It's often said by world-class Backgammon players that the best way to improve one's play is to analyse one's games with a computer program such as gnubg, JellyFish or Snowie.

gnubg has a "tutor mode" that evaluates the quality of the player's proposed move, to know before making a move if this one is bad. I assume the same exists for chess programs (hence the tag).

Alternatively, it's also possible to do all the analysis at the end and walk through our moves to look at the errors. My question is whether one of the two methods is better.

A third way I often use is to play normally and ask the computer to analyse the move when I'm unsure of the best answer. I do this just after my thoughts, but without following the computer's advice (to stay motivated).

Do you know of any public statement from top Backgammon or Chess players, or a study regarding the best way to improve oneself by playing against a program, using these or other methods?

  • What are you really looking for? The best way to play against the computer? If playing against the computer will make you better? Quotes from top players? For backgammon or chess? Nov 9, 2011 at 17:28
  • I think the OP is asking for an authoritative source of advice for a systematic approach to playing chess or backgammon against a computer, that most efficiently allows you to learn and improve. Nov 9, 2011 at 22:55
  • @ire_and_curses, yes, it's that. The best way to use a computer to improve. Winning against the computer is not the goal itself at all.
    – Clement J.
    Nov 10, 2011 at 12:16

3 Answers 3


I have no sources to cite, but I think it depends a lot on your own learning style. You will certainly not learn well if the computer just tells you the best play and you follow it, but if you can carefully think through your move before being told by the computer if it's bad, and if the computer has a better move you can reason out why it's better, immediate feedback is fine. However, if you find yourself getting impatient and cutting back on your thinking time, perhaps you should wait until the end.

I do Go problems on the computer a fair bit. With Go problems, usually the first move is the key, and often is the hardest move to find. Some programs will tell you immediately if the first move you've made is correct, but I much prefer the ones that will let me play out a bad sequence without telling me it's wrong until the end. That way I see exactly why it doesn't work, and it reduces the temptation to just try something that looks promising to see if it's right.

  • Thanks shujaa. Maybe there isn't a way that is the best for every one. I asked this question to know if researcher in learning (or top player, it's similar) have studied it. Maybe not... Of course, it's important to think and search before asking the computer.
    – Clement J.
    Nov 10, 2011 at 12:20

For chess there are many ways to use both chess playing programs and database programs:

  • Play over your own games (you should always record your games, even skittles/fun games, to see where you need to improve) in "infinite analysis" mode (where the computer doesn't make moves, it just evaluates the position and finds the best sequence of moves for both sides) to find moves where you (or your opponent) made tactical mistakes. When the score changes by more than 1 pawn of value that is usually considered a mistake. More than 2 or 3 pawns of value change is like loosing a bishop/knight, etc.

  • I find using a computer program (like Fritz) to analyze my games and to go over master games much more efficient than doing so on a board because you can easily try out different variations and go back to the main line without having to physically move the pieces around on the board.

  • Practice technique (one of my favorite uses of a computer): play out theoretically winning positions against the computer to make sure you can always win them. (Sometimes programs make very silly moves when they're losing by a lot and there aren't many pieces left on the board, but if you set them up to use "endgame tablebases" then they make moves which put up the most resistance). Practice the basic checkmates until you can do them on auto-pilot. Practice pawn endings, etc. Play out master games where one player resigned to see if you can win the 'winning' side.

  • Set up positions from books (especially endgame studies work well), and play them out against the computer, to see if you can win won positions, and draw difficult positions.

  • Use database programs to research the openings you like, to see what kind of games arise from various opening move options.

  • Not only master games, but amateur games from databases are very instructive to go over, to see if you can spot the mistakes and how you would take advantage of them. You can find large numbers of amateur games on sites like: http://team4545league.org/

  • 1
    +1, I really like the idea of playing out master games where one player resigned. Nov 18, 2011 at 17:23

You may also want to look at the following two books from Gambit Publications:

  1. How to Use Computers To Improve Your Chess by Christian Kongsted.

    The second half is specifically on how to use chess computers to improve our play.

  2. Secrets of Practical Chess by John Nunn

    The new edition has an expanded chapter on chess computers.

You can download PDF samples from their respective pages.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .