Correcting Grandmasters’ Analyses in Elementary Endgames
L. Rasmussen · ICGA Journal · 1987
node, and D is the number of drawing moves.Provided that at least one node in the tree has a score of -1, even if due only to the opponent's blunder, a search of the resulting 7-ply tree will provide a greater practical chance of success against an imperfect opponent than will selecting a drawing root move at random.Various refinements to this concept suggest themselves, but are beyond the scope of this note.Although the application of this concept to chess-endgame databases is new, so far as I am aware, the idea of tree search assuming imperfect play by the opponent has been described earlier (e.g., Michie, 1981).In other words, when desperate, it is good tactics to mislead one's opponent.The opponent, if not omniscient, i.e., if a human being rather than a computer database, is therefore classified as being, in some sense, contemptible.The notion of a contempt factor has been broached in literature by Slate and Atkin (1977) for full games.Without any risk (as opposed to its use in full games) it may now be applied in endgames tabulated in a database.References