stratechery
stratechery

Netflix and Net Neutrality

This article is by definition very US-centric, but the general principles apply broadly

For anyone remotely connected to technology, the idea that net neutrality is an unabashed good seems incontrovertible, and one of the most popular examples of why it matters is Netflix. Consumers get a video competitor to their cable provider over said cable provider’s pipes; surely the end of net neutrality would mean the end of Netflix! For example, consider this section from Netflix CEO Reed Hastings’ 2014 letter to shareholders:

Unfortunately, Verizon successfully challenged the U.S. net neutrality rules. In principle, a domestic ISP now can legally impede the video streams that members request from Netflix, degrading the experience we jointly provide. The motivation could be to get Netflix to pay fees to stop this degradation. Were this draconian scenario to unfold with some ISP, we would vigorously protest and encourage our members to demand the open Internet they are paying their ISP to deliver.

And yet, just last week, Netflix – without much protest – did exactly what Hastings described: they paid Comcast to stop the degradation of Netflix’s services, and are expected to reach a deal with Verizon and other ISPs soon. The stock market promptly punished Netflix, sending the stock down 3.4 percent the day after the announcement.

Oh, wait, never mind: the stock was up 3.4 percent, hitting an all-time high of $447. This may be the opposite of what most tech observers expected, but Wall Street is not stupid: this is a great deal for Netflix, a company who has every incentive to not support true, end-to-end net neutrality.

Defining Net Neutrality

The first problem with the net neutrality debate is that there are three competing definitions:

  • The public definition – For most people, particularly those of us in the tech industry, net neutrality means non-discrimination against packets from origin to destination. A packet from Netflix or YouTube or PornHub or the New York Times is treated and priced the exact same from server to client and back again.

  • The legal definition – The FCC’s Open Internet rules, which were ruled as overreaching by the U.S. Court of Appeals in Washington, yet still apply to Comcast due to an agreement they signed as part of their acquisition of NBC Universal, only ever applied to traffic within an ISP’s network; in other words, once data is within Comcast or Verizon’s network, they can’t discriminate, delivering some data faster or slower. The Netflix/Comcast deal, on the other hand, is about peering: the point at which data enters the Comcast network (there is a useful overview of peering available here). This is not (and never was) covered by net neutrality, as many geeks are now learning to their dismay.

  • The Netflix definition – Netflix has a subtly different view, best articulated by Reed Hastings himself in a Facebook post two years ago:

    Comcast no longer following net neutrality principles.

    Comcast should apply caps equally, or not at all.

    I spent the weekend enjoying four good internet video apps on my Xbox: Netflix, HBO GO, Xfinity, and Hulu.

    When I watch video on my Xbox from three of these four apps, it counts against my Comcast internet cap. When I watch through Comcast’s Xfinity app, however, it does not count against my Comcast internet cap.

    For example, if I watch last night’s SNL episode on my Xbox through the Hulu app, it eats up about one gigabyte of my cap, but if I watch that same episode through the Xfinity Xbox app, it doesn’t use up my cap at all.

    The same device, the same IP address, the same wifi, the same internet connection, but totally different cap treatment.

    In what way is this neutral?

    What Netflix is most concerned about from a non-discrimination standpoint are broadband caps, and, more broadly, usage-based broadband pricing. It’s not that their position differs on a point-by-point basis from most net neutrality advocates; rather, the priorities are different.

Why the Comcast Agreement is Good for Netflix

This deal is in many ways a win-win for Netflix: they are likely paying less for better quality.

Previously Netflix paid backbone providers such as Cogent for transit; it was then Cogent’s responsibility to interface with last mile provides such as Comcast or Verizon. Cogent made a lot of noise about Comcast and other ISPs wanting to get paid on both sides – by customers for Internet access, and Cogent for peering agreements – but the truth is Cogent was just as duplicitous: they wanted to be paid by Netflix on one side, and effectively subsidized by ISPs on the other.

Free peering agreements between Internet providers were premised on the idea that a roughly equal amount of traffic was going in both directions, meaning there was no net increase in cost as a result of a peering agreement. Netflix, though, changed that equation by moving as much as 30% of all the United States’ internet traffic in one direction (on Cogent’s backbone). Cogent’s insistence on “free” peering, then, was not at all consistent with such previous agreements: Cogent was not carrying an in-kind amount of traffic in exchange for the traffic they were dumping onto ISPs.

As an analogy, suppose my friend Bob and I agreed to watch the other’s dog in the event of a trip or vacation. We both travel about three weeks of the year, so while I have to watch Bob’s dog for three weeks, he watches mine for three weeks as well. While the specific amount of travel may vary year-by-year, it all evens out in the end, and we’re both happy. A few years later, though, I take a job as a consultant, and am suddenly traveling 30 weeks a year. Wouldn’t it be unfair for me to insist that Bob hold to the terms of our agreement, even though it entails him watching my dog 27 weeks more than I watch his? Yet that is exactly what Cogent was demanding for direct access to ISPs’ networks, and the ISPs in turn demanded compensation (keep in mind, Netflix has always been free to use the open Internet to reach customers; they simply find the performance unacceptable and wants shortcuts into ISP networks).

With this deal, Netflix has effectively cut out the middleman Cogent, and is sending traffic directly from their servers onto Comcast’s network. Not only will this mean better quality for Netflix customers on Comcast, but it also raises the barrier of entry for potential Netflix competitors. Netflix currently has unique leverage over Comcast due to Comcast’s proposed merger with Time Warner, which, combined with their brand name and favorability amongst customers and regulators likely meant they got a great deal; future Netflix competitors, forced to go over the open Internet or rely on providers like Cogent will be at both a cost and quality disadvantage.

Most importantly, though, Netflix has to be thrilled that Netflix – not end-users – is paying for better Netflix video, shrouding the extent to which end-users are subsidizing Netflix.

Who Pays for Broadband?

There’s no question, at least in my mind, that broadband is just as much a requirement for day-to-day life as is electricity, water, sewage, paved roads, etc. And, like said utilities, broadband lends itself to a natural monopoly; the cost of capital for building out a network are so great that the economics demand a single provider.

The primary way to deal with natural monopolies is to either have said service provided by the government or provided by a private firm that is heavily regulated with strict requirements about widespread access combined with (relatively) high prices. This is indeed the case with electricity, water, sewage, and roads.1

The problem with regulating broadband in this way, though, is that the definition of acceptable broadband is much more of a moving target. As Marc Andreessen memorably put it on Twitter:

Remember, the United States is a country where one of the two major political parties routinely threatens to default on the nation’s debt to score political points. Infrastructure investment is embarrassingly low in things like roads and bridges, much less in environmentally sustainable power like nuclear;2 to put the future of broadband, something that requires continual investment, into the hands of such a dysfunctional government seems foolhardy at best.

And yet, the fact that wired broadband in particular is a natural monopoly remains, raising the question of how you incentivize investment in ever faster broadband? There are three main options:3

  • Government mandate – Given the assumption that broadband is a economic necessity, this is the prescription that follows. Unfortunately, the same pragmatic problems that make government-provided broadband a likely non-starter plague this as well; Republicans in particular have actively opposed any sort of telecom regulation, even before you get to the incentive problems of mandates versus markets.

  • Discriminatory pricing – Companies like Amazon know that every 100ms delay causes them to lose sales; that makes guaranteed access to end users exceptionally valuable. It’s the same thing with Google, Netflix, and most other Internet companies. Comcast and other ISPs would certainly be incentivized to improve their networks if they knew that said companies would compensate them accordingly.

    To some extent, this is exactly what just happened with the Netflix peering agreement, although true discrimination within their network would incentivize Comcast even more.

  • Usage-based pricing – With usage-based pricing, if you use more data, you pay more; use less, pay less. As we’ve seen with wireless, this strongly incentivizes network providers to increase broadband capacity. It’s no accident that the rollout of LTE in the US was combined with the imposition of data caps, just as it’s not an accident that the US has far better LTE penetration than anywhere else in the world. This despite the fact that providing wireless service in the US is much more difficult than just about anywhere else in the world due to sheer physical size and effective NIMBYism.4 AT&T, Verizon, etc. want you to use as much data as possible as quickly as possible, and to charge you for the privilege.

Making Tradeoffs

In the end, each of these options presents a different set of tradeoffs among three competing ideals:

  • Continual investment in faster and more accessible broadband
  • Non-discriminatory treatment of data
  • Unlimited access

There is no approach, at least given the United State’s political realities, that allows for all three; this is “Fast/Good/Cheap Choose Two” applied to Internet access.

Thus, we need to make choices based on priorities. From my perspective, the most important of these ideals is the non-discriminatory treatment of data. This is what makes the Internet so profound, and what enables new companies to disrupt the market and improve the lives of millions. It must be protected not just within an ISPs network, but all across the entire Internet including peering.

The second most important is continual investment in faster and more accessible broadband. The flip-side of the Internet being so profound is that improved access has an exponential return both from an economic as well as from a societal impact perspective.

That leaves unlimited access on the chopping block. While I love the idea of unlimited data, I also am aware that nothing comes for free; in the case of unlimited data, the cost we are paying is underinvestment and/or discriminatory treatment of data. Therefore I believe the best approach to broadband is usage-based payment by both upstream and downstream, with no payments in the middle.

The way this would have played out in the case of Netflix is that:

  • Netflix would pay more at the point of origin to compensate backbone providers for the massive amount of data they generate
  • ISP customers who watch the most video would pay more

It’s the latter result that terrifies Netflix, and is why, in the end, they are not an ally of those of us who desire true net neutrality. Currently non-Netflix broadband subscribers are effectively subsidizing Netflix viewers; they use much less capacity, yet pay the same price. This needs to change for the sake of true net neutrality, and if it results in Netflix losing subscribers, so be it.

Unfortunately, this agreement and the others that are soon to follow makes such an arrangement unlikely. Comcast and company are getting paid, so they’re happy, and Netflix is disguising their true cost to end users so they are happy as well. It’s non-Netflix users, and, more distressingly, the startups and services that have yet to be created who are ultimately paying the price.

  1. Problems with these models arise when pricing becomes unregulated, or not included at all
  2. If there is one thing to take away from this article as a whole, it is that everything is a tradeoff. I know my aside on nuclear just upset a lot of you, but when you consider the relative cost and capacity of wind, solar, etc, and the environmental destruction caused by fossil fuels, non-polluting nuclear and its spent fuel risks makes a lot of sense. Tradeoffs.
  3. Google Fiber is a fourth: building a straight-up competitor, natural monopoly economics be damned. It’s possible because Google already has its own backhaul network, but I have trouble seeing how it will scale at least in the near term
  4. NIMBY = Not In My BackYard