Hello,

Im getting my ass kicked by this computer to about 20 pips diffrence does this mean im a beginer?I only get out 2 or 3 pieces out.

If you're losing regularly to GNU or to Extreme Gammon, that's nothing to be alarmed about. Both are very strong programs. Even good players can lose to them regularly.

But yes, I assume you are a beginner. Have you read any books on backgammon? Do you play in tournaments? Do you play at a club? Do you enjoy solving backgammon puzzles in your spare time?

No ive only played agains people and ive never read any backgammon books.I played on free sites and managed to win most of time.That is why i tought i was intermediate ive also won 6 out of 10 games on backgammon free rolls on play 65.

GNU backgammon is stronger than any human player. Everyone loses to it in the long run. It is a good learning tool: I suggest you use the hint button often.

You may want to switch to manual dice, and roll some real dice on your desk and input it. I find I lose much more if I use the computer generated dice. I'm not accusing bg software makers of cheating, but it is odd that my win ratio goes up when rolling real dice.

This can be a real effect, but has nothing to do with the dice. When playing against a bot, it is easy to get into a rhythm of playing very fast, which promotes errors. Rolling dice manually and entering the rolls into the computer takes time. This forces the player to slow down, and as a result sometimes find better moves.

That is most likely true. Even using random.org's app on my phone, my win ratio goes up. Though it does take an extra couple of seconds, like you said, to go back and forth between my phone app or real dice, and the pc.

gnubg, extreme gammon are training devices for people who need to improve the quality of their play (especially highly ranked professionals). Why do you think they could possibly feature unrealistic dice behaviour?

I understand that. I'm just saying my win ratio goes up, some times dramatically, when rolling real dice, or using a separate dice app.

Everyone's win rate goes up when inputting real dice to a bot, because the time it takes slows down your play, which gives you more of a chance to spot and avoid an error.

Try playing against the bot when it rolls the dice and keep track of your error rate. Then input manual dice for a while and keep track of your error rate. You'll probably find that your error rate is also going down when using manual dice. Q.E.D.

Getting back to the OP, I recommend activating "Tutor" mode in GnuBg. That way, the computer will interrupt you when it thinks you have made a significant error. I often practice this way against GnuBg. Because my errors are usually flagged and corrected, my win rate goes up.

The threshold I use is 0.02. If my errors are less than this, the GnuBg tutor does not correct me. For larger errors, it does.

Many times, of course, GnuBg is wrong. That's because I have the tutor set to use GnuBg's "grandmaster" setting for checker and cube decisions. ("World class" would also be a good setting.) A rollout will occasionally overturn a recommendation made by the tutor. The truncated rollout I use when my intuition tells me to check the tutor is described here. It runs very quickly.

I have not kept records, but with an assist from the tutor, I think I win slightly more often than I lose against GnuBg. That's because of the short rollouts I make in the middle of a game. When I think my move is better than the tutor's, I roll it out. The end result is that I play about as good as GnuBg when I follow the tutor, and I play better when I add in the plays that I have rolled out.