The Center for European Policy Studies recently released a 'Special Report' on network neutrality in the European Union. It is quite elaborate and well argued, and author Andrea Renda has made a great effort to translate further this hot topic into European affairs. Here's my take on it:

The main thing that Renda and I agree on, is that the network neutrality debate should not be framed as a binary opposition between neutrality and network management. It would indeed be more productive to focus on which forms of network management are appropriate, and which are not. I share Renda's doubts about the feasibility of absolute end-to-end, though without the pathos Renda invokes here. By stating emphatically that the Internet is not, and has never been, neutral, Renda falls back into the the binary pitfall that he argues against.

This signals a common strategy used by opponents of network neutrality—dismissing proponents as fanatics who stubbornly hold on to absolute neutrality against technological reality. While this gives opponents of net neutrality a nice straw man to beat up, it isn't based on more than clever rhetoric. So when Renda brings up Dave Clark as stating that the net "is not neutral, and has not been for a long time", this only affirms what has been common knowledge for proponents of net neutrality for a while already. Even one of the most partisan proponents of end-to-end, Lawrence Lessig, acknowledges this in one of his earliest writings on the topic (section F). Given that Lessig also wrote one of the more pessimistic works on the inevitability of normative technology on the Internet, this isn't really surprising to me.

Another rhetorical trick that I take issue with is the 'wrong metaphor' strategy, which has been exposed by Frank Pasquale at Seton Hall. Opponents of net neutrality pull the 'wrong metaphor' card whenever proponents compare neutrality of the Internet to infrastructures regulated under common carriage, like the electric grid. Instead, it is argued, why not compare the Internet to postal services or advertising, where prioritized treatment when paying more is commonplace? Renda does exactly this, by asking the question why the Internet should be neutral and non-discriminatory, when prioritized treatment in the airline industry—between business class and coach for instance—is such a standard practice. Now what Renda doesn't mention, is that business models based on prioritizing cannot go without elaborate cross-subsidizing structures—for instance: by keeping the prices of express mail high, the prices of regular mail can be kept low.

Cross-subsidizing is very prone to abuse, and allows incumbent network operators vertically integrated into retail markets an unjust competitive advantage. Remember that far-reaching cross-subsidies in American telecommunications led to the break-up of AT&T in 1984. Now I am aware of the current debate on how much of a bad thing cross-subsidizing in telecom really is, but Renda passes by the discussion altogether. In any event, simply stating that your metaphor is better doesn't cut it.

Renda's otherwise well-developed analysis becomes problematic when he addresses congestion issues on the network. He relies on work by George Ou, who works for an industry-funded think tank and casually compares his adversaries to nazis. When we actually look at the figures, we see that the data as provided by Ou solely represent cable architecture. Without even going into the data as such, the problem is that only about 15 per cent of Europe is covered by cable. When looking at figures provided by the OECD, you can see that (with a few exceptions) cable coverage in European countries is rather insignificant. This makes Renda's argument problematic, simply because congestion of the network is much more of an issue in cable than it is in DSL—the latter being much more common in Europe. Cable architecture is built such, that congestion can arise pretty much anywhere on the network, to such an extent that neighboring cable connections influence each others' speed. DSL connections in contrast remain independent of each other until they reach the central office. Christopher Yoo writes about this distinction at length.

What Renda does here, is transplanting congestion issues in cable, which forms a minority in Europe, to the European Internet architecture at large—even though such congestion is very unlikely to occur in a DSL dominated architecture.

When talking about congestion moreover, it is important to realize the inherent bias of this subject. Note that network operators can only make money by offering prioritized services in case of congestion. This leaves them with a strong incentive not to upgrade their network to such a level where congestion would not occur.

What really surprised me, is that Renda suggests European regulators follow the precedent set by the FCC, in providing an Internet Policy Statement of loosely formulated principles against traffic discrimination. I don’t think this would be a good idea. The FCC’s Policy Statement immediately created a lot of confusion regarding its enforceability, which culminated in the by now infamous Comcast case—currently being fought out in the DC circuit. A similar European Policy Statement would result in the same regulatory confusion, which would indeed be a solution in search of a problem.

Notwithstanding my critique on Renda's report, we agree on some key things. Most importantly, I also think that overall European antitrust is robust enough to deal with the most blatant discriminatory practices on the Internet. Blocking of traffic forms an easy case for art. 82 of the EC treaty, and competitive abuse in terms of essential facilities is likely to apply. This conclusion seems to suggest for Renda that European regulators should approach the issue of net neutrality with due prudence. Here I also agree. However, a few issues remain over which Renda glanced all too quickly:

What to do with degradation of traffic? Deliberate degradation of data packets seems clearly anti competitive, while Bronner seems to suggest that an essential facilities claim cannot be brought here—simply because access is not completely blocked off. It is moreover likely that degradation practices will be implemented differently across European member states, which can have a fragmenting effect on the market. Finally, it is questionable whether mandatory disclosure by network operators of which services are being degraded and which are not, is going to provide the transparency that the European Commission envisions here. Isn't this too much to ask of consumers?

And what to do with prioritizing, or as Renda refers to it, access-tiering? This is clearly less anti-competitive than degrading or blocking, and possibly even welfare enhancing. However, more research is needed to determine what effect it has on small-scaled market entry, and how likely it is for network operators to slip into degradation of traffic while prioritizing.

The disclaimer here is that I hope to address these open questions of degradation and prioritization in my future PhD work. And I hope to have the opportunity to share thoughts with Andrea Renda soon.