Author: Thorsten Czub
Date: 12:44:33 05/22/98
Go up one level in this thread
On May 22, 1998 at 12:48:53, Don Dailey wrote: >>Why ? I don't see that these 3 programs are much stronger than >>Hiarcs or Mchess. I do not believe that these 3 programs understand >>chess better than Hiarcs or Mchess. Do you ? > >I do. If they are winning they understand chess better. It might very >well be that we need to modify our view of what understanding chess is. >You seem to strongly believe that tactics are not a part of chess. Don - Tactis IS a part of chess. Where did I said it isn't. I can only QUOTE yourself: IF they are winning they understand chess better. IF ! Yes . And even WHEN they win, they will not win forever, or? For how long was Rebel/Hiarcs or Mchess or Genius superior ? And where was Fritz? Fritz 4/3/2/1 ? >I have seen the phenomenon in people, where certain "clumbsy" players >seemed to get better results than they deserved. They didn't know >thematic plans, didn't understand opening theory etc. but they got >good results compared to much "better players." But they did more >thing right than the apparently more knowledgable players. They >fought hard, didn't make errors and were good tacticians. So I ask >you, who really understood chess better? My friend Bernd understands chess. :-)) There are several way to play chess. Accepted. STYLE is a thing that can be differenciated between players. Some play elo 2500 and have style A. Some play elo 2500 and have style B. A is opposite of not subgroup of B. This is HUMANS. Chess programs have styles too. But the whole thing is not the same degree complex. Humans are much more complex than software. >But I even dispute your belief that Fritz and Nimzo are dumb programs. I have no problem with you having a different opinion. >As I've observed, Fritz seems to come up with consistantly good moves, >otherwise how could it possibly win unless you admit these other >programs are not "smart" enough to deal with this. Fritz does NOT come up consistently with good moves. Guess I am stupid. Like a worm. But I do always know 10 seconds earlier than you about the future. How could I fool you and imply that I am more intelligent than you are ? If Genius5 searchers 4 plies and plays against Genius5 with 6 plies search depth, I guess 6 plies will more often come with "consistantly good moves" although Genius5 on the other machine is the same KLOWLEDGED program. Fritz and Nimzo and Junior have (I do not doubt this) a SEARCH advantage. The new search-ideas have made them compute deeper than their competitive programs. It is always good to look intelligent if you have e.g. insider information. Insider information is often equal 2 plies ahead in a search. Sometimes even more. > >>>In my mind it >>>takes more skill to write a program in this way. > >I even disagree here (since I'm being disagreeable today.) The magic >of programs like Fritz is the great engineering involved in making >such a fast program so smart. Do you think this is easy? I don't think that Fritz is an UN genius program. Frans is a genius. And he and the chessBase team are all very clever guys. I have no doubt that this is true. Fritz is very very smart. But fritz - despite whatever strength it may have - plays ugly. It plays strong. But ugly. I have great respect for people working on such a smart program. I could not do this. I would be completely disapointed about the ugly moves and the senseless mainlines and would not stand to work on this. I have tried. Long time before. But failed. >Focusing on positional play and neglecting tactics is just plain >foolish, chess is a highly tactical game. You have to conform to >what works if you want ultimate chess strength, not what pleases you. Nobody does this. CSTal e.g. is not that stupid at all in tactics. We never tried to make CSTal play positional good. This job is for Mark and Ed. You are right that anybody should find HIS way to get the best out of his creativity. I only say that MY opinion is that the way Fritz is doing it stops me from doing chess. I can live with Nimzo98. I live with Junior. Maybe the style of it does not fit to my values. Maybe I have seen to many games were Bernd killed Fritz (and others) that I am too much disapointed. >I think the burden of proof is on you. Why. Because I have a different opinion ? Fritz has to prove that it can stand as long on top as all the others have done for years. Also it could try to win a championship :-)))) > So far, we've seen that both >approaches are very reasonable, there are good slow (but not too slow) >programs, but the fast approach is currently working the best. Why ? Which "god" or pope has shown this ? Which prove do you refer in this claim ? >>Big words. How can you prove them ? >>I have not seen Fritz showing positional understanding. > >It must make winning moves purely by accident. How does it >beat programs that understand more? Because it searches deeper. It finds the WINNING moves by accident in the tree. Completely right. >Again, you are not taking a wholistic view of things, but >pretending chess involves only a subset of skills that YOU >consider important. No - if I would do so, I would not like Nimzo or ChessTiger. Or even understand that Junior won a good championship. I would attack these programs too. But i do not. Since these programs don't produce such an amount of nonsense. Is this evidence enough ? Or how can I concinve you that i do differenciate ? How do you think was I able to outplay many 40/120 tournament games with Chess Tiger if this program would play as miserable as Fritz ? I can asnwer for you: the other programs play better (hint: better is subjective and means: IMO) so I can easily watch them playing chess. >Why don't you just come clean, recognize the facts and then >try to learn something from them? I see no facts that would convince me. I have Junior, Nimzo98, Tiger and Fritz. Fritz is the weakest program. It plays the worst moves. And plays senseless moves. It has no idea about the game. The other programs do better although they are fast-searchers. > Obviously, Fritz is doing >SOMETHING right, and whatever it is wins chess games. Obviously Fritz is doing something right. HERE we agree again. I have never said they are looooosers. They (ChessBase=Fritz) are very interesting and strong competitors. > You >seem so very close minded about how things should be done. I take this as an insult. I am not close minded. If you write me a chess program that is a fast searcher but plays good chess, I will be delighted. To call me close minded because I do not like 1 from 4 (ChessTiger, Nimzo98, Junior) fast searching programs looks a little offending in my mind. But i forgive you because you are a nice guy. > I >have no problem with the approach you advocate, I'm completely >open to it or anything that work, or look like it has a chance of >working. Like your approach. Why can't you be equally open to >other viewpoints? I am. I guess you misunderstand me. You have overseen or forgot that I was completely positively surprised about Nimzo last year when I bought Mchess7 and Nimzo98. I was even more enthusiastic about Nimzo98 than about Mchess. Maybe you forgot this. No problem. I am sure you are more open than I am. You have more knowledge and massive practise about computerchess. I am only doing a hobby. And I have no goal to become always right about a topic. But why do you believe I would not be open to someones viewpoints ? This hurts me. Did you forgot out talks at Aegon ? There I found - yes : don is very open minded. For me the WAY a program plays is important. Not how it gets this way. I am not that much focussed only on RESULTS. If a program gets the same results (against computers) but plays worse I will of course buy the other programs. >You will notice that Cilkchess is not fast, at least when you >consider the hardware. I use what I consider a balanced approach >of speed and knowledge. I'm willing to consider that I need >more knowledge, and I also consider that I may just need more >speed. But there is one thing I know is a sure thing, and it's >that more speed will ALWAYS help. In my mind the real trick >is getting as much well balanced knowledge in a program with the >very least speed penalty. You cannot admit that Fritz and Nimzo >have done an amazingly good job at this. Why do you berate them >for this. Again. I do not like 1 of 4. I don't see your point. I like Nimzo98 and have very FAST told so in public. I cannot repeat the fritz5 results and not the good games others see, and therefore cannot subscribe or like this program.
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.