AbigailII: your opponents play will not take gammon considerations into account.

Is that really true? Do you start playing doolally moves if your opponent has already lost or do you continue playing consistently? If you continue playing consistently, how many times are the moves the same between gammon go and merely winning? It's not zero percent, is it?

It's fun when playing a bot to see if you might get a backgammon but, unfortunately, if the match is already lost then they move randomly and don't seek to avoid the backgammon. A human opponent is very unlikely to do so. They will move out of your home just because they can - because it avoids the virtual backgammon. Yet if they leave the pieces there and lose by a backgammon, it will still tend to be intentional. People don't often do random.

Conversely if a human opponent is playing on, even if they have won, there's a high chance that they will play to see if they can get the gammon, just as the loser is playing to see whether they can avoid it. It's a mini competition above and beyond the game itself. Call it fun, call it foolishness, call it what you like. Bots don't give a damn about such things, they can't understand the point. So, it would seem, is the case for certain humans. The bots, however, don't feel a need to argue that the funsters are wrong!

Even if you avoid the gammon, it does not mean you would have avoided the gammon if the gammon had mattered (because your opponent might have made different moves).

Granted, they might be making different moves. In a bearoff situation that's how likely? But anyway, if they might then they might not, and might not is more likely if you're playing a human of the kind outlined above. And if they might not then the exercise is not fruitless.