I'd say I can get behind pair-programming to a point... it helps get new ideas, different view points and leads to rubber-duck-programming like effects.
I'm not sure I'm a fan of Mob programming... as others mention, over a phone/video conference, it seems inadequate. Turns more into an ad-hoc video lesson by the stronger personalities (for better or worse - I happen to be one of those in my small group at work).
Do you have problems with both? 2 vs 3+? or is it one or the other?