The limits to net neutrality

What is net neutrality? There are as many definitions as there are stakeholders in the net neutrality debate. But proponents generally agree on some form of the following: Internet service providers should treat all traffic equally. If you break that down further, there are actually two parts to it: no traffic should be advantaged over any other (what you might call “positive” non-neutrality), and none should be disadvantaged against any other (what you might call “negative” non-neutrality). As with almost any debate, there are many positions on this issue, on a sort of spectrum. A simplified version, with extremes at each end and an illustrative in-between position, is shown in the diagram below. Net neutrality spectrumsThere are certainly people at each extreme on both of these topics, but the debate is playing out very differently on each.

Blocking and degrading (“negative” non-neutrality)

On blocking and degrading, there is no-one with a real stake in the industry that believes outright blocking or degrading of non-nefarious traffic should be permitted. There are less than a handful of proven cases in which this behavior has occurred (the Madison River and Comcast/BitTorrent cases are probably the best known) and major providers today disavow any intention of doing this. So the left-most extreme on the spectrum above is essentially unoccupied.

However, there is a middle camp, including the providers, that argues that reasonable network management should be permitted, to prevent serious congestion caused by one application from degrading all other traffic, for example. The FCC wrote such an exception into its original Open Internet rules. There are, however, those who believe that even this “reasonable” network management should not be permitted, and that the solution to network congestion is to build fatter pipes. This ignores a fundamental facet of how TCP/IP works, namely that it is designed to use whatever bandwidth is available so that there will always be some danger of congestion. But it is nonetheless a position some advocates take, though I would definitely characterize them as being on the fringes of the debate.  As such, much of the debate on the negative side is in the center – what constitutes reasonable network management? Who is to decide? What are the standards and principles that should apply? Is it OK to degrade all applications of the same kind (e.g. all video streaming regardless of the specific application) to prevent congestion, or is even that out of bounds? One of the biggest challenges here is that intent plays a huge role. A provider might claim to be slowing certain traffic to preserve performance for all other applications, but others might believe the degradation is being done for competitive or other reasons. How would you find out? How would you prove your case?

Prioritization (“positive” non-neutrality)

On the prioritization question, there is also a divide. However, on this question the extreme neutrality position is actually fairly mainstream. The bulk of net neutrality advocates takes the position that providers should never prioritize any traffic over any other no matter what the reason. There are a number of reasons put forward for this position, among which these two are frequently cited:

  • broadband providers shouldn’t be allowed to favor their own content. We might call this the unfair competition argument.
  • wealthier content and service providers shouldn’t be allowed to pay for preferential treatment, which would by default disadvantage poorer providers. Let’s call this the startup argument.

The other end of the spectrum is more inhabited on this question, too, though most providers have been careful to stay away from the absolute extreme. Verizon, though, has talked recently about new business models under which it would charge content providers for prioritization. And providers have in general argued that it would benefit users to prioritize certain latency-sensitive application classes (e.g. video calling or gaming) over less latency-sensitive ones (email or web browsing). Then there is a middle ground, which is mostly made up of a number of broadband providers and their advocates, which sees a role for transparent, open, paid-for prioritization of content. In other words, the business model would be open to anyone and the broadband provider would be transparent about which services are being prioritized. The challenge comes when we start to pick apart two of the reasons for banning all prioritization, mostly because almost they would apply to things that no-one finds objectionable today as well.

Unfair competition argument

Cable providers, AT&T and Verizon all offer TV services alongside their broadband services. These services travel over the same wires to the home, but are typically delivered over separate channels with different technical characteristics. These TV signals do not have to compete with other TV traffic because of the way they are delivered technically. As far as I know, no-one considers this objectionable. However, over the coming years it is entirely possible that at least some of these companies will find it more efficient to have their TV signals use the same technology and channels as all the other broadband traffic to the home, but they will likely want to prioritize this traffic to ensure that it is delivered at the high quality the customer expects. Would that be objectionable? What if the provider offers an entirely “over the top” service to any broadband customer, but prioritizes the traffic when delivering to its own broadband customers? Would that be objectionable? The technically separate delivery of TV signals today seems to exempt these services from net neutrality scrutiny – would that continue if the architecture changed? Why? By the way, the same arguments might equally apply to telephony, which has historically been delivered over dedicated PSTN networks, but is increasingly delivered (even by incumbent telecoms providers) as VoIP.

Startup argument

As a reminder, this argument says that broadband providers shouldn’t be allowed to charge for prioritized access because this would disadvantage smaller providers. The history of successful startups is often drafted to support the cause here 1. The challenge is that this is the way the world works: larger, wealthier companies always have advantages over smaller ones, because of all the infrastructure, resources, people and technology they can afford. Even if we narrow things down to the world of the Internet, content delivery networks (CDNs) stand as the prime example of existing business models no-one finds objectionable under which larger companies have a distinct advantage over smaller ones. When companies such as Apple use Level 3 and Akamai to host their popular content close to users, it means that content gets to users faster and more predictably. When Google or Apple build their own in-house CDN infrastructure, the same results flow, and in both cases these wealthy companies are paying for priority placement of their content, which enables faster, more reliable delivery. How is prioritizing traffic over the last mile (which CDNs don’t help with) anything other than a complement to CDNs which applies exactly the same principles? 2

Where things stand today

As I summarized above, the debate on net neutrality is really a debate about two separate questions – blocking and degrading traffic on the one hand (negative) and prioritization on the other (positive). Mainstream positions are far more diverse on one than the other, as shown in the revised chart below, where the darker shading indicates more occupied positions on the spectrum: Net neutrality spectrums with shadingWhen people say they’re for or against net neutrality, they really need to be more specific: where do they stand on these two questions? Under what circumstances should blocking or degrading traffic be permissible, and under what circumstances should prioritizing traffic be permissible?

Regulatory approaches to net neutrality

Given all this, what should regulation have to say about net neutrality? Here are some thoughts:

  • Regulation needs to clearly distinguish between these two behaviors. They are linked, in that prioritizing some traffic would effectively lead all other traffic to be degraded if there were congestion, but they are fundamentally different behaviors, and need to be treated as such.
  • There is much agreement around allowing reasonable network management when it comes to blocking and degrading traffic. Yes, there are extremists, but reasonable people acknowledge the need for some management, especially on mobile networks, which are much more prone to congestion. This should be done by managing classes of applications rather than individual applications where possible, and there should be transparency. There needs to be a debate about intent and its importance, but ultimately any regulation which takes intent into account will be essentially unenforceable.
  • On the positive side, however, there is a much wider range of opinions, and the extreme neutrality position is popularly held by many people and consumer rights’ organizations. I have pointed out above the flaws in this approach to the extent that it relies on the unfair competition and startup arguments. Any regulation which relies on these principles will find itself either thrown out on the basis of inconsistency or applied to a much wider range of behaviors and business models including CDNs. Does anyone really want that? I believe that regulators will have to permit business models under which content providers pay for prioritization, but the key here too will be transparency and openness.

There are no easy answers when it comes to regulating net neutrality. In the US, the historical system of addressing abuses as they occurred actually worked reasonably well until challenged in court, which led to the regulatory overreach that was recently itself overturned in the courts. As the FCC goes back to the drawing board with a new approach, I hope it bears these principles in mind. On that basis, today’s announcement from Tom Wheeler is promising.


  1. Boston Globe: “The next Facebook might never take off; conceivably, Verizon or Comcast could slow traffic to and from the site if they’d already inked an agreement with one of the earlier, rudimentary social networking services like Friendster.” The Daily Beast: “Without Net Neutrality, the next Google being built in a garage somewhere will never get off the ground.”
  2. I’ve been writing about CDNs and net neutrality since 2010, but a recent (I think tongue in cheek) tweet from Benedict Evans sparked an online conversation I wanted to pen a fuller response to).