Adam Candeub – Daniel John McCartney

MSU College of Law

Network Neutrality & Network Transparency

I.  Introduction

Network neutrality, though existing only in draft legislation and a handful of Federal Communications Commission (FCC) statements, has inspired a voluminous debate, a debate in which theory and ideology has outpaced fact. Defined as “a bit is a bit” or “like content is treated the same,” network neutrality constitutes a type of non-discrimination requirement, similar to the FCC’s other open access regimes, like open network architecture (ONA),[1] Section 68,[2] or local competition under the 1996 Telecommunications Act.[3] Network neutrality, however, leaves key distinctions unexamined. How and who will determine what content is alike?[4] Is different treatment for spam and other malware “reasonable traffic management”? Does the discrimination involve intranet internet (the “internal internet”), i.e., discrimination of traffic within one ISP or backbone, or does it involve inter-internet (the “external internet”), i.e., discriminatory peering and transit agreements among ISPs and backbones? These questions are particularly nagging as very little about internet management is publically known. As the recent BitTorrent controversy reminds us, very few know very much about internal internet management protocols. Similarly, very few details of peering and transit agreements are public.

Neutrality-advocates tend to support their position by reference to the importance of an open internet to civic society and human communications.[5] They also make economic arguments concerning neutrality’s importance to prevent types of vertical foreclosure and to encourage second-order dynamic efficiency.[6] Intuitively appealing, these insights fail to translate into any workable, even identifiable, rule. Further, neutrality advocates lack any metric for determining whether neutrality makes sense given the potential costs it imposes. Neutrality critics have used this vagueness to paint neutrality as dirigisme, destined to throttle and weaken the United States communication industry.[7]

Neutrality-critics typically rely on economic arguments to show that any regulation would produce inefficiencies. The economic literature has produced elaborate, often impressive models, but as we will discuss, often makes question-begging assumptions that render their insights and predictions of debatable significance to public policy. Further, economic models of vertical foreclosure and double-sided markets often point in different directions, lacking definitive policy prescription. Legal scholars often use broad market arguments to make explicit policy recommendations. But they repeatedly fail to examine whether their broad economic arguments have any applicability to the internet, given our understanding of the way the market works—or the assumptions and ambiguities in the models they espouse.[8] Our limited understanding of the contractual relationship among the content-providers, backbones, backbone access provides, and the ISPs—and the relative market power of each frustrates application of these analytic devices. Finally, the internet has manifest values relating to social capital, cultural expression, civic involvement, political infrastructure, and social meaning that are difficult to measure, indeed, perhaps intractable to economic quantification.[9]

Then, what, if anything, must be done? We maintain that (i) internet policy cannot be made, if it is to be made at all, in an informational vacuum and (ii) disclosure can serve the dual purposes of informing any potential policy and curbing anticompetitive behavior with minimal regulatory cost either to the government or firms. Calls for disclosure, however, have been ad hoc, with various internet controversies like BitTorrent prompting specific disclosures.[10] More important, these calls have failed to describe disclosure’s purpose or justification, leaving unanswered how disclosure would better inform consumers or curb anti-competitive abuses.

Identifying internet disclosure’s nature, purpose, and goals is particularly important given that the traditional economic and regulatory justification for telecommunications disclosure seem inadequate in the current internet market and regulatory environment. First, disclosure’s economic justification—that mandated disclosure is appropriate when information can be more cheaply provided by the supplier and will allow for better consumer choice—seems inapplicable to the internet for the simple reason that given the lack of competition in the last mile connection, consumers lack choice, informed or otherwise. One has little hope that more information will ameliorate that market’s structural inadequacies or increase consumer surplus.

Second, historically, the basis for telecommunication disclosure has been regulatory, to render FCC and state public utility commission oversight more effective. This purpose also seems inappropriate given the agency’s lack of meaningful process to address internet abuses as well as the lack of any generally accepted regulatory goals that would allow for non-controversial FCC regulation—a point the BitTorrent order discussed below makes clear.

Third, the typically “static” type of disclosure—like cost data that the old AT&T system provided—seems to be an inappropriate form for the internet. Unlike telephony, in which routes and carriers were largely fixed, identifiable, and relatively few, internet traffic travels, at least ex ante, through unknown numbers of carriers on largely unspecified routes. Internet connection quality is a probabilistic function (mathematically sketched-out below) not meaningful captured by a handful of metrics, as in the old AT&T system.

Last, markets face a collective action hurdle to providing useful information about internet access quality because the “quality” of a consumer’s internet experience depends not merely on his or her relationship with the ISP, but with the ISP’s interconnection relationship to other networks. No one ISP has the ability or incentive to provide information about other networks so as to provide a more complete picture of internet quality.

We argue that mandated disclosure should create “network transparency.” We look to network engineering to propose a dynamic disclosure regime that exposes network management configurations in machine-usable form and enables consumers to react and even circumvent anticompetitive network managements. Rather than reproduce a static, regulatory disclosure, we encourage a “bottom-up,” flexible type of network disclosure which reacts quickly to changes in network technology and provides the specialized, even individualized, revelations that traditional regulatory disclosure cannot. Such disclosure should provide better informed consumers; and, perhaps more important, a sort of netroots antitrust enforcement.

To achieve this end, we would require public disclosure of (i) each network’s QoS policy, with reference to some industry standard like the widely used DiffServ architecture and (ii) all peering and transit agreements as well as (iii) the implementation of these interconnection agreements by mandating network’s registration in the Internet Routing Register (IRR) or equivalent public registration.

Our disclosure’s immediate goal is neither improved consumer choice nor more effective regulatory oversight. Instead, tracking the incentives that function in open source software collaborations, our disclosure would be aimed at high level network users, what we term with no doubt flattering vainglory, the “internet vanguard.”[11] Our disclosure is aimed at those with pecuniary, personal and reputational interests to track and counter efforts to in open networks to work more effectively, strengthening a balance, even an ecosystem, between network managers’ centralizing tendencies and users’ decentralizing tendencies.

The Essay proceeds as follows: First, we show how the shared ignorance about the contractual governance structures among and between the internet’s various parties and networks vitiates claims about neutrality’s desirability or lack thereof. More broadly, given the vagueness of most regulation proposals, as well as the current state of knowledge, claims about these proposals’ costs and benefits are speculative. Second, we defend our type of dynamic disclosure suggestion its institutional suitability given the internet’s nature. Third, we specify our disclosure.

II.  Internet Ideological Certitudes and Economic Fables

Economics-inspired neutrality debates have swirled around two questions: (i) whether the last mile provider, i.e., the ISP, such as Comcast or Verizon, will likely block or lower the quality of access to content providers with which they are unaffiliated (“vertical foreclosure”)[12] or (ii) whether neutrality will increase or decrease investment and innovation (“dynamic efficiency”).[13] Despite the conviction with which these points are argued, neither economic theory nor empirical evidence offers many definite, non-controversial conclusions, given what little is known about the contractual basis of internet interconnection.

a. Is foreclosure a reasonable concern?

Neutrality advocates fear that ISPs will engage in a strategy of “vertical foreclosure” will block or degrade connection with non-affiliated providers The neutrality’s advocates’ feared scenario involves (i) an ISP refusing “interconnection” (a term further described below) with a provider that offers a competing service, i.e., a telephone company that is an ISP will refuse interconnection with a competing VoIP service, or, (ii) an ISP charging exorbitant prices for interconnection acting as an oppressive toll keeper depressing the incentives for online production or providing a de facto refusal to interconnect.[14] In both instances, the ISP uses its market power over access to the consumer to foreclose competitors in prior stages of production and distribution.

To the contrary, neutrality critics argue that (i) there is no incentive for such foreclosure because the ISP has a market incentive to provide as diverse and consumer-enticing platform as possible and (ii) somewhat contradictorily, there are great efficiencies in vertical integration as well as greater profits and, therefore, greater investment incentives. Variations of these arguments characterize the entire legal-economic debate.[15]

Let us assume, perhaps not so heroically, that ISPs have some level of market power with regard to the consumer. The question is what they do with it. Network neutrality advocates fear some type of leverage or foreclosure, working like this:

One problem with this scenario is the “one monopoly rent” theorem, which states that the monopolist would have no incentive to expand its control into a market into which it could only gain competitive returns.[16] Here, the notion would be that even if the ISP had a monopolist position with respect to consumers, the ISP would have no incentive to eliminate competition among content providers. Because consumers are willing to pay only a certain amount for internet services, the single monopoly rent can be extracted by the ISP or the content providers, not both.

Nonetheless, the theorem has many exceptions.[17] Most straightforward, it does not apply in instances when the adjacent market offers additional rents that the monopolist can extract.[18] In situations in which revenue from the product can be garnered from different sets of users, exclusion may be a reasonable strategy. If the monopolist, due to network effects or other efficiencies can achieve greater income than its rivals in the complementary market, it would have the incentive to exclude. In instances the monopolists refuses interconnection in order to discourage possibly monopoly-threatening innovation by other firms and thereby protect monopoly rents.[19]

Yet, no one really knows whether these exceptions apply or not. Such knowledge would require rather detailed revenue and cost information about networks and their applications. While such large scale investigations are, in theory, possible, one is hard pressed to think of a regulatory body capable of such a study.

b. Are ISPs double sided markets?

Other analysis has viewed the internet as a two sided market. These markets involve networks with two distinct user groups. Each group provides each other with network benefits. For instance, credit cards are such a market. They provide credit to consumers and consumers to merchants. The value of the card to both groups, consumers and merchants, depends on the size of the other group, i.e., credit cards accepted by more merchants are more valuable to consumers; conversely, credit cards that more consumers hold are more valuable to merchants.

The optimal pricing in two sided markets is not clear. Indeed, there are certain circumstances in which charging more than a competitive price to one side of the market leads to optimal social welfare. The models point both ways.[20]

The fundamental problem with either the vertical foreclosure or the two-sided market story is that is not clear that the ISPs and content providers are in adjacent markets. Vertical foreclosure requires an ISP to directly control the input of some good, here internet content. Similarly, two-sided markets require that a platform bring together two distinct sets of users. But, are these scenarios realistic? Consider this schemata of internet interconnection:

It is not clear that any ISP has any direct contractual, or even physical, relationship to any content provider. Rather, content is received through a complicated contractual network between numerous parties. Market power in content provision requires not simply market power over the last-mile, which the ISP arguably has, but market power in the network itself, a more complicated proposition.

Finally, in the end, conceptualizing network neutrality through the lenses of the single monopoly rent theorem or even double sided market often reduces into a trade-off between adopting regulation that limits the monopolist from obtaining rents but lowers its incentive to invest in infrastructure and the application incentive to innovate.[21] Despite claims in the literature otherwise, there is no empirical consensus emerging as to where the balance on this question falls.[22]

Consider an example, Ebay. Presumably, the network neutrality story is that the ISP would siphon off Ebay profits through higher interconnection charges. But, does the ISP interconnect with Ebay? Most likely it does not directly. It would have to siphon off fees through its backbone connection—and there may be several jumps to get back to Ebay’s servers. Each participant in this link could demand and potentially extract economic ranks.[23]

An ISP could try to block all traffic to or from Ebay subject to receiving a fee. Implementing this block, however, would raise significant transaction costs for ISPs—requiring them to both track traffic and negotiate with an almost infinite number of parties whose traffic transits their network. This transaction would continue to increase as content providers attempt to avoiding detection, creating a cat-and-mouse game.

Perhaps more important, any plan of discrimination would complicate its interconnection with other backbones, which are in the business of carrying bits that will be received by their intended recipient. An ISP must be in the business not only of delivery, but also of sending. A backbone, therefore, would likely not accept traffic from an ISP or other network that would not deliver the packets it is handed off—or (more likely) would charge more. In short, discrimination likely would make peering or transit more expensive.

At the same time, these dominant modes of economic analysis—which concentrates on the role of the ISP—distracts from the real possibility that concentration of the backbone in conjunction with ISP market power could have anticompetitive results. The backbone continues to be ever more concentrated, and those who control it are often those who dominate the late mile, ISP connection. It is not clear what theoretical model should apply to these complex, but ever more concentrated, network markets. Given the value the internet provides to society, even a small level of anti-competitive behavior, however, could have significant social welfare consequences.