Internet Architecture and Innovation.

• Author: Candeub, Adam
• Position: Book review

Barbara van Schewick's Internet Architecture and Innovation surveys broad areas of computer engineering and economic theory to argue that some types of network neutrality regulation may be necessary to optimize innovation. My central critique of her argument is its use of economic theory to bolster one side of a highly politicized debate, rather than using economic analysis to clarify that debate. Van Schcwick relics on an impressive array of economic approaches, but fails to acknowledge their ambiguity. Her argument strings together a succession of questionable economic generalizations, thereby greatly weakening her conclusions.

Van Schewick is not alone in using economics in this way. Too many law professors rely on theoretical models but ignore their limiting assumptions, failing to sort through the massive ambiguity inherent in their application. A close examination of van Schewick's argument, therefore, leads to general recommendations for legal interdisciplinary research methods.

On December 23, 2010, the FCC released its landmark "network neutrality" Report and Order ("Order"). (1) It prohibits a dominant Internet service provider, such as Verizon or Comcast, from discriminating in favor of traffic or content that it owns or with which it is affiliated. (2) For example, the Order prohibits Comcast (or any other broadband provider) from blocking or degrading competitor content, such as Netflix downloads. (3)

Washington, D.C. policy pandemonium greeted the Order, releasing the bile accumulated over the past decade of ideological debate over Internet openness. A Washington Post headline blared, "FCC Approves Net-Neutrality Rules; Criticism Is Immediate." (4) On the right, House Energy and Commerce Committee Chair, Representative Fred Upton, vowed to "use every resource available ... to strike down the FCC's brazen effort to regulate the Internet." (5) Senator Kay Bailey Hutchison promised to slash funding to the FCC, (6) and on the Wall Street Journal's opinion page, John Fund labeled the move "a coup" fomented by a cabal of communications communists. (7)

On the left, the reaction was also hostile, with the consensus that the Order had little substance and failed to deliver a meaningful network neutrality regulation. (8) Given that President Obama's campaign promised to enact strong network protection, cries of betrayal echoed through the Internet activist community. (9)

The beating of breasts and gnashing of teeth seemed excessive, even bizarre, given the esoteric regulation at issue. After all, as insiders and telecommunication specialists know, the Order itself has big loopholes, such as exempting wireless, rendering it close to toothless. (10) More significantly, it stands on weak jurisdictional grounds that the U.S. Court of Appeals for the D.C. Circuit is not likely to uphold. (11) That the Order, likely never to be law, should cause the policy "Chicken Littles" to see falling skies suggests mass confusion or perhaps, more cynically, efforts to influence the appellate outcome.

In this context, Stanford University professor Barbara van Schewick's book, Internet Architecture and Innovation, (12) could not be more timely. It defends Internet regulation, like that which the FCC recently promulgated, as necessary to control Comcast's, Verizon's, and other broadband providers' anticompetitive tendencies (13) and, above all, to ensure the optimal amount of innovation. (14) Van Schewick is a one-time colleague of Larry Lessig, who popularized the "law as code" idea (15) and is a long-time advocate of the "end-to-end" Internet design principle discussed below. Van Schewick's prominent role in Internet debates--the FCC's Order cites and relies on her work in several instances--will make Internet Architecture an important and influential contribution within Internet policy debates.

In a nutshell, van Schewick argues that (1) the Internet's "original" so-called "broad" e2e architecture (16) enables all applications; (2) real options economic theory suggests that Internet architecture should enable as many different innovators as possible; (3) "modular" Internet design reduces coordination and transaction costs of innovation; (4) dominant network operators have the incentive to discriminate against competing applications providers even if they are more innovative; and (5) evolutionary economic theory suggests more diverse innovators will provide better innovation. (17) Van Schewick then concludes that Internet regnlation--of the sort the FCC just mandated in its Order--is necessary to protect Internet innovation. (18)

Internet Architecture's fatal weakness is that none of the areas of economics on which van Schewick builds her argument lends itself to simple conclusions. Disregarding ambiguity in the economic models on which she relies, van Schewick strings together overbroad claims and questionable conclusions to construct Internet Architecture. She uses economic theories to support conclusions, but too often neglects to show in a rigorous way why the assumptions that allow the models to produce her results are more believable than opposing ones. Instead, she uses anecdotes to demonstrate theory. While case studies and narrative may be highly effective tools of persuasion, formal economic models are designed for rigorous econometrics (i.e., properly sampled and controlled data). As the old saw goes, the plural of anecdote is not data. When legal academics claim the power of the social science game, they need to keep its rules.

Van Schewick shows that some economic models suggest that Internet regulation may enhance innovation. But she does not attempt to address her myriad models' ambiguities or examine data rigorously to conclude whether this result is at all likely. Given that Internet Architecture uses economic analysis to justify the specific public prescription of regulating the Internet, this shortcoming is not minor. To paraphrase David Hume, "maybe" does not imply "ought." (19)

Professor van Schewick is not alone. Internet Architecture sadly enters a long syllabus errorum of interdisciplinary legal telecommunications scholarship. (20) Thus, a close examination of van Schewick's argument leads to recommendations applicable to legal interdisciplinary research method generally.

Building on Lessig's work, van Sehewick's work first distinguishes between what she calls the "broad" and "narrow" versions of the e2e network design principle. (21) For nondevotees of Internet wonkery, e2e network design requires that "intelligence" (i.e., application functions) should be located at its "ends" (i.e., the users). (22) The protocols that exchange information between users do not do much except transport. (23) This allows individuals to build-on new applications that can still network and communicate.

To get a sense of what this means, compare e2e network design to its predecessor, the Bell Public Switched Telephone Network ("Bell PSTN") (i.e., how we all communicated before 1995 or so). With e2e, the network smarts are widely distributed and modular. Taking advantage of the Internet's common transport function, users can independently add applications and functions. In the Bell PSTN, by contrast, all the intelligence resided at the switch (i.e., the central place in the network that the telephone company controls). (24) End-users could not add functions; instead, the phone company provided all services (such as voice communications, voicemail, call waiting, speed dialing, and 800 services) centrally. (25)

Notice the tradeoff between e2e and centralized, integrated designs such as the phone network or cable system. Some network functions require greater coordination and centralization than e2e can provide. For instance, the Internet can neither provide the old telephone network's level of security or quality of service, nor match cable systems' video delivery capacity. But the telephone system limits potential uses that e2e permits, (26) Different networks perform some functions better than others.

Larry Lessig and Mark Lemley famously argued that e2e plays a necessary role in Internet renovating. (27)...