Thinking about Bill Herman’s argument that network discrimination threatens freedom of speech, and his broader point that broadband ISPs can use their control over the “last mile” to force Internet users to do or not do certain things (mostly, give them more money), it occurs to me that Hayek’s discussion of coercion in chapter 9 of The Constitution of Liberty has some relevance:
Coercion occurs when one man’s actions are made to serve another man’s will, not for his own but for the other’s purpose. It is not that the coerced does not choose at all; if that were the case, we should not speak of his “acting.” If my hand is guided by physical force to trace my signature or my finger pressed against the trigger of a gun, I have not acted. Such violence, which makes my body someone else’s tool, is of course, as bad as coercion proper and must be prevented for the same reason. Coercion implies, however, that I still choose but that my mind is made someone else’s tool, because the alternatives before me have been so manipulated that the conduct that the coercer wants me to choose becomes for me the least painful one.
Now, I should say at the outset that Hayek would not have regarded an ISP trying to manipulate its customers as coercion, as he was thinking about cases where a coercer was able to exert broad control over peoples’ lives through threats of violence. But I think that what Herman claims AT&T will do to Internet users is analogous (albeit much less severe) to Hayek’s description of how states can coerce their citizens.
Hayek’s crucial insight here is that coercion generally proceeds by taking away alternatives (or making them more painful), rather than by by directly compelling people to do what the coercer wants. The state doesn’t burst into stores every few hours and confiscate a fraction of their revenue. Rather, they threaten the stores with fines, and their owners with jail terms, if the stores fail to submit the required fraction of revenues as a sales tax.
The same principle applies in the network neutrality context. In our hypothetical case from the last post, AT&T has no way of forcing their customers to read pro-choice blogs. Rather, they try to accomplish the goal indirectly, by taking away the option of reading as many pro-life blogs as possible. Likewise, AT&T can’t force anyone to use its VoIP products, but it might be able to manipulate circumstances so that for many users, AT&T’s VoIP offering becomes “the least painful” option.
I think this highlights an important point: in order to be effective, coercion must remove all (or at least nearly all) the alternatives to the action the coercer wants to promote. Shutting down half the pro-life blogs in the world probably won’t have much impact on the total number of people reading pro-life blogs because the former readers of those blogs will shift to reading other pro-life blogs. Likewise, blocking one VoIP application won’t do AT&T much good if there are lots of other VoIP offerings available that are superior to AT&T’s own product. Only if they can block almost all of the VoIP products on the market simultaneously do they have any chance of getting a significant number of customers to use their product.
That’s particularly true because people absolutely hate being manipulated. So if they discover that AT&T is deliberately blocking Internet services in order to promote their own offering, they’ll actively seek out alternatives just to spite AT&T. So either you have to block absolutely every competitor, or you have to do the blocking in a way that no one will detect.
And that’s a serious problem for a company seeking to profit from network discrimination because despite Herman’s name-checking of “sophisticated, packet-level network-management tools,” it’s remarkably difficult to block an entire class of applications. As I’ve written before, in college I saw firsthand just how challenging it was for a big state university to block peer-to-peer applications from their networks. There are all sorts of ways to disguise one type of traffic as another type, or to obscure all of your traffic so that the authority doesn’t know what to block.
The fundamental problem is one of information overload. Precisely because the Internet is inherently an end-to-end medium, there are over a billion people coming up with new and creative ways to transmit information. Just as it would be incredibly labor-intensive to read every blog on the Internet and classify them by political view, it’s also labor intensive to study and identify every communications protocol (and variant on communications protocols) that each of the billion people on the Internet use. And there’s no way to automate the process, because it involves matching wits with other human beings.
So it just doesn’t strike me as plausible that, in general, the ability to filter traffic would be a terribly effective way to increase profits. The one possible exception is introducing jitter to degrade realtime applications. Ed Felten has a great discussion of that here, and I offered some reasons to think introducing jitter would be counterproductive for a telco here. Other than that, I’ve yet to see anyone suggest a plausible explanation for how a network discrimination regime would work.
Governments have the crucial advantage that they can shoot people who try to evade their controls. AT&T has no such alternative the worst they can do is cut of someone’s service, and even that is unlikely to be especially effective, because you’d have to publicly announce that you were cutting off his service for trying to use a non-approved application, which would generate a lot of negative publicity and make a lot of other customers more determined to evade AT&T’s filters.