Single Comment

Greg, I always add 'preliminary estimate' after my evaluations because it is a continuous process. I recently discovered that I had probably overestimated my Ladon, so I modified its movement. When playtesting I use different criteria. In computer-computer games I can study how much the piece involves itself in the game. If it makes about the same amount of moves like the other pieces, then it seems to be a tactically useful piece, and its value could be on a par with the other pieces. Another method is to use different armies. On one side there are only traditional pieces, and on the other two traditional pieces less, but instead new pieces. While we know the values of the traditional pieces that have been removed, we can assume that the values of the new pieces are the same if the games tend toward equality. I have, using this method, often discovered that my new pieces could not compete with the traditional pieces, so I had to upgrade their powers, while I have restricted myself to the traditional piece context. As the traditional Western piece values are today perfectly established, its possible to tweak Zillions according to this. Zillions sligtly underestimates the knight, the rook, and the queen. This makes it exchange a queen for rook plus bishop sometimes, and it often avoids exchanging the bishop for a knight when this is advantageous. In most of my latest implementations I have corrected this. If one makes these tweakings, then Zillions is an excellent help when evaluating new pieces. /Mats