Author: Thorsten Czub
Date: 18:12:59 02/28/98
Go up one level in this thread
>To conclude something useful you'll have to have enough data. Getting >data (games) is not easy, it is very difficult and time consuming. Ok, I present you results: Fritz 300 points. Hiarcs 6 points. These are my RESULTS . And now you can do the statistics. I needed 2 seconds to present you with my results. I have no data ! >I can have a "feeling" after a few seconds. To have a proofed therory >I'll have to work much longer. So where is the difference ? The difference is: Results do not play chess. They don't tell us about which opening was chosen. Not if it was an endgame or if ONE machine stopped autoplaying nor if it was adjudicated. Results do only show NUMBERS. I do like to replay the game and get a feeling for what has happened. Maybe I replay all games with my friends and correlate with them. After this, I feel ! This takes the time it costs to replay the games. But I guess it is much more precise than statistics. SO: why should I believe in your numbers when I can replay your games. It doesn't make sense. And if you don't have games, than why not believing my above results 300:6 ?? >Why do you play chess with machines if humans could do this much better? Because I cannot learn from humans. I can learn better myself by watching others not playing perfectly well machines HOW to do. When I ask Bernd about something he explaines it to me. Together with all the main-lines and evaluations of the computers I have seen, the impression is much saver. I get Bernds idea about a strong human watching or commenting the game and my conclusions from watching the main-lines and the fail-high/fail-low correlations or WHEN which program showed WHAT. This is more than any RESULT can show me. I guess Ed is doing it the same. I know many computerchess people doing it this way: watching the computers play against each other, and maybe asking a strong chess player (Bernd has 2430 and I know him since I was 15) you are used to understand. >But you can never really prove that one of two good wines is better. >Three people, three different "feelings". I do trust Bernds statements. And ask my computers if they can refute Bernds ideas. if the machines say: no - bernd is right. I am satisfied. I could never believe only in numbers. Sorry. Statistic is IMO a science like an alibi after you have killed somebody. Statistics gives ideas about what could have happened AFTER the real games have happened. I prefer watching the games, not the numbers out of them. >You'll have to use your feelings in a number of occasions. But I don't >like you trying to tell us that - when it comes to computer chess - your >feelings are more relevant than thousends of games + the statistical >conclusions. Thats all. Nothing more, nothing less. Why ? Why don't you like feeling beeing superior over science ? Bad experience ? Whatsoever. As I said. Your statistic is worthless unless we have all the games. I create my feelings OUT of the games. I do not meditate without data. >Andreas In my opinion you behave like somebody telling me Stephan King is a bad author, and when I ask you, WHICH BOOK did you read and which thing you don't like ? And you answer me, oh - I have read a statistic about his books. I never read one book myself. But i read about people have read his books. I prefer reading the books myself and give a subjective feeling afterwards. I know that my feeling is not precise. But i trust more THIS special feeling than the hear said of people who present data I have not seen with own eyes ! Thats the difference. YOu believe in numbers, I do believe in what I have seen and what I discuss with other people.
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.