Overhaul the backhaul?

I just re-read "Overhaul the backhaul" by Gaby Junowicz and was thinking about how bad the problem of wireless backhaul in cellular/data networks is going to be.  I think what the article probably leaves out is the the effect on congestion and serivce associated with usage.  Increasing the bitrate for a data service frees up packet data spectrum only if usage becomes constant.  The promise of HSDPA seems to be high bandwidth to the user device. 

I remember when I got to college, and I was on the fastest broadband network I had until that point ever encountered.  You better believe I dowloaded eveything in sight and expected the netowrk to keep up!  This is anecdotal, but I think the data back me up: when people get higher badwidth connectivity their usage goes up.  So perhaps I am missing something.  I promise to find the data and report back- how much do user habits change compared to increased available bandwidth at any given time, and how many additional users could be served if the higher-bandwidth service does indeed shave the MhZ * users total?