Wednesday, December 08, 2004

Beyond Deep Blue

Eric Meyer, a self-avowed non-fan of chess, has a few suggestions for how to improve chess computer user interfaces. To summarize, he has three major recommendations:
  1. The computer should not move instantaneously so as to appear more human and methodical in its "thought" process

    Eric's perspective is clearly limited by the one program he is using. Most commercial chess engines include time controls for the game and the engine will typically take advantage of that time to search deeper in its tree to find the best move. Contrary to what he states, engines do not complete their full analysis in seconds, but instead will continue to search further as time permits.

    Update (12/10/04): I got this one wrong. Eric's actual point is that the computer should evaluate for the amount of time appropriate to the level/strength setting selected but then wait for a period of time before making the selected move. He believes that for beginning players this will be less intimidating than programs which move instantaneously. Interesting...

  2. The computer should show all the squares which it currently attacks so the player is less likely to move a piece to those squares.

    This certainly seems like a useful training mode for beginners to help them visualize the opponent's threats. I know that Fritz has several training modes including one that shows all the moves a piece can make, but I'm not sure if showing attacked squares is one. The idea of tinting the squares to show how many pieces are attacking is quite an interesting idea. Now if I could just get my OTB opponents to color the board for me after each move, I'd be all set.

  3. The computer should play chess variants.

    When I first started reading this section, I thought Eric might be thinking about Fischer-Random, Suicide, CrazyHouse, etc. In fact, there are specialized engines that play these variants. However, he had something completely different in mind (which left me rather amused):

    My favorite was Nuclear Chess. In that one, any piece could instead of moving choose to self-detonate, destroying itself and any pieces in adjacent squares....And then there was Thermonuclear Chess, where any piece could detonate after making a normal move.

nuclear explosion

No comments: