Author: Uri Blass
Date: 01:45:39 05/26/02
Go up one level in this thread
On May 25, 2002 at 21:40:47, Miguel A. Ballicora wrote: >On May 25, 2002 at 16:24:21, Uri Blass wrote: > >>On May 25, 2002 at 15:02:51, Miguel A. Ballicora wrote: >> >>>On May 25, 2002 at 11:07:16, Vincent Diepeveen wrote: >>> >>>>On May 25, 2002 at 05:07:17, Uri Blass wrote: >>>> >>>>>On May 25, 2002 at 03:38:11, Jeroen Noomen wrote: >>>>> >>>>>>On May 24, 2002 at 18:34:18, BORIS YUDOVIN wrote: >>>>>> >>>>>>>Please answer my question. >>>>>>>If Fritz 8 play against different people with Elo rating >>>>>>>2200-2300 (corresp.chess ) what result can we expect. >>>>>>>(score %).Amount of games 500. >>>>>>>Thank You.Boris >>>>>> >>>>>> >>>>>>If you play against correspondence chess players of 2200-2300, >>>>>>there is a big chance you play Fritz, Shredder, Tiger, Rebel, >>>>>>or another chess program :-). >>>>>> >>>>>>Jeroen >>>>> >>>>>I think that this is mainly the case with higher rated correspondence players. >>>>>I believe that most 2200-2300 correspondence players or not use a computer or >>>>>use programs only for a short time. >>>>> >>>>>I expect top programs to get 80% against 2200-2300 correspondence players. >>>>> >>>>>The match of Steve Ham against computers when Steve Ham(2508 ICCF) lost 2.5-1.5 >>>>>proved that programs can get a correspondence performance of more than 2500. >>>>> >>>>>Steve Ham did not play anti computer but most correspondence players do not try >>>>>to play in anti-computer style and if you say that the correspondence players >>>>>cannot use hardware for 24 hours per move in every game then I can answer that >>>>>today it is possible to use better hardware then the hardware that was used >>>>>against Steve Ham. >>>>> >>>>>My experience in correspondence games also suggest that computers can get >>>>>performance of more than 2500(I used also my head because performance of more >>>>>than 2500 was not enough for me to be the Israeli champion) >>>>> >>>>>Uri >>>> >>>>Most 2200-2300 corr games i see are 100% computer moves Uri. >>>> >>>>That you 'can' get higher with a computer doesn't say anything >>>>about reality here. >>>> >>>>Another thing where i have to laugh loud for always is that all >>>>correspondence chess players who are themselves rated pretty low, >>>>that they conclude that a certain program X is best program. >>>> >>>>The reason behind this is that they suck so much that they usually >>>>go for the biggest patzer move. So if they use more than 1 engine >>>>to analyze with, then what happens is they play the minimum strength >>>>of the both engines, resulting obviously in worse play than simply >>>>using 1 engine and giving it 24 hours of computing time. >>>> >>>>However the strong world top corr players i know, they all distrust >>>>the computer completely. These are rated obviously way higher than >>>>2200-2300, and their own rating OTB rating is not always saying >>>>something about how strong they play their corr games. >>>> >>>>Analytical insight, systematic analysis, and taking time at the right >>>>moment are the bottom lines. >>>> >>>>No computer has a chance against them. Some are rated 'only' 2400 >>>>which obviously means that playing with a computer only is not garantueeing >>>>a high rating anymore in 2002. >>> >>>Let's not forget that correspondence ratings are probably inflated a lot. >>>The best players at this specialty probably never played a single game. >>>For instance, I have no doubt in my mind that Kasparov will eat alive any >>>correspondence player by a huge margin. What would be the rating of Kasparov if >>>it decides to really go into it? 3000?. >> >>I do not think that it is going to be 3000. >> >>Remember that the facts that both sides use computers reduces significantly the >>advantage of kasparov. > >>I guess that he is not going to be more than 100 elo higher than the best >>correspondence player and I doubt if he is going to be better than the best >>correspondence player. > >Do you really doubt it? Yes If there is a correspondence match of 10 games between kasparov and irena krush I expect kasparov not to do better than 6 out of 10. Kasparov failed to prove superiority in a correspondence match against the world and could get a winning position only after the advice of irena was not accepted because it came too late if I understood correctly. I also know that there are GM's who play correspondence chess and irena is not a GM. Peter Leko plays correspondence chess and he had not good results. It seems that he did not waste a lot of time in correspondence games and the only GM who takes seriously correspondence games may be GM Ulf Anderson but if kasparov can get 3000 in correspondence games then it means that Ulf anderson can get at least 2900 because the gap between players when both players can use computers is clearly smaller. >Once you said that a computer could have a better static evaluation of an ending >than Capablanca (it was IIRC KBPPP vs KNPPP 3 vs 1 on each flank). I think that >you really underestimate the knowledge and power of those kind of guys. I think that the main advantage of humans except better selective search is not in better static evaluation but in learning from search to change the evaluation. I have some small knowledge about it in Movei and the evaluation is not based only on the leaf position but today(if I have no bugs) the change in the evaluation based on other factors is very small. Uri
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.