Author: Don Dailey
Date: 14:55:04 06/10/98
Go up one level in this thread
On June 10, 1998 at 05:58:53, Carlos Adan Bonilla wrote: >Hi! > >I'm developing a new computer chess program and I have some questions >that I will post here when having enought time to. > >Actually, I'm developing a Suicide Chess program, but my questions may >apply also to regular chess... > >Just to start my series of questions: > >I wonder if it is ethical to develop a feature in my program in order to >let the computer know if its opponent is a human or another computer >program. > >I mean, a button to allow the operator to switch from one position to >other position (human or computer) in order to play a different kind of >search for each case. If the operator selects the "HUMAN OPPONENT MODE" >the algorythm will be different from the normal alpha-beta search and >also another differences... > >The main point of that is to get advantage of the knowledge that its >opponent will eventually play bad moves. Thus, the computer may select a >complicated line rather than an "easy-to-see-for-humans" line, even if >that line is not so good. >Also, in lost positions, the computer will select the lines in which >there is an opponent move that is a mistake and could turn the game back >to a drawing position (or winning). >You can also take that information into account for having a different >time managing schedule, not letting the human player think vary much >when he has few seconds in his clock and its the computer's turn to >move. So what you're really talking about is opponent modelling techniques but I believe most of these are going to get you into trouble. I have found that in my own personal chess games, I always do best when I consider that my opponent can see everthing I see, even when much weaker. I think the same thing applies to computers. One idea though that may have some merit was suggested years ago, I think by Hans Berliner but don't quote me on this. At the point you see a huge drop in score to a seriously losing position, play the move you WOULD HAVE played before you saw the score drop. There is some chance this is good because computers usually play a horribly desparate move when they realize they are lost. Often the move makes it obvious to the opponent that he is winning when there is at least a chance the opponent would not have noticed it otherwise. However if this is an improvement, I doubt it's more than just a very minor one, after all, your position is dead lost in either case. - Don > The ethical questions appears when I realise that the operator has to >decide whether the computer's opponent is human or not, because that's a >choice that the program is not able to do. >I could also develop a feature in the code to allow the machine to guess >the type of opponent by comparing the moves selected by its opponents >with its searching evaluations... but that has to be made when the >opening has ended and maybe too late... ??? >However, a human gets that info before he starts to play, so , why don't >let the operator to supply that info to the machine at the beginning? > >What do you think about all these thoughts? > > >- Carlos Adan Bonilla >- CarlosAB on FICS. >- Developing MoiSui(C) on FICS. >- I love Suicide Chess!!!! >- "I eat your bishop and count 20 moves."
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.