Please turn on javascript in your browser to play chess.
Only Chess Forum

Only Chess Forum

  1. 30 Jan '08 20:12 / 1 edit
    For a while I have been able to avoid more and more losing due to tactical mistakes. But of course I still lose games because of not so obvious mistakes.

    For analysis I will depend on chess engines. For me in certain positional positions it is hard to understand why the best move is indeed the best move, even when it's pointed out to me. And it's even harder to understand why the move or variation the chess engine recommends is good. Especially because I doubt the chess engine. Is a move that is 20 cents better according to the chess engine actually better?

    I have been using Rybka 1.0 beta 32 bit. I have noticed that Crafty disagrees strongly with Rybka's analysis of certain positions, of course Crafty being a weaker engine. I have an AMD Athlon 1600+ processor.

    Also, I don't really know which settings to use or how deep to calculate a position to get a good judgment. Should I use another chess engine as a second opinion? Which one? I know Rybka should be strong positional-wise.

    Any tips here?
  2. 30 Jan '08 20:26 / 1 edit
    Originally posted by Prometheus4096
    For a while I have been able to avoid more and more losing due to tactical mistakes. But of course I still lose games because of not so obvious mistakes.

    For analysis I will depend on chess engines. For me in certain positional positions it is hard to understand why the best move is indeed the best move, even when it's pointed out to me. And it's e second opinion? Which one? I know Rybka should be strong positional-wise.

    Any tips here?
    I don't understand the popularity of crafty, it's by far weaker than some other free engines.

    FWIW, the last versions of fruit and toga (which are free) are actually stronger than rybka 1. just google them.
  3. 30 Jan '08 20:39 / 1 edit
    I'm not sure what your playing strength is, but you may not be at a point where you really need to distinguish between two moves that are only different by 0.20 in evaluation. For moderately-ranked players it should be sufficient to concentrate only on the 1-3+ point blunders--if you could eliminate those with a high degree of consistency you'd improve greatly. At that point, two moves that are 0.20 apart might as well be the same.

    On the other hand, say your moves are off by 0.20 but no more on almost every move, and then you go on to lose. Can you at least jump ten moves ahead, when the evaluation has fallen from, say, 0.0 to -2.0, and see what is bad about the position then? Then maybe you can work backward and see why you should have tried to avoid that position.

    If you lost the game you must have wound up way behind in the evaluation at some point. Even if the individual errors are very small, somehow you should be able to see how they've led you to a bad position.
  4. 30 Jan '08 21:37
    Originally posted by Prometheus4096
    For a while I have been able to avoid more and more losing due to tactical mistakes. But of course I still lose games because of not so obvious mistakes.

    For analysis I will depend on chess engines. For me in certain positional positions it is hard to understand why the best move is indeed the best move, even when it's pointed out to me. And it's e ...[text shortened]... second opinion? Which one? I know Rybka should be strong positional-wise.

    Any tips here?
    You didn't mention which GUI you use to run the Rybka engine in. I usually use Fritz 8 for blunder checking my otb games, only because I already have the Fritz 8 program. For simply blunder checking your tactics, I don't think engine strength is terribly important, as long as it's a fairly recent engine.

    While the free Rybka is very strong, it has several drawbacks for checking tactics. Its endgame tends to be a little buggy. Also, it can only display one principle variation. (That is, it doesn't have multiPV capability.) And finally, I don't think it has capability of using Nalimov endgame tablebases, which could be useful with not many pieces left on the board.

    If you don't need to use tablebases, I think Toga II 1.3.1 would be a good choice for a UCI engine for use with the Arena or Fritz GUI. (Note that it's a rar archive file.)

    http://www.superchessengine.com/Toga131.rar

    I should mention that the latest (and strongest) Toga version is Toga II 1.4 beta 5c, but the programmer hasn't yet implemented multiPV capability in this beta version.