My general position is fairly neutral, I'll probably use voice within a close circle of friends, who already use voice extensively via Skype. Yet, I'm a bit disappointed in the implementation for a number of reasons. Here's the comment I posted:
It's interesting how proximity-based seems to mimic hard-wired phones. In RL, we live in a world where wireless communication allows us the ability to move about. That's one of the frustrating things about those Vivox phonebooths. We *have to* be there, and no other place. Group IMs and IMs in general, do afford us the luxury of moving around.
It won't be surprising to see Skype, TeamSpeak, or Vent continue to be used, because they exist regardless of where we may be.
The same exists with music streaming... the community over time, had decided to share URLs for people who couldn't get to an event, or had someplace else to be-- even though the SL client and the music streaming parts are separate, music is tied to a parcel. From my understanding, this has been done for purposes of immersion, and vetoed before (even before my time).
The thing I'd encourage everyone to watch is the emergent uses of the voice aspects of SL. It might be used for things far more creative than simple communication, since it is limited in its design.
While writing this, another issue came to mind: when building or camming about, I don't hear proximity-based sounds like typing and effects, especially if I'm zooomed out. Music, however will still play because my avatar is located on a parcel, and that system is not affected by my camera movements.
Will the voice system operate like this or is it true proximity that will be voided by the position of my camera? My thoughts are (if I read this correctly), that no, I'd not be able to hear you talk if I'm cammed out. Time will tell.