Skip to Content

Cybersecurity & Standards, Open Internet

Neutrality and Caching

There has been lots of discussion in Internet neutrality circles this week about Monday’s Wall Street Journal article claiming that Google, in seeking to enter caching deals with ISPs, is departing from its stance in favor of Internet neutrality. Google and a number of commenters (here, here and here) by now have explained why the article is off base. Like so many arguments in the Internet neutrality debate, the article is based on fundamental misconceptions about what Internet neutrality, properly conceived, would require. In effect, the article takes aim at an exaggerated, straw-man version of Internet neutrality. But maybe it offers a “teachable moment.” Specifically, clear thinking about this issue requires recognition that:

(1) a neutral Internet does not require some kind of utopian “equality of results,” in which the resources of different speakers have no impact on the prominence and technical sophistication of their communications or services; and (2) a neutral Internet does not in any way preclude or conflict with the use of caching and content delivery networks (a la Akamai).

Regarding the first point, it is true that caching confers an advantage on those who can afford to pay for it. Caching involves storing Internet content on a distributed set of servers, so that when users request the content it can be downloaded to them from a nearby server instead of a remote one. That translates into speedier download times. But this is just one of many ways that entities with resources can attempt to improve the quality and performance of their online offerings. A service or site that buys more content servers or a faster Internet connection will be able to handle more traffic volume with less delay than a competitor that does not. Better software, snappy web design, better advertising or marketing — clearly, an entity with money to spend has ways to try to make its voice heard or to make its service work more smoothly than smaller rivals. Sensible Internet neutrality advocates aren’t trying to create a world in which the monetary advantages are entirely cancelled out. Significantly, purchasing these kinds of advantages doesn’t degrade the absolute performance level of anyone else’s traffic. Nor do such edge-based tactics require special inputs available only from a single source; computer servers, software programmers, etc. are generally available in the competitive marketplace. CDT sought to explain this in our 2007 comments to the FCC. (We said much the same thing in prior comments to the FTC, as well.)

“Another practice that seems unobjectionable is the provision of caching services. These services differ from the other practices discussed in this section in that they are available from companies like Akamai rather than (or in addition to) ISPs themselves. In addition, caching services do not cause some packets to be prioritized over others during the transmission process; rather, they speed delivery by storing certain content closer to potential recipients. It is as if, rather than giving one company’s vehicles the right to cut in front of others on crowded roads, the company simply established more local offices so its vehicles would have shorter drives. Unlike prioritization in routing, therefore, caching services improve the delivery of some traffic without having any negative impact on other traffic. The only possible concern CDT could envision would be if a network operator were to try to exclude third party caching providers by denying them convenient local interconnection for their data storage facilities, and perhaps then offer its own caching service on a selective basis.”

Based on the final sentence above, there could be a real concern if Google was seeking exclusive deals with the ISPs — but it has expressly said it is not doing so. In short, it’s time to retire the idea that Internet neutrality rules would potentially preclude caching. Nobody is seriously saying that neutrality principles call for such an outcome, and ensuring that any policy steers clear of prohibiting caching shouldn’t be very difficult. The caching question is just a red herring that clouds the debate.