At a previous job, our standard pairing setup was a shared screen session and some kind of voice connection (landline, VOIP, whatever). It's a bandwidth-efficient setup that works very well. screen might be rather arcane for folks who don't have a *nix background, though.

Too, we found that graphical screen-sharing stuff like VNC or Remote Desktopcan be made to work, but it's more bandwidth-intensive and can be annoyingly laggy depending on how good the network connection between collaborators is. That was a few years ago now, though, so hopefully things have improved since. =)