Author: Robert Hyatt
Date: 18:04:51 11/10/02
Go up one level in this thread
On November 10, 2002 at 13:12:18, Bob Durrett wrote: >On November 10, 2002 at 13:01:51, Robert Hyatt wrote: > >>On November 10, 2002 at 04:22:16, Sune Fischer wrote: >> >>>On November 09, 2002 at 22:31:34, Robert Hyatt wrote: >>> >>>>On November 09, 2002 at 17:42:07, Russell Reagan wrote: >>>> >>>>>On November 09, 2002 at 17:22:49, Gerd Isenberg wrote: >>>>> >>>>>>The rule proposal is not random, but based on experience. >>>>> >>>>>I'm just saying that the rule proposal does nothing to prevent anyone from >>>>>cheating, and it also acts in a counter-productive manner by preventing some >>>>>people from participating. So, it's a matter of whether you want to add the pro >>>>>with the con, or have neither. >>>>> >>>>>Russell >>>> >>>> >>>>There is no way to _prevent_ cheating. But one significant problem we had >>>>in the past was a programmer using a manual interface while in book, then >>>>switching to automatic after out of book. The excuse "my interface won't work >>>>with a book" is pure nonsense, it just lets the human choose the opening lines >>>>as he wishes. >>>> >>>>If we require kibitzing time, score and PV, it will be very hard for an operator >>>>to make the program play a different move while keeping the scores and PV in a >>>>consistent state. It won't eliminate it, but it will make it harder. >>> >>>What score and PV do you display while in book then? >> >>I display a score of <book> and the PV contains the book move and the >>most popular book response to that move. But the main point is that I >>play that move _instantly_. No manual operator can do that... >> >> >>>There is no output from my engine while in book, I don't see what information >>>could be relevant, perhaps statistics on how often the chosen moves are played, >>>but that is no trivial matter to implement (and I would find it rather boring to >>>be honest, so hopefully no rule about that). >>> >>>-S. >> >> >>I don't see any reason to kibitz or whisper something while in book. When >>two programs are in book for 10-20 moves, who could follow kibitzing at that >>speed anyway? :) > >This may seem a bit radical, but do you expect to have any human observers of >these games? If you do, then let me suggest a counter-intuitive idea: Yes. First, I will be observing the game(s) Crafty plays in. And I will also be observing other games that are interesting. But more importantly, at the previous CCT events, we have had _plenty_ of non-programmers watching and asking questions about "what is it thinking, what is the eval, etc." > >Deliberately make both engines delay exactly one minute between moves while they >are still in opening book. > >OK, there may be a few practical objections. > >But, for the sake of the observers, it would be a nice feature since the >observers would have a chance to think about and discuss the opening moves. Can't disagree, the problem is that we are playing using the ICC clock, and that would mean the programs are getting zinged for time that they are really not using... > >Games between top human GMs, played on ICC, are typically discussed in the >opening phase. Everybody is looking for the "Theoretical Novelty" move. >Sometimes the opening theory is discussed, too. Often, the opening repertoires >or styles of the players become a topic of discussion. Opening moves are >predicted. Etceteras. Depends. Games between some GM players and computers are played almost instantly for the first 10-15-even-20 moves... > >But if there are to be no observers, . . . forget it. > >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.