For an important chess game, perhaps a world championship, a large number of other players of (nearly) comparable strength are hired to "kibitz", specifically, to play guess the next move.
If one of the actual players is frequently making good moves that are not being guessed by the kibitzers, then we suspect cheating by computer assistance.
The implementation is tricky, as the kibitzers don't get to choose how much time to spend on a move. Should they be forced to guess exactly 1 move, or can they submit a list, or perhaps even give an evaluation of every possible move in a position? How should compensation work to encourage correct guesses and discourage unrealistic play?
Despite the likely exorbitant cost of this anti-cheating measure, it might not work. The actual cheating player might avoid wildly "computer" moves, so even assisted moves will be guessed by some kibitzers.
Perhaps the kibitzer is asked (or induced) to label a degree of speculativeness of guessed moves: an expected value and a variance. If an actual player seems to be playing getting "lucky" too often about moves guessed speculative, then we suspect cheating by computer assistance.
Of course, we need to prevent the kibitzers from cheating as well.
My first thought was some complicated UI for the kibitzers to enter guesses, but a better possible format could simply be a videotaped talk-it-through live-analysis, with actual determination of cheating a difficult subjective analysis of all the videos later. They talk about and demonstrate what moves are possible, what moves might be strong but seem difficult to evaluate, and so forth. Strong chess players can already do this naturally.
Related idea but for a broader audience and different purpose
No comments :
Post a Comment