Author: margolies,marc
Date: 20:26:27 01/23/04
Go up one level in this thread
HI Bob, Thanks for responding. I get it now. You know New York there are lots of actuaries who play chess, usually in the metropolitan league or the bankers league. So be careful who you say that to about statisticians or your premiums get hiked :o( The salt of my question, while not negating the powerful tool of post-ortem game analysis as a means of improvement, .."can we expect our results to be analytically useful when all players are not attempting to achieve victory when we enter a contest designed for that purpose. And if not, is it honest advice to ignore results?" My sensibility-- my human sensibility-- is that I do not have to win each game because I also have long-term goals to meet in competition. And I know that winning all games, or striving for it leads to very safe, boring play. I personally want to grow as a player and as a person-- also creatively. So I find ineresting acceptable risks ( these are situational, one doesn't take the same risks always it depends on tournament position, personal energy, etc). While I am not sure how to apply this to computer chess tournaments, I do take as an ethical given that competition itself should strive to be honest and directed. In that regard I would want to know that each player is striving seriously to achieve a win in each game. This is how I came to ask whether you think winning is important or not in Comp Chess matches. Because as an end user of chess engines I do take it seriously. I also think that once people do not accept that these are serious contests, there will be less observers and consequently less purchases and programming in this arena (no pun intended towards FQ). -Marc On January 23, 2004 at 20:29:33, Bob Durrett wrote: >On January 23, 2004 at 15:55:38, margolies,marc wrote: > >>I am trying to figure this, Bob. Did you suggest that tournament results would >>be more or less useful for program development when participants and observers >>cared less about the tournament outcome? >>thanks, Marc >> >> >> >> >>On January 23, 2004 at 11:26:42, Bob Durrett wrote: >> >>---snip--- >>>I see a tournament as being a tool too. It helps us to improve our odds of >>>being on the right path. If we seek to determine who [or what] is best, who is >>>second best, etc., then statistics provides us with useful tools to improve our >>>chances of "getting it right." But there is no guarantee. >>> >>>It appears absolutely true that every tournament provides some information. The >>>problem is in identifying that information, measuring it, and finding a good way >>>to express or communicate that new information. >>> >>>This forum seems to have a problem coming to grips with the idea that there is >>>useful information in every tournament. People seem "hung up" on the mandate to >>>determine a "winner." That causes all kinds of misunderstandings. >>> > >No, Marc. My bulletin was more in the nature of a private joke between me and >good 'ol Rolf. > >It is possible that I was also taking a "friendly jab" at another fellow here as >well, but that was not you and the guy is one of the "good guys" too. > >Truly, my feeling is that drawing conclusions based solely on win/loss/draw >statistics is a losing proposition right from the start! > >If I were the one who played those games, or if my brainchild chess-playing >program were to have played those games, you'd better believe that I would >post-mortem analyze those games to death! All this hoo-rah about wins, losses, >and draws would be down in the noise level. I would be searching for test >positions where my engine didn't play as well as it should have and my main >concern would be in fixing any bugs or weaknesses in my program. > >Having studied probability and statistics as a math major in graduate school [I >later got a graduate degree in engineering], I got more exposure to statistics >than any SANE human would ever want. When I retired, I took Emanual Lasker's >advice and immediately set out to purge as much of that nonsense out of my mind >as possible. [As I recall, he said he memorized a bunch of chess "information" >as a young man and then spent the rest of his life trying to forget that stuff.] > >Do you realize that there are people who actually devote their entire careers to >statistics [I call it "sadistics"]? Talk about an esoteric or arcane >profession, . . . that's it! It's worse than being a chess engine book maker or >an endgame tablebase guru! : ) > >So, in a sense I sort of agree with what you said, sort-of, assuming that you >don't mean that too literally. Winning a chess tournament can be a great ego >booster. I know from personal experience, having won a few. But I think >in-depth analysis of the games would return a lot more information useful for >improving the design of the chess-playing programs. > >Only IMHO, of course. I am not a chess programmer so cannot be sure [with the >right confidence level, etc., etc. . . . ] > >Bob D.
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.