Information Policy recently posted the AIE-Brookings manifesto "Economists' Statement on Network Neutrality Policy." They claim that differential pricing is more efficient, provides an incentive to increase bandwidth (overall), and that any "abuses" should best be dealt with through anti-trust lawsuits. IP also links to another AIE-B paper ("Network Neutrality") that essentially repeats these points.
It seems to me (and no doubt others) that there are serious flaws in their logic. As Cory Doctorow points out (mp3), folks with a lot of experience providing ultrahigh-speed Internet discovered that it's far cheaper to throw bandwidth at congestion than to bias content delivery. Also, the marginal cost of greater bandwidth is, well - marginal. Once lines are laid (huge dormant capacity is already in the ground), increase is more about switch design than fiber.
Finally, while I understand economists' affinity for "institutional tidiness," shunting abuse prevention to anti-trust regulation and litigation seems both naive and irresponsible. Naive, because "anti-trust" law becomes effectively moot in the context of global-scale markets (i.e., local concentration is absolved in the face of multiple producers worldwide). It is irresponsible, not only because they ignore wealth effects on the ability of citizens to bring claims against far more well-resourced telecoms, but also because they seem to ignore the potential net costs of such litigation.
Police powers based entirely on liability seems more like a jobs bill for lawyers than anything to do with justice, or even that holiest-of-holies, efficiency.
Addendum: Read E W Felten's (Princeton) "Nuts and Bolts of Network Neutrality" for a readable and well-balanced explanation.