Author: Thorsten Czub
Date: 05:24:24 11/11/00
Go up one level in this thread
On November 10, 2000 at 23:27:47, Bruce Moreland wrote: >You showed a game between Crafty and Tiger and I said that it didn't demonstrate >anything about what was in the title. The title spoke of accuracy vs >inaccuracy. What happened in the game was fairly accurate defense, an >aggressive attack, and an endgame blunder. therefore i showed the score of gambit-tiger. to show the difference in the score between crafty and gambit-tiger. in the second game (1.a3 a6) the different score behaviour ws exactly the same, only that i did not presented the data this time. >I also made mention that you can get an attacking position more easily in an >opening where you are afforded one by choice of opening. > >You took this second statement and ran with it, and did this a3/a6 thing. I >think that was a fine game. Gambit got a good attack and won the game. You >showed that it is aggressive and plays well, and Crafty crashed and burned, >which I'm sure will appeal to a lot of people, but I hope that the games aren't >viewed as personality battles. bob and i have to calm down. so much that has been said was without thinking, only because different point of views cannot be combined or linked together. >There is so much other stuff swirling around here that it's difficult to discuss >what is actually presented. What is going on is much more than a discussion of >a chess game or a discussion of evaluation philosophies or anything else. A lot >of this stuff is people being pissed off at each other right. excatly. > because they can't deal >with the simple fact that nobody is right 100% of the time. :-))) for some people accuracy seems to be a god. as a human beeing, we should understand that we cannot be 100% right. maybe spock can sometimes, if he controls his emotions. but at least each 7 years he runs out of control... i wanted to present the things i see different in gambit-tigers behaviour in relation to other programs. >My god, who cares. right. >And pavel makes a mistake to try to get you to specify some match condition that >will solve everything once and for all. It's silly to make you jump through >hoops rather than just talking about what these game-analogies are supposed to >prove. they are examples. we canot discuss all the games i have played on my autoplayer so far. that would take time :-)) >Do you want to prove that Bob is stupid? i want to show that christophe goes a different way than bob believes in. and others believe in. gambit tiger is more than tuning a few evaluations. > That is kind of nasty and I don't >believe that that is what you are after, it's not up to me to prove this. bob is a doorkeeper. he is the old paradigm IMO. i think its time so say goodbye. i am on this "trip" for a while. as long as i remember i searched for the different in chess-programs. early 1992 it was with mark uniacke on hiarcs. i remember very good how much i was impressed when mark send me a latest version and it killed almost anything on my machines. he said: oh - i never tried it out against others. is it good ? wow ! he only tried it out against MCHESS. because he liked how Marty put knowledge into computerchess. and mark tried to copy this approach. the hiarcs of today is the result. a knowledge based program, the version 7.32 gets normally outsearchered by the latest fritz and junior programs. but hiarcs is still very good. if you use it on a 1 Ghz machine, it can really be a good tool. i remember when i told him: you will win because hiarcs kills all the other opponents. he first did not believed it. that was the same way with christophe years later. when it came to the championship in munich, i was sure hiarcs will make it. so my travel to munich was to visit how hiarcs will gonna make it. i operated Mchess in munich, and it when i had time, sat next to mark to keep the fingers crossed for him. also i had the typical battle with ossi weiner when it came to the mchess-genius match in munich. stefan meyer kahlen was also there.he said he believed while watching the dialogue that ossi and i would strangle us in the moment nobody looks to the desk. but it was only arguing about : is the evaluation of mchess (the very high speculative) accurate, or is genius right. this discussion is so old. doing it with ossi, with bob or with anybody else... mephistoIII,superconny, mchess, hiarcs, cstal or gambit-tiger. its always the same discussion. the ones try to held back any kind of CHANGE and the others like the new approach. >but if it is, that's probably won't be >very productive. right. we will stop this. it leads to nothing. >Do you want to prove that Tiger is better than Crafty? One game won't prove >that so why even bother trying with one game? :-))) >Do you propose to prove that speculation is better than a more conservative >evaluation strategy? as i said (did i?) speculation is NOT the trick IMO. if it would be, i would not call it new paradigm. > The first game didn't show this, there was no speculation >in it. ?! gambit - tiger says 3.42 for move 39. so you believe too that this score is real. > In the second game Crafty went pawn-grabbing. I think that even a >conservative eval should have been able to decide that putting the queen offside >and going after a bad pawn is a bad idea. And without having played it through, >I don't know if the pawn grabbing was just something you do after you realize >you are lost, or if it was the cause of the problem, or something else. >I think what you have the best chance of proving is that a program that is tuned >for aggression can bowl over other computer opponents, and make itself look >great in the process. aggressive play isn't it at all. i remember nimzo experimented with aggression. if you have no knowledge, agressive playing style can be at least a method to get a similar kind of thing like a "plan". aggression can replace knowledge, but only to a certain small degree. than it backfires. >No argument there, certainly. I think that aggression is a great evaluation >term. >Perhaps in Tiger this is also linked to speculation, but I haven't seen much >speculation out of it yet. I do like the idea of speculation, and want to make >mine more speculative.. I think this can make a program play more interesting >chess, and in the long term might make it play stronger chess as well. right. >bruce we have to wait for more examples.
This page took 0.01 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.