Thoughts on Scott Hemphill’s Network Neutrality paper

Scott Hemphill has posted an excellent, thoughtful paper on network neutrality. I’ll post the abstract along with a few comments on the paper below the fold:

Abstract: This Article examines zero‐price regulation, the major distinguishing feature of modern “network neutrality” proposals compared to traditional regulation of infrastructure industries. A zero‐price rule prohibits a broadband Internet access provider from charging a content provider to send information to consumers. The Article differentiates two access provider strategies thought to justify a zero‐price rule. Exclusion is anticompetitive behavior that harms a content provider to favor its rival. Extraction is a toll imposed upon content providers to raise revenue. Neither strategy raises policy concerns that justify implementation of a broad zero‐price rule. First, there is no economic exclusion argument that justifies the zero‐price rule as a general matter, given existing legal protections against exclusion. A stronger but narrow argument for regulation exists where socially produced content—content (such as Wikipedia) produced by collaboration without anticipation of financial reward—competes with ordinary market production. Second, prohibiting direct extraction is undesirable and counterproductive, in part because it induces costly and unregulated indirect extraction. I conclude, therefore, that recent calls for broad‐based zero‐price regulation are mistaken.

Let me offer a few thoughts:

• According to Scott, “Network neutrality proposals share a distinctive feature, what I call a zero-price rule. A zero-price rule prohibits an access provider from charging content providers to send information to consumers.”

Note that the label is somewhat misleading in the sense that it is not saying that access providers must charge a price of zero. I know that is not what he means exactly because the real thrust of the rule is that the AP cannot charge content providers for accessing their customers. It is a zero price rule only with respect to a particular type of potential commercial relationship.

• Scott correctly notes that network neutrality (NN) is different than common carriage because it would constrain price discrimination in some situations where common carriage would not. He is right, but that is why no one is calling it common carriage. It is analogous in form and function but not identical, because the nature of transport is different: networks of networks running TCP/IP where end-users are both senders and receivers (and producers!) of content that traverses multiple access providers’ networks.

• Scott bases his analysis in antitrust economics. This framing misses the mark from the outset in my opinion. NN is NOT justified on antitrust economics alone and analyzing access provider strategies for anticompetitive effects misses the forest for the trees. At best, the antitrust/regulatory economics story is a “tie” à it doesn’t mandate NN or no NN. As I’ll note below and have argued elsewhere, the social value from a nondiscriminatory “mixed” infrastructure derives from a wide diversity of private, public and nonmarket goods produced by users. (see various papers, x, y, or z)

• Scott notes correctly that “Much anticompetitive exclusion is already prohibited by existing antitrust law.” He then argues that “To the extent that antitrust law as currently enforced is successful in identifying and remedying exclusion, there is less need for a new layer of regulatory intervention.”

There are a few leaps of faith in this argument, perhaps the biggest being “successful” enforcement. (See my Revitalizing Essential Facilities paper just published in the antitrust law journal, draft here) In the end, “less need” isn’t “no need,” and so I am not sure where relying on antitrust law to get it right gets us. (Actually, as I noted above, I think the NN debate is overly focused on antitrust as an effective substitute for regulation; Scott fairly acknowledges that antitrust law does not solve all problems.)

• Scott recognizes a “relatively stronger argument for zero-price regulation … in the narrower case of socially produced content.” He seems to define socially produced content broadly – content produced when individuals collaborate without anticipation of financial reward – but then concludes that socially produced content “raises distinctive issues for regulatory policy where such content competes with ordinary market-produced content.” The distinct issues have to do with the fact that antitrust law will not remedy inefficient exclusion of socially produced content.

His “socially produced content” exception is an exception that swallows the rule, in my mind. It may appear to be a narrow exception that antitrust doesn’t deal with very well, but it is a wedge that pries the cracks in the dam open – the Internet as we currently experience it is flooded with socially produced content, some of which coheres in content or a product that directly competes with an existing market actor’s content or product and much of which does not directly compete yet or in a manner that is easily observed and accounted for in an antitrust economics framework. Experts are trying to figure out what is happening in various media, where old and new forms of production meet, merge, compete and struggle for relevance. The Wikipedia vs Encyclopedia Britannica battle may be easy to frame, but it is only an example of a rather narrow subset of social production.

E.g., small scale productive behavior that is widespread, as it is on the Internet, generates significant social value, some portion of which may very well have direct economic benefit, regardless of whether it is sufficiently organized and collaborative to yield a product that competes directly with a commercial product; various niches are filled where commercial products simply do not venture because niches may be small in scale. … Moreover, many information-communication-social activities that take place online have spillover effects for offline social systems (whether economic, cultural or political).

• Scott adopts a rather common position regarding the parties involved and focuses on how NN would affect access providers and content providers. Who are the content providers? Well, Google, of course, and others like Google, like Yahoo and other search engines. Not content providers really, but application or service providers. But wait, they provide content too. Anyway, you get the point, the big companies that send a lot of traffic to consumers.

But this highlights one of the problems I have with the NN debate. Consumers are users that produce apps and content, and Google is a consumer and user and producer, like me. Yes, we are different in some ways, but not necessarily in a way that matters. I could be the next Google (I wish!) or the Google of my own “space” on the net. I could go on about this for quite a bit but will avoid the temptation. (a few questions come to mind: Are “content providers” defined by the nature of the content they provide? whether they are commercial or not? if/how they generate revenue? whether they are big or small? whether they “compete” (whatever that means exactly) with “ordinary market actors” (whatever that means exactly)?)

Sidenote: I guess I am not ready to abandon a rather strict NN rule (no discrimination among packets based on the identity of user or use) in part because I don’t see how lines can be drawn around “content providers.” Of course, there are other reasons and there are drawbacks to such a rule, but I am heading too far from Scott’s paper … back to the point about google and I both being content providers, users, consumers, etc.

• Scott notes, for example, that “In addition, extraction has a dynamic efficiency consequence if it alters the investment decisions of content providers.” I agree with this and think that he needs to substitute “users” for content providers, for two reasons. First, I am not sure how he can draw a line around “content providers” that excludes users. Second, making the substitution makes an important point:

Extraction has a dynamic efficiency consequence if it alters the investment decisions of users, and such investment decisions should be understood to include investment of money as well as time, effort, and reputation; the point is that extraction has a dynamic efficiency consequence if it affects user behavior.

If it were merely an exercise in transfering wealth, then we might overlook extraction, but it plainly is not just wealth transfer between the Comcasts and Googles (and their respective shareholders); NN, exclusion and extraction strategies, prioritization, etc. is very much about shaping the investments and behavior of many different actors.

• Scott also raises a concern that NN would preclude access providers from pursuing a possible strategy of subsidizing consumer adoption of broadband. If he made the case that this type of strategy was itself essential, then I could see where this is going; but the fact that preventing network providers from extracting rents from some content providers may (or may not) impact their build out and may (or may not) prevent them from lowering prices to encourage adoption does not seem like a strong argument against NN; there are alternatives to this strategy that might also be worth considering as well; i.e., adoption can be subsidized via other means.

Alright, this post is MUCH longer than I anticipated. So let me end with a suggestion: read Scott’s excellent paper and then come back and comment here. Let us know what you think.