005 at 12:16:52, chandler yergin wrote: >>> >>>>On April 22, 2005 at 12:01:09, Mihaly Szalai wrote: >>>> >>>>>On April 22, 2005 at 09:29:13, Andrey Popov wrote: >>>>> >>>>>>This is the last position of the last game of Kasparov-Karpov match. >>>>>>If Karpov (Black) reaches a draw, he wins the match. >>>>>>Of course, White has a huge advantage. >>>>>>However, I do not see any way to win. >>>>>>5n1k/5Q2/4p1p1/2q1P2p/7P/6P1/5PK1/3B4 b - - 0 64 >>>>>>[D]5n1k/5Q2/4p1p1/2q1P2p/7P/6P1/5PK1/3B4 b - - 0 64 >>>>>>Why did Karpov resigned? >>>>>>Can anybody see a series of White's moves leading to win after Qc5-b4? >>>>>>Engines quickly go to score about +3.00 and stick on it forever. >>>>> >>>>>According to Kasparov this is the winning line: >>>>> >>>>>64... Qb4 65. Bf3 Qc5 66. Be4 Qb4 67. f3 Qd2+ 68. Kh3 Qh6 (68... Qb4 69. Bxg6 >>>>>Nxg6 70. Qxg6 Qxh4+ 71. Kg2) 69. f4 Qg7 70. Qxg7+ Kxg7 71. Bc6+- >>>>> >>>>>and we can agree with him: >>>>> >>>>>71... Kf7 72. Kg2 Ke7 73. Kf3 Nh7 (73... Kd8 74. Ke4 Kc7 75. Be8 Kd8 76. Ba4 Kc7 >>>>>77. Kd4) 74. Ke4 Nf8 75. Kd4 Kd8 (75... Nh7 76. Kc5 Nf8 77. Bb5 Nh7 78. Kc6 Nf8 >>>>>79. Bd3) 76. Kc5 Kc7 77. Be4 etc. >>>> >>>> >>>>Neither of you took the line to Mate! >>>> >>>>The +- & etc.. is not convincing! >>>> >>>>What engine can 'demonstrate' a Mate? >>>> >>>>The PV evals are static positional values, and meaningless unless a Mate >>>>is found. >>> >>>What you wrote is meaningless....Chan think before you type. >>> >>>First you don't know whether a computer was even employed, and even if it was, >>>it's not important to show checkmate, which is some distance away. >> >>This was posted by Uri: >>"I could probe clearly higher score than +3 by yace after Qc5-b4" >> >>[Event "?"] >>[Site "?"] >>[Date "????.??.??"] >>[Round "?"] >>[White "New game"] >>[Black "?"] >>[Result "*"] >>[SetUp "1"] >>[FEN "5n1k/5Q2/4p1p1/2q1P2p/7P/6P1/5PK1/3B4 b - - 0 1"] >>[PlyCount "16"] >> >>1... Qb4 2. Bc2 Qc5 3. Kh3 Qxc2 (3... Qb4 4. f4) 4. Qxf8+ Kh7 5. Qf7+ Kh8 (5... >>Kh6 6. Qxe6 Qxf2 (6... Qb1 7. Kh2 Kg7 8. Qf6+) 7. Qf6 Qxf6 8. exf6) 6. Qxe6 >>Qxf2 (6... Qe4 7. Qf6+) 7. Qf6+ Qxf6 8. exf6 Kg8 9. g4 * >> >>A computer was used to evaluate the FEN position. >>My comment: >>"The PV evals are static >>positional values, and meaningless unless a Mate is found." >> >>If the engine cannot find a Mate; what good is the eval? >>Showing a Plus or a +- is misleading, if a win is beyond the capability of the >>analysis module. You need more moves to take it a proper ending. >> >>There are many positions where a +- is shown, or even as much as a +(2.00) >>and the position is still a Draw. > >You must see the flaw in your logic? I see the flaw in yours! Of course there is a margin of error, but >Chan, a machine need not see mate to see it is clearly winning. No, the Eval is just that, an 'eval', and every Engine will evaluate the position differently. > >I don't need to see mate 30 moves down the road to know whether I'm winning, nor >do you. As Yogi says.. "It ain't over til it's over"! Any Programmer will also tell you that. Ask any GM or Master. > >If the position is won, you can rely that a computer will usually see the win, >but mate may be too distant. If the Mate is beyond it's horizon, it has not found the win. It will evaluate based on Material first, Mobility, King safety, and other Static Positional values in a quiescent position, that are programmed in. > >The same can be said about us. >> >> >>> >>> >>>And if it were all computer, to state and I quote, "The PV evals are static >>>positional values, and meaningless unless a Mate is found." Isn't correct >>>regardless. I repeat: This is true: The PV evals are static positional values, unless a forcing line is found and that will be reflected in the PV, usually going to Mate.
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.