Interesting and potentially far more menainful than elo based ratings.
en.chessbase.com/post/ranking-ch … heir-moves
i beg to differ. if you’re going to use a computer to compare moves in a sterile setting, fine, but in the heat of battle, anything can happen. it’s about competition. mischa tal probably wouldn’t even rank in the top 100 with that method.
Artistry, imagination, and daring are difficult traits for AI to evaluate. Chess is not just pattern recognition or calculation or use of memory. The players mentioned all have styles that their opponents have found difficult to cope with because each has an approach based on experience and a conceptual understanding of chess which is sometimes out of the ordinary. Since we cannot really have some of the greats play against Stockfish or some other program running on a supercomputer, it is really difficult to say how they would stack up. For example, during his time Emanuel Lasker baffled many of his opponents with play that they struggled to understand. In a real game, one of the objects is to make your opponent so uncomfortable that he is bound to make errors.
Second best moves are mistaken for being inaccurate when they may be another choice in a position where there are many options, but the “right” move is difficult to discern and calculate. Players often make choices based on their style, temperament, or philosophy of chess. What is rational, logical, and perfect to one player or to a computer, is ugly and irrational to another. That is why we play the game, to test our ideas and abilities to think.
Much of what you say is true. Still, a supercomputer running Stockfish will distroy any GM presently walking the Earth. What interests me is that this is a quantifiable way of measuring ability that does not depend on win, loss or draw. Winning may not be everything or the best way to rate players, particularly if this system can also accurately predict who is likely to win and encounter.
The study looks at GM games played in the heat of battle using the number crunching logic of Stockfish runnng on a super computer. In the heat of battle anything can indeed happen. That’s why the study gives percentages of best moves. Some greats made more best moves than others in thier best years. Percentage of best moves doesn’t always equal wins. All it takes is one blunder after making 35 best moves…Elo is just one numbers game. This is an interesting and potentially more useful number game.
It is also possible to say that the person that plays “best” moves all of the time is cheating. There is not that much difference in the methodology of the article and how cheaters’ games are examined. Each looks at the quality and consistency of moves produced.
True. The difference is that the analysis for the rating is for the player over time and would not show the level of correct moves that the cheater would produce in one game.