Author: Mark Young
Date: 11:21:19 09/06/01
Go up one level in this thread
On September 06, 2001 at 13:13:22, Robert Hyatt wrote: >On September 06, 2001 at 12:49:57, Mark Young wrote: > >>On September 06, 2001 at 11:45:32, Robert Hyatt wrote: >> >>>On September 06, 2001 at 11:08:57, Uri Blass wrote: >>> >>>>On September 06, 2001 at 10:06:48, Robert Hyatt wrote: >>>> >>>>>On September 06, 2001 at 06:31:48, Uri Blass wrote: >>>>> >>>>>>On September 06, 2001 at 05:39:22, Geo Disher wrote: >>>>>> >>>>>>>On September 05, 2001 at 22:43:29, Jay Rinde wrote: >>>>>>> >>>>>>>>On September 05, 2001 at 22:32:52, Chris Kantack wrote: >>>>>>>> >>>>>>>>>On September 05, 2001 at 15:24:56, Roy Eassa wrote: >>>>>>>>>>I think that was the original point. Common people who make bets DON'T know >>>>>>>>>>much about chess. The common conception is that computers are better than >>>>>>>>>>humans. So the masses will bet against Kramnik, making the odds favorable for >>>>>>>>>>those of us who know better! >>>>>>>>> >>>>>>>>>I completely agree with what has been stated in this thread. But what if Vegas >>>>>>>>>was looking for some kind of "point spread"? That is, what will the final >>>>>>>>>match score be? I believe most of us know Kramnik could go undefeated if he >>>>>>>>>wanted to. But how will he play this match? Will he keep the score close to >>>>>>>>>make it look good? In other words, will Kramnik throw a few games to keep the >>>>>>>>>interest level up? I would not be surprised if that was the plan. >>>>>>>> >>>>>>>>You must be joking!! Yes, you are joking. >>>>>>> >>>>>>>I don't believe he is joking and I agree that it would be in Kamniks interest >>>>>>>not to win all the games if he could. It would appear that this match was much >>>>>>>diffrent than the Deep Blue Match if he put all his energy into this and won all >>>>>>>the games which I believe he might be able to do. This could jeopardize the >>>>>>>interest in another match. Anyway all this being said I would not like to bet >>>>>>>on the spread but on win or loss. Kramnik to win. >>>>>> >>>>>>I believe that if kramnik can win all the games there is no >>>>>>reason for him not to do it. >>>>>> >>>>>>I do not think that there will be no interest in another match >>>>>>in the future if kramnik gets 100% >>>>>> >>>>>>Kasparov also got 100% against Deep thought and later >>>>>>there was an interest in another match of kasparov >>>>>>against deep blue. >>>>>> >>>>>>I expect kramnik to win the match but I do not believe that >>>>>>he can win all the games when he does not have the exact program. >>>>>> >>>>>>I think you simply underestimate the machine. >>>>>> >>>>>>A weaker version that did not use 8 proceesors drew 3-3 with >>>>>>GM heubner and I do not believe that >>>>>>kramnik can get 100% against huebner. >>>>>> >>>>>>Uri >>>>> >>>>>Chess is not necessarily transitive. Nothing says that if A beats B 100% of >>>>>the time, and B plays evenly with C 100% of the time, then A will beat C 100% >>>>>of the time. Even the Elo calculations don't suggest that. >>>>> >>>>>I don't think Kramnik will win every game. But not because he can't. Because >>>>>the match is set up in such a way that taking any risk should be avoided. I >>>>>suspect he is smart enought and wise enough so that he will adopt a match >>>>>strategy that attempts to maximize money won without taking undue risks that >>>>>could backfire. >>>> >>>>If he believes that he can win every game then he can try it. >>>>I do not see a significant risk because it is enough for him to win more than to >>>>lose and a strategy of probability of 90% to win every game means almost 100% >>>>chances to win the match >>> >>>New game. I am going to pull the pin on 6 hand grenades, one at a time, and >>>toss them to you. If you catch one, you "win". If you get blown up, you >>>"lose". >>> >>>My strategy will be to _not_ try to catch all 6. I am going to watch you >>>very carefully and notice how quick you pull the pin and toss the grenade. >>>I am going to pick the one where you move the quickest, giving me the best >>>chance to catch it before it explodes, and still leave time to toss it safely >>>away after I catch it. After I catch _one_ I am going to run like hell and >>>ignore the rest. I have already won the match. >>> >>>I think any smart player will play the match just like that. Based on the >>>current prize rules, I probably would just forfeit all games and take home a >>>small fortune to retire on, anyway. But to be a bit smarter, I would really >>>try to draw game after game, safely, unless my opponent makes a positional >>>blunder that I believe I can exploit without much risk of losing. I will try >>>to win that game, then go back to drawing, which will win the match and the >>>$1M. >>> >>> >>> >>> >>>> >>>>The probability of Deep Fritz to draw the match against this strategy is >>>>0.9^4*0.1^4*70<0.5% and kramnik has more than 99% chances to win the match. >>>> >>>>It is not a significant risk and the fact that kramnik needs to work for less >>>>games is enough compensation for it. >>>> >>>>90% to win every game means less than 50% to get a 8-0 result and less then 60% >>>>to get 5-0 result. >>>> >>>>Uri >>> >>>I don't know that I buy the 90% probability first. The real question is >>>what is the probability that the human will make a result-altering mistake in >>>a game and lose a game that should have been drawn or won? It is not zero. >>>The more complicated the game is, the greater that probability becomes. I >>>believe he will try to keep the probability very low by playing safe chess >>>waiting on (a) a mistake or (b) the end of the match giving him a huge cash >>>bundle no matter what happens. >> >>I agree, I think GM Kramnik will have no trouble beating Fritz 7. I do think he >>will try and win one game regardless, then coast if he can. >> >>I do think Deep Blue is stronger then Fritz 7 dispite what GM Kramnik and other >>say here and else where. The question for me is will we be able to judge the >>strength of Fritz 7 from the upcoming match. I fear the answer will be no, >>unless we see Fritz 7 win. As you point out we may see 1 Kramnik win and the >>rest of the games will be draws, that would be the smart way to play the match, >>but it will tell us little about the true strength of Fritz 7. > > >Ok... > >Reasonable comments. > >Here is my take: > >If DF wins convincingly (IE by a more than one-game margin) then I will have >to re-think my GM evaluation. If DF wins by .5 or 1.0 points, it is possible >that Kramnik was too conservative, then lost one game by accident, and couldn't >catch up. If the match is tied, it won't say much either. If Kramnik wins >by some narrow margin, it could mean one of two things. He was lucky. Or he >was ultra-conservative. > >I think the match result will therefore, in my opinion I will add, not be >very important. I think the _games_ will reveal far more. IE if they are >kasparov/DB like games, where it was obvious Kasparov was _not_ trying to >draw, then the match result will be much more significant. IE if we see >Kramnik trying to win, and DF trying to win, then it will be exciting and >revealing. If Kramnik goes way conservative, the result will be less >informative. > >I would not expect Kramnik to lose, but strange things can happen in an 8-game >match. If he loses one game, the last one, that won't say much. If he loses >an early game, and then can't catch up, that will say a _lot_. > >I'll certainly give my impression after it wins, not that my impression carries >a lot of weight. > >I think, therefore, that the final match score will have to be looked at in >light of the actual moves played in the games. If Kramnik tries to win every >game, and the match is drawn or he loses, that will be _very_ revealing. It >all depends on what he tries to do... And it also depends on what the DF >guys try to do as well, as they can try to steer for very drawish positions >as well, although against a GM a "drawish position" can be a "losing position" >if your endgame skills are not up to snuff. May be the only way to judge the strength of computers under match conditions is to use the old WC match rules of winning 6 games draws not counting, but this is certainly unfair for the human player if the match last over 30 games. I think and I know you disagree but the only fair test for judging the strength of computer programs seems to be playing tournaments. Even if the players choose not to prepare in some way for playing against the computer programs.
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.