you can convince someone much more readily if you talk to them privately instead of before an audience

I suspect the optimal size for a discussion where minds are actually changed is not two people but three. I find a group of three is less prone to diversions from the factual to the personal level than a group of two, yet remains intimate enough to minimize the loss of face that disincentivizes changes of opinion.

This hypothesis seems difficult to test and become sure about, but maybe some of you would like to entertain the notion and play with it sometime.