Network neutrality - why it matters, and how do we fix it?

My friend Tim Wu (presently guestblogging at Lessig) has a great essay on why "network neutrality" matters -- that is, why it's a bad for the Internet for the cable companies and Bells to charge money for "priority" delivery of some companies' packets. A non-neutral ISP could guarantee better delivery to Yahoo than Google, for example.

I agree with net-neutralists -- the Bells' and cable-companies' plans to put toll roads on the Internet's pipes are evil incarnate, and the Bells' arguments that they're currently delivering packets for Google "for free" are steaming BS. Google pays for its bandwidth, and pays handsomely.

That said, I remain skeptical of the idea that this is a problem with a regulatory solution. The FCC is slow, often captured, and breathtakingly dumb about technology (this is the agency that passed the initial Broadcast Flag rule, after all). Asking them to write a set of rules describing "neutrality" and then enforce them seems like a recipe for trouble to me.

For example, say that your university maintained a pool of DSL lines for students, and a data-center for courseware, and created dedicated connections between them -- is that "neutral?" What about Akamai: they put servers in ISPs' NOCs around the world, and then sell mirror-space on those servers to people who want optimized delivery to those ISPs' customers. Is that "neutral?" How will you tell, from the outside, whether an ISP is delivering slow packets to you because it's "non-neutral" as opposed to badly managed, overloaded, or staggering from some kind of net-quake?

At the end of the day, we're talking about a set of rules governing networking configurations. Network configurations aren't something that we have ongoing, permanent consensus on -- rather, they're a hodgepodge of each admin's idea of the best way of provisioning her network for her customers and users. Trying to write a regulation -- or even comprehensive best practices -- for a "neutral" network is going to be really hard. Getting it wrong could mean screwing things up even worse -- imagine if the FCC could be convinced to create a neutrality rule that preserved Akamai's business-model but punished their innovative competitors.

The plans to put toll-roads on the Net are terrible and we need to do something about them. I just don't know what we should do.

This is the basic case for network neutrality–to prevent centralized control over the future of the Internet. But there's a long-standing rebuttal that goes like this: A broadband company already has incentives to make the network neutral, because it's a better network that way. If AT&T makes money on an exclusive deal, they'll lose it somewhere else. Whatever money AT&T earns by prioritizing Google rather than Yahoo!, it will lose by making its product–broadband service–less attractive to consumers. By this logic, regulating the Bells is a waste of time. AT&T and Verizon also say that they must be free to discriminate to justify their investments in building networks. If you don't let us discriminate, they say, we won't build.

It's true that the Bells might make extra cash by discriminating. But AT&T can extract cash in other ways, too, like charging its customers higher prices. I believe that it's better to have consumers pay more for service than to have AT&T picking and choosing winners on the network. Both are a cost to the economy, but the latter is a double cost. It creates costs that are passed on to consumers anyhow, and it also distorts competition between eBay, Yahoo!, and the like. Building networks at the expense of network applications has a logic O. Henry would enjoy, for it's akin to selling a painting in order to buy a better frame.