Author: Walter Faxon
Date: 16:19:25 02/03/03
Go up one level in this thread
On February 02, 2003 at 16:10:37, Graham Laight wrote: >On February 02, 2003 at 11:14:35, Uri Blass wrote: > >>On February 02, 2003 at 10:10:04, Graham Laight wrote: >> >>>Since I made the pattern recognition algorithm "more aggressive" yesterday, >>>nobody has posted any results! >>> >>>It looks like people will only paste their results here if they win - am I >>>right? >>> >>>One minor modification today - the program will start playing you on the third >>>turn now, not the seventh. The sooner proper play starts, the sooner it can >>>understand your personality. Have another go... >>> >>>http://mysite.freeserve.com/grahamlaight/jscript/GuessWhichHand.htm >> >>I simply did not play. >> >>I take my games seriously and after winning only by result of 101-100 before the >>modification I was afraid to try again. >> >>You said that it is easy to win the computer in short games so I thought about >>the possibility to try a gambit in the first moves to confuse the computer. >> >>A simple idea that gives good chances may be to get an advantage in the >>beginning of the game and to continue by random choice. > >You are, of course, most welcome to play the game in any way you wish! > >Your method proposed above would give you a better than even chance of winning - >but it strikes me as being just a teeny-weeny bit unsporting - a little bit like >saying to a human opponent, "We'll play chess over the board so long as, at a >moment of my own choosing, we can switch to using computers to select the >moves". > >>It may be interesting if somebody can find a simple formula to get a convincing >>win against the computer. > >This is certainly possible. At the moment, ANY feedback would be welcome! :) > >I know that I, personally, do not play my own game very well. Now that I've >changed the pattern-matching algorithm, I personally can do much better in the >early stages of the game, but I find that the computer becomes clever in >guessing my choices sooner. I wanted to know if this finding applied to other >people as well. It could be that the changes I've made actually make it easier >for other people to win rather than more difficult - at this stage, I can only >guess. > >Oh well. > >>I believe that it may be possible to define some pattern that the computer does >>not consider. > >Maybe someone out there is, even as we speak, writing a program which will play >the game better than mine does (which is undoubtedly possible), and we can have >the world's first "Computer "Guess Which Hand" Competition"! :) > >-g > >>Uri Hi, Graham. Your idea of a computer "Guess Which Hand" competition is interesting. But just as I suggested in a CCC thread that computer chessplayers not over-rely on search, here it's important that every move be "motivated" -- completely random moves should not be allowed. (That goes for pseudo-random methods, too, of course.) I would sanction a small amount of randomness added to an essentially deterministic procedure. Seeing computers play should prove more fun than playing this game myself (unlike chess). I would be interested in seeing an "anti-computer" strategy, employed by a computer -- another challenge for you. Although "guess which hand" is one name this game is known by, a much more extensive bibliography can be found under "matching pennies". Do a web search on that, and enjoy. There is also organized international interest in "rock-paper-scissors", and programs to play that game, no doubt. -- Walter
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.