I was playing backgammon against my Palm Pilot this morning on the bus on the way into work. I play against it in advanced mode and usually end up splitting the games against the computer.
Recently, I’ve started playing a little more agressively than I used to. The result has been that while I still manage to keep even overall, there is a wider swing in the point difference between myself and my Palm opponent.
That being said, I know that the general rule with regard to accepting the cube is to accept it if you have a 25% chance of winning. However, I still sometimes have trouble deciding when it is the right time to offer the cube.
The example at hand was as follows:
It was the very end of the game. I had a man on my six point and a man on my five point. The Palm had two men on the five point. It was (obviously, if I’m thinking about giving the cube) my roll. I decided to give the cube, even though the chances of my taking both men off this turn were slim, and the Palm had a slightly better chance on his (its?) next roll. I already knew that the Palm would accept (intuition based on many games against this opponent) if I offered. What’s the right call?
Result: I offered the cube, and promptly threw double 4s to win.
I think standard cube protocol is to offer it only if you have an 8-10% pip count advantage and you don’t have appreciable gammon chances. Of course, elements of reading y our opponent and position come into this; you may want to offer the cube earlier or later depending on how you think he will react, and a pip count lead with blots everywhere and a terrible home setup isn’t really an advantage.
Your cube offer in the above situation was essentially just a decision to play for two points instead of one; there is no way the computer would drop, and both players winning chances were virtually identical. I don’t think I would have doubled in your situation unless there was more going on you didn’t mention (e.g., game count in the match mandated it, etc.).
Some backgammon programs are very unsophisticated in the way they accept (and even issue) doubles. Some of them seem to go only by the pip count. Imagine the computer with five stones on the 1-point, while I have one stone on the 9 or 10. (An unusual situation, but possible.) The computer will probably take three turns to clear his board, and I’ll be off in two. If it’s my turn, I’d certainly offer a double. I’ve seen cases like this where the cube gets all the way to 64. It’s rare, but it happens.
It’s sounds like your program may be doing something like this. In the situation you described, it was ahead on pips. (Although it should make some allowance that you get to roll first.) Try it sometime when you have two stones left, 1-1 or 1-2, and the computer has a single stone on the 1-point. Any roll will clear your board, you can’t possibly lose; but if the programmer took the easy way on figuring doubles, the computer might accept.
If you find a weakness in your opponent’s game, exploit it.
I would say you did the right thing, zev. The computer can win on 6 out of 36 possible rolls: 3-3, 4-4, 5-5, 5-6, 6-5, 6-6. Of course, you could still lose by not getting a total of 11 on your 2 rolls.
No, this one is pretty good. When it presents me with the cube, it’s usually in a position where I probably would seriously considered doing so if I were in its position. It’s well advanced from the standard “count the pips” approach.
Myself, I figure the doubling cube has two purposes: a) to hasten the end of a won game, or b) to collect more money (or more points if in a tournament).
On point #a, I never use the doubling cube so late in the game. In fact, I’d never even considered such late usage before reading this post, although it does make for some interesting possibilities for point #b.
On point #b, I’m assuming you were playing for no money, and also were playing single games (as opposed to “first player to 7 points” or such). And therefore I’m not sure the cube helped. Although now that I think of it, in non-money single games, there’s really no reason to ever offer or accept the cube, but that’s another story.