Keeping the internet neutral?: Tim Wu and Christopher Yoo debate.

Tim Wu is Professor of Law at the Columbia Law School. He is the author of Network Neutrality, Broadband Discrimination (2003) and coauthor, with Jack Goldsmith, of Who Controls the Internet? (Oxford University Press 2006). Christopher S. Yoo is Professor of Law and Director of the Technology and Entertainment Law Program at the Vanderbilt University Law School. He is the author of Beyond Network Neutrality (2005) and Network Neutrality and the Economics of Congestion (2006) and the coauthor, with Daniel F. Spulber, of the forthcoming Networks in Telecommunications: Economics and Law (Cambridge University Press).

Yoo:

During the past year, network neutrality emerged as one of the most controversial issues in Internet policy. Although the details of specific proposals differ, as a general matter, a network neutrality mandate would prohibit network owners from discriminating against particular applications and content providers. Network neutrality played a key role in the debates over communications reform legislation that was pending before the Senate and the House of Representatives, as well as in the Federal Communications Commission's ("FCC") clearance of the SBC-AT&T, Verizon-MCI, and AT&T-BellSouth mergers. (1)

I am not convinced that deviations from network neutrality will necessarily harm consumers and innovation. On the contrary, competition and innovation might be better served if policymakers embraced a "network diversity" principle that would allow different network owners to pursue different approaches to routing traffic.

Simply put, deviations from network neutrality may represent nothing more than network owners' attempts to satisfy the increasingly intense and heterogeneous demands imposed by end-users. (2) The early Internet was dominated by applications such as email and Web browsing, in which delays of half a second were virtually unnoticeable. These are being replaced by newer applications, such as Internet telephony and streaming video, in which such delays can be catastrophic. One obvious solution would be to give a higher priority to traffic associated with time-sensitive applications. Unfortunately, this is precisely the type of discrimination between applications that network neutrality would condemn. (3)

Another interesting innovation is the emergence of content-delivery networks like Akamai, which reportedly serves 15% of the world's Web traffic. Suppose that an end-user in Los Angeles attempted to download a Web page from CNN.com. If CNN.com hosted the content itself, this request would have to travel thousands of miles to the server in CNN's headquarters in Atlanta and back, passing any number of points of congestion along the way. The speed with which the request is filled also depends on the number of other queries being directed at CNN's server. Akamai minimizes delay by caching content at thousands of locations throughout the Internet and routing requests to the server that is the closest and/or the least congested. The catch from the standpoint of network neutrality is that Akamai is a commercial enterprise, which means that those who are willing to pay more get faster service. (4)

Employing different protocols might also provide more competition among network platforms by permitting multiple networks to survive by targeting subsegments of the overall market, in much the same way that specialty stores survive in a world dominated by low-cost, mass-market retailers (or, more properly, given the scale necessary for a telecommunications network to be viable, in the same way that department stores compete by developing strengths in certain types of merchandise and becoming the exclusive distribution outlet for particular product lines). For example, deviating from network neutrality might make it possible for three last-mile networks to coexist: one optimized for traditional Internet applications, such as email and Web site access; a second incorporating security features to facilitate e-commerce; and a third that facilitates time-sensitive applications such as streaming media and Internet telephony. Network neutrality, in contrast, threatens to foreclose this outcome and instead forces networks to compete solely on price and network size--considerations that favor the largest players. (5)

At this point, it is impossible to foresee which architecture will ultimately represent the best approach. When it is impossible to tell whether a practice would promote or hinder competition, the accepted policy response is to permit the practice to go forward until actual harm to consumers can be proven. This restraint provides the room for experimentation upon which normal competitive processes depend. It also shows appropriate humility about our ability to predict the technological future. (6)

Wu:

Network neutrality is a useful way of talking about discrimination policies, on networks or otherwise. Whether it comes to employment, networks, or just about anything else, no one really believes in systems that ban discrimination completely. In employment, for example, you want to be able to fire people who are lousy--to discriminate on the basis of ability. When government chooses who gets to vote, we accept that it can say "no" to twelve-year-olds.

Yet I don't think that the fact that an absolute ban on discrimination would be ridiculous undermines the case for discrimination laws. It's like what nutritionists say about fat: there are good and bad types. And what I think is going on in the network neutrality debate--the useful part of it--is getting a better grip on what amounts to good and bad forms of discrimination on information networks.

Christopher, you've done a good job of suggesting some of the reasons that types of discrimination can be useful on a network, like dealing with congestion problems and offering different types of networks altogether. These are valid points. But sometimes you seem to be arguing that based on a few good examples of discrimination, that there's no such thing as bad discrimination--particularly where a network gatekeeper has market power. That is where we part company.

I'll start with the clearest network example: blocking. So yes, in general, a Bell or cable company has some interest in giving you as broadly useful a network as possible, because then the product is more valuable, and the company can charge more for it. But that interest in neutrality holds true only to a point. If a product being offered over the network--say, Internet voice ("VoIP") for $5 a month--competes with an established revenue source (telephone service, offered at $30 a month), the temptation to block it is strong. It is true that, in theory, the provider might start charging the customer $25 a month extra because the network is now more valuable. But that means taking on the costs of changing business models and establishing new consumer pricing patterns, which companies are loath to do.

I am not sure if you would go so far to suggest that blocking is fine because either companies won't do it or will have good reasons when they do. As to whether they will, we don't have to make guesses, because incumbent providers in the United States and in many countries around the world, including Mexico, have blocked or wanted to block competition from VoIP. The United States Trade Representative's office has an ongoing practice, in fact, of trying to talk to countries and their incumbents about such blocking. They don't call it network neutrality or anything of the sort, but it is the export of network neutrality policies.

What's bad about blocking, then? At an extreme, blocking can keep a better or cheaper product (VoIP) from coming to market at all, and often it can prevent such products from being offered in an effective form. That's a problem, in turn, because if you believe that market entry and innovation are linked to economic growth, we're ultimately talking about such policies hindering the growth rate of the country.

Now I admit blocking is the clearest case where discrimination is bad, and it provides the strongest justification for network neutrality rules. That's what Michael Powell thought too, and that's why he announced such blocking would be illegal. (7) But I also think there's another type of bad discrimination--picking favorites, or choosing one company out of many to favor. I'll explain why in the next post, but I better let you back on to see what you have to say.

Yoo:

To date, the debate has focused primarily on a type of discrimination known as "access tiering," in which network owners charge Web sites and application providers more for premium (i.e., higher speed) service. Access tiering could provide benefits similar to those provided by the emergence of premium mail services like FedEx. Instead of taking three to four days to send a letter from coast to coast, FedEx made it possible to send the same letter overnight. FedEx customers were more than happy to pay more for faster service, since it opened up new ways of doing business that were impossible when everyone paid the same amount for a single class of service.

The same logic applies to the Internet. The Internet is currently dominated by a suite of protocols known as TCP/IP (Transmission Control Protocol/Internet Protocol). TCP/IP has two notable features: first, it routes traffic on a "first come, first served" basis, which provides no guarantees as to how quickly a packet will arrive; second, it also routes traffic on a "best efforts" basis, which provides no guarantees that a packet will ever be delivered. Companies developing applications that depend on guaranteed throughput rates (e.g., 100 Mbps) have indicated that they would willingly pay more to ensure better quality service. This has led leading technologists to point out that TCP/IP is a thirty-year-old technology and that network owners should be permitted to experiment with new...