Well, after almost a year’s downtime I’ve finally found time to get my blog up and running again and even actually write an entry. Spurred by articles on Netflix’s rampant traffic growth last month, I’ve decided to lay down my thoughts on the related topic of Net Neutrality, which may be somewhat at odds with many in the technology community. This is a fairly old topic these days, but one that I think will be relevant for some time to come yet.
You’ve probably heard of the online video rental service Netflix — in case you hadn’t, they’re a company who started as a flat rate DVD rental service in the US but are now best known for their online streaming offering, making their entire catalogue available for watching online for a flat monthly fee.
Well, Netflix have enjoyed some pretty significant growth. Last month, for example, it was announced that they comprised almost 35% of downstream traffic and even 10% of upstream ACKs — that’s a massive proportion of bandwidth for anyone, not least of which a company whose market cap is only around 5% of Google’s. This growth in traffic speaks to Netflix’s rampant popularity, but this success has also brought them some pretty stern opponents — primarily ISPs1.
ISPs seem rather bitter about the success of companies such as Google and Netflix. This is because the ISPs feel that these companies are only able to make their profits because of the network infrastructure that the ISPs have built out at their own expense. This is a rather debatable position, which I’ll come to in a moment, but whether justified or not in recent years the ISPs have become increasingly desperate to somehow jump on the bandwagon and monetise the rampant success of online services.
Clearly the primary source of revenue for an ISP is subscription fees — could they just charge their customers a higher flat rate? Well, increasing fees is generally unpopular, especially in areas where there’s little competition, and would most likely attract a lot of negative attention from regulators. Another possibility for making additional money is to provide their own services, but in practice these are almost invariably significantly less compelling than existing players and gain very little market traction. It shouldn’t be a surprise, since the existing providers make this their sole business — in short, they know what they’re doing.
Faced with these avenues being frustrated, ISPs have instead made moves to try to leach some of the revenue streams away from service companies (e.g. Google and Netflix). They have a trump card to play to achieve this, which is to put roadblocks between their users and the servers used by these companies such that the service gets degraded or removed completely. End users typically don’t know the reasons for this and assume the service is poor, placing their pressure on the provider of the service to fix it. Indeed, exactly this has happened to Netflix where the ISP Comcast in the US started slowing down their data — the problem got so bad that some users were totally unable to watch movies and cancelled their Netflix subscriptions.
Understandably the likes of Google and Netflix are none too happy with what they view as unfair business practices. These fears were proved somewhat justified in Netflix’s case where they were forced to cave in and start paying ISPs for unthrottled access to their customers. This turn of events is a concern to a lot of the online services companies who feel that it’s the thin end of a wedge that’s going to leave them at the mercy of powerful ISPs. As a resut, for years now they’ve been lobbying governments worldwide, but primarily in the US, to pass laws enforcing what’s typically known as Net Neutrality — put simply, the notion that nobody gets to artificially accelerate, slow down or block traffic from one particular source relative to others.
Under Net Neutrality, Comcast wouldn’t be allowed to slow down Netflix movie streaming, for example, although they would be able to, say, offer a uniformly slow service for a lower cost to consumers (i.e. “fair” throttling). As well as these companies, a lot of grassroots Internet advocates also support Net Neutrality, believing that it will help protect the Internet in its current fairly open form from undue corporate interference which could harm the quality of services now and in the future.
Now the actions of all of these parties are quite understandable from their own points of view, but frankly I believe they’re all rather misguided — for the remainder of this post, I’ll try to explain why.
Firstly, as much as the ISPs tend to be demonised in these debates, it can’t be denied that they have a potential problem. As public companies they have shareholders, and like any shareholders they want to see growth in the company and a return on their investment. If ISPs are blocked from innovating to achieve this growth then they’ll stop attracting investment; and that means they’ll be unable to afford to upgrade their infrastructure; and that means gradually degrading service for everyone — that’s a lose/lose scenario. Hence it’s totally natural to see why they’d oppose major legislative constraints on their business.
Secondly, and conversely, it is certainly a problem that ISPs are in the position where they can so easily inflict untold damage on the bottom line of companies who sell their services online. At its most extreme this coud develop into a form of protection racket, where the ISPs can extract arbitrary sums of money from other companies in return for access — I’m not suggesting that it’s anything like this at present, but even the possible risk is clearly not an attractive state of affairs.
Thirdly, a lot of people seem to be putting a lot of faith into legislation to sort this out — they believe some strong Net Neutrality laws will block the ISPs from “abusing” their networks and interfering with other people’s business. But they apparently forget the atrocious record that governments have in passing laws that intersect with business, and especially technology. Those who make policy just do not understand the issues enough to make informed choices, so it becomes a battle of the lobbyists. Let us not forget that ISPs are typically well established companies with plenty of lobbyists of their own, so it’s not at all clear that some awful tangled up mess of a compromise won’t emerge at the end that doesn’t particularly please anyone except the lawyers who’ll get to wrangle over it in legal disputes for decades.
Finally, even if we could rely on government to pen some good legislation now which is fit for purpose and achieves the stated goals, how can we assess the potential for stifling future innovation? There are quite legitimate uses for traffic prioritisation — for example, a cable company might decide to deliver all of its TV services over its IP infrastructure and customers are clearly going to expect to receive this without interruption even if their neighbour is downloading a hundred movies over BitTorrent. Or perhaps a hospital decides to allow doctors to perform operations remotely via the Internet and requires the ISP to give this important traffic priority. Preventing ISPs from implementing this sort of mechanism risks harming both their service and customers in the future by preventing them from making best use of their infrastructure.
In summary, I accept there are problems to solve in the ISP business, but I don’t accept that Net Neutrality legislation is necessarily the best solution to them. So what is?
What about good old competition?
The root cause of these problems, in my mind, is not that the ISPs are able to make decisions about their own networks which affect other companies — the problem is that their consumers are, in many cases, lacking any viable alternative provider they can move to. This problem appears to be particularly acute in the US, where some areas really have no practical choice at all for high speed Internet access. This means the ISPs have more or less carte blanche to implement whatever draconian traffic management policies they like, and their customers can do very little but complain about it. Were there viable alternatives, customers could move away in droves and effectively force a reverse of the policy. For this to work adequately we could perhaps have some light legislation that forces ISPs to make public their traffic shaping measures, but that’s really just tightening up existing regulations on requiring products and services to be accurately described.
Let us not also forget that ISPs don’t just look to exploit other companies — they’ve shown a willingness to exploit their customers as well, admitting that data caps are more about extracting additional revenue than managing network congestion, which is the reason they typically cite publicly. Competition and free market economics handily resolve these sorts of issues as well, whereas Net Neutrality doesn’t say a lot about ISP charging structures, subscription fees or fairness to customers.
Of course, it’s undeniable that competition in this market is incredibly hard to engineer — a lot of basic infrastructure has to be built for even a basic service, and that’s ignoring ask those difficult issues like getting permission to dig up thousands of roads. There are ways this can be done, however, such as “local loop unbundling” or LLU, where the government instead force the incumbents to open up their “last mile” infrastructure to the competition — that’s the connection to your house, and it’s the expensive part that prevents competing entities starting a business.
This might seem unfair to incumbent ISPs but let’s not forget that, for example, the US cable companies only got their local monopolies with a little help from the government in the first place. It’s also important to note that while ISPs may be jealous of the profits of service companies, they’re still large companies with a healthy bottom line — compare Comcast’s profits of $2.6bn last quarter with Netflix’s rather more modest $60m, so incurring some capex in the short term to unbundle their services isn’t going to send them bankrupt.
The use of LLU has been quite successful in countries like the UK, for example. Indeed, earlier this year the incumbent BT asked the regulator to allow it to charge more for LLU access, which is probably a pretty good sign that competition is working. Also, LLU is just one example of an approach that’s been shown to work — I’m sure there are other forms of government intervention which could encourage competition.
Overall, therefore, I would argue that although fostering competition may be the harder path, ultimately the market will be a lot healthier for everyone, not least of which the consumer, if fair market economics is allowed to assert itself instead of relying on the vagaries of government legislation to swing the balance of power around one way or another.
Whilst we’re on the subject I should also mention that this isn’t regarded as a purely legislative or policy issue by everyone — some people believe that a technical solution is a feasible alternative. The most likely candidate seems to be switching to some sort of P2P system, where users share video streams among themselves as well as from Netflix’s servers. This approach is used to implement the popular music service Spotify, for example. It’s not without its potential issues, but P2P company BitTorrent have said they think this is the best approach. Well, OK, so they’re a P2P company, of course they’d say that, right? But actually not so crazy — looks like Netflix has advertised for someone with the relevant skills already, so it seems as if they’re at least investigating this possibility.
Personally I think that’s a mixed bag. On the one hand, it would undoubtedly make it a lot harder for ISPs to selectively block or throttle Netflix’s traffic; on the other hand, ISPs like Comcast have shown in the past that they’re quite willing to take on the challenges of blocking P2P traffic and if Netflix decided to stop paying them then it’s quite possible they’d be up for the fight of doing so again. I think it’s also debatable that a catalogue as large as Netflix’s would benefit from P2P’s swarm effect — the efficiencies will tend to come when a large number of people are concurrently watching a small amount of content. This might work with popular new releases, for example, but the long tail of Netflix’s content may mean that much video just gets streamed from its servers anyway due to lack of other peers. Finally, there are complex issues surrounding use of storage on customer machines and concerns from rights holders over storing their content for longish periods on computers outside of Netflix’s control. I’m sure all these issues are quite resolvable, but it’s certainly not a simple topic.
In conclusion, then, like many issues in life it’s complicated. Parties on both sides have legitimate concerns; and parties on both sides have, at times, been quite disingenuous with their petitions to the government and the public. I think it boils down to whether we can find a way to allow the tried and tested mechanisms of the free market take control; or whether we’re going to roll the dice and let the a handful of government policy makers come up with some sort of legislation that may help or hinder and whose long-term effects are hard to predict.
I know which I’d prefer, but I think there’s only one thing that we can all be certain of — this debate is likely to rumble on for a very long time yet before it’s decided one way or another.
Internet Service Providers (ISPs) are those companies that provide Internet access to the public. ↩