It required near forty years of concerted effort by computer scientists and chess players to create code that could play a reasonable game of chess. Once that was accomplished, it became terribly difficult to weaken these programs (and minimal resources have been given to the task).
At no point in the history of AI have computers succeeded in playing like humans, although some Hiarcs users claim there is a humanlike quality to it compared to Fritz, and Kasparov thought that Deep Blue was suspiciously humanlike.
The weakest Chessmaster personalities are programmed to give away their queen in the first few moves. Those above 1800 or so, do not make egregious tactical blunders as often, but still lack positional sense. They are good training for play against humans only over many games with many personalities, as then spotting different sorts of weaknesses and errors becomes second nature. Individual games against handicapped software do not resemble games against humans.
Originally posted by ivan2908I noticed that also with CM. It plays well for the first 2/3 of the game then blunders.
I noticed that when I play against the Chessmaster 10th edition personalites (1400- 1600 rating), it is very difficult to play tactical against them. I feel often it is nearly impossible to set up a fork, discovery, skewer, CMX anticipates it almost every time, so I can't beat it with my tactics. I ussualy win, but only because at certain point of game (if ...[text shortened]... aying aganist computers? I think playing with human opponents is far more fun and instructive.
Yes but if it is you who make the mistake, you will certainly loose, so the main training is to keep a game without major blunders, since a human could not take advantage of them in the proper way, and you could be doing the same mistake without noticing.
I have corrected a few things that I see chessmaster spoting in my game, and a lot of humans I have played didn't noticed. so in that way it could be good for training but I agree that it is not the same.