I haven't done much in-person pairing, but I think the main difference is not having to share a keyboard. Each person can use a system they're comfortable with; no fuss about dvorak vs qwerty, Kinesis vs Natural 4000, standing desk vs sitting, or Ubuntu vs Macosecks. It does mean you need to use auditory cues to indicate who is in control of a given session. The biggest drawback is probably that you can't share a web browser over SSH, but since we are doing backend coding it doesn't affect us.
For me it's been a huge win. We do also have week-long meetings a few times a year to bring the whole team together in-person to promote the team's cohesion. But working from home is great; as long as I'm able to switch things up occasionally by working from a coffee shop a couple times a week, I'm happy.