Random Posts

Friday, July 9, 2010

TobyTal Chess Engine

I saw a small ad in the last issue of Chess Life listing the website for this “new” chess engine. I visited the site and they claim it plays at 3200 and it costs $59.00. They also advise they market a vast database collections, specialized training programs and cutting edge resources for coaches & teachers as well.

The developer also makes a claim that in 4-6 years there will be no more chess engines as chess will be solved and the moves will simply be looked up in tables. I won’t bother with commenting on this other than to say it seems like a nonsense statement considering the enormous size of 6-man tablebases, let alone a 32-man tablebase.

The developer claims that his engine easily defeats Rybka 3 and gives some sample games. I’m not sure what conditions the games were played under. Nor could I find a decent review of this engine when I Googled it. I also went to the SSDF rating list to see how TobyTal fares against other engines…it’s not listed. One would think that if the engine is as good as the author claims he would have entered it in one of their competitions.

Personally, I wouldn’t bother with it until there’s more proof available than the unsubstantiated claims of the developer. If anybody has any more information on this engine, please post it!

6 comments:

  1. It is in the Fire/Ivanhoe/Robolito/Ipp family which one can get for free.

    The Deep Tactics GUI is nice though.

    Some more info

    http://www.open-chess.org/viewforum.php?f=7&sid=d556865b29ebe2f64aa4bce9df58a6f5

    ReplyDelete
  2. P.S. 2: Strength of an engine (which compares engine2engine) is not always the answer to analyzing your games. Use a cross-section of engines that have unique programming. Some will find brilliant moves at times while others do not. Run an analysis of your game, for example, three multiple times, using a variety of engines such as Rybka, Ivanhoe, Fritz, ProDeo, Stockfish, Ktulu (and hundreds more). There seems to be the crafty family, Robolito family, Stockfish, and Fruit inspired family engines. So using the same engines from the same family can often give redundant analysis (useful), even with a slightly different ELO. Myself I usually run Hiarcs for positional, Spark or Rybka for tactical, and ProDeo for finding brilliant moves

    ReplyDelete
  3. Using multiple engines is good advice and agrees with top level cc players who advise using more than one engine from different families.

    This is where I ran in to problems playing in engine tournaments on IECG. I am not good enough to evaluate which engine output is the most accurate for a given position.

    Also it proves my point that playing cc at the titled level where engine use is accepable, even expected, involves more than just buying an engine. In fact it requires a measure of skill just to get past the gaggle of engine users lurking at just below that level!

    ReplyDelete
  4. Yeah, chess engines have been a blessing and a curse. And I do not always know which one is best, unless a unique move comes out and I let the computer finish the game to see what the outcome is.

    As for playing with engines it turns some games into just engine matches so you might as well just use the ones that succeed that way like Rybka, Ivanhoe, Houdini, etc etc.

    But for human games I would trust the evaluation of engines such as Hiarcs, Fritz, Shredder, Spark and ProDeo

    While extra strong, some of those engines (Like Ivanhoe) are not as "polished" as these since they have been long term projects by the programmers. Even strong programs like Stockfish have gaps in their endgame knowledge. If I had to pick just one, my most trusted is Shredder

    ReplyDelete
  5. P.S. My guess is that most internet players using engines are using Houdini. I do not often see TobyTal as I am sure this derivative is not as strong as they try and claim, and the ones being worked on like Ivanhoe are free. For internet play might as well just run one engine (Like Houdini or Ivanhoe with as many CPU's as you have) and for studying human games use the cross-section of engines

    ReplyDelete