Too bad AT&T doesn't have a femtocell product that can route iPhone traffic over local internet connections. Oh, wait, they do! It's in testing with their own employees. See: http://www.dslreports.com/shownews/ATT-Femtocells-in-2009-99... Not deploying a few to SXSW would seem a major blunder and lost opportunity -- unless the units are currently so awful they'd be even worse then the current complaints.
AFAIK a femtocell can support fewer than a half-dozen phones; you'd need a ton of them to handle the Austin Convention Center. It's not even clear that such density is possible from an RF engineering perspective. OTOH, T-Mobile is fine and the wi-fi works well (which is impressive with ~3,000 simultaneous clients).
If the femtocells are to work in an urban environment, they'll have to work with many to a city block -- or even single apartment building. That is, about the same density as wifi access points. I can see 7 SSIDs right now from my kitchen table; from the roof of my building I can see 20-something. (Of course only a fraction of these are usefully connectable.)
So why not a femtocell in the ACC everywhere there's a wifi access point? That would seem to mirror the best case for deployment once they're for sale to home users.
Or at least a half-dozen in the corners of the venue? So at least a few people can report back about islands of good coverage?