Subsidization really makes sense because the carrier is using the customer’s bandwidth to provide 3G access.  Charging a monthly fee, in this case $20/month, for the customer to use their own bandwidth to support AT&T’s customers seems like just a bit of a dick move.   Maybe AT&T should be paying people to use their microcells?

Sure the box costs money to build and support, but AT&T could certainly use the access bandwidth and certainly the good PR.  Their efforts so far haven’t yielded any fruit for iPhone users, especially those in big cities.  Plus, for the densely populated areas like New York and San Francisco, where their networks are failing the worst, they could add a block of coverage with a few microcell spots.  Immediately.  Not at some point in the future.  Maybe.  Also, as an aside, why wouldn’t AT&T just be working on a VoIP solution that runs in the background and switches on when the signal strength isn’t so hot and a Wifi signal exists? Almost every American has Wifi in their home and workplace – where they spend most of their time.  This would be a better long term solution.