The Economist has run an article this week that’s left me quite disappointed. In A tangled web, they make an attempt to explain the question of Net Neutrality without taking sides. In doing so, some critical details of the history of the debate are omitted and biased terms are allowed to stand without serious discussion. This leaves the reader in a position where forming an unbiased understanding of the facts is impossible. The result is a rather more biased viewpoint than I’ve come to expect from The Economist. The article’s major failings are twofold. It forms a broad analogy which doesn’t hold up to scrutiny – that “fast-tracking” and “pay for priority” are the same thing – and fails to appropriately distinguish between backbone “transit” (bandwidth between ISPs) and end-user bandwidth.
Fast tracking, they explain, is the practice of being able to pay more money to an ISP to get a faster internet connection. “Pay for priority”, in contrast, is a term used by service providers to describe selling ad-hoc services to a consumer. The problem with the latter is in the details. To implement “pay for priority”, they must limit bandwidth on a per-service basis for other services – either at the transit level or at routers that service individual consumer connections. This necessarily means interfering with network connections, stifling connections that don’t originate at their services in order to provide services to their customers faster. There are several methods which can be utilized for this, ranging from simply providing insufficient transit bandwidth to actively tampering with user’s connections.
The transit method requires that the ISP is also fulfilling the role as content provider, which is the normal case in practice. This method is to simply have much more bandwidth between the ISP’s content servers and the customer than the amount of bandwidth customers demand to other content providers, or to artificially restrict bandwidth between the ISP customers and outside content. This is what they refer to as “reasonable network management”, but it’s anything but reasonable. Under this scenario, an ISP customer buys (for example) 20Mb/s of download bandwidth. But what the customer actually receives is 20Mb/s of download bandwidth from content their ISP hosts, and severely limited bandwidth to other content providers. This is essentially what Comcast wants to do with Netflix bandwidth – they want Netflix’s ISP, Level 3, to pay extra to send data to Comcast’s internet customers, or Comcast will limit their customers’ access to Netflix’s streaming video data.
What the cable companies want agency to do concerning their ISP customers, however, is not limited to merely extortive practices with other content providers’ data. Nor have they attempted to charge their ISP customers more in order to provide more bandwidth to outside content. When Comcast decided that bittorrent was in competition with their TV and movie content interests, for example, they didn’t simply block all bittorrent traffic and demand a fee from their ISP customers for access to the service. In this case there was no individual company from which to extract fees, since bittorrent is a so-called peer-to-peer service, where traffic originates at individual ISP customers both inside and outside of Comcast’s network. Instead, they intentionally modified their ISP customers’ bittorrent traffic by introducing fake packets which reset their customers’ connections, dramatically increasing the protocol overhead for any bittorrent users to the point where the service was not usable. The customer saw what was a technically functioning service, but with performance so degraded that nobody would want to use it. This is what “pay for priority” means in practice. In short, it’s institutionalized fraud: in theory the consumer is paying for bandwidth, a commodity, from an ISP, but in practice the ISP has control over what the customer can use that bandwidth for – thus relieving it of commodity status.
While the article’s conclusion – that lack of competition among service providers is a large problem which must be addressed – is correct, the authors fail to understand that allowing service providers to discriminate among services being carried through the connections leaves an Internet where price discrimination for access to bandwidth is irrelevant.