More Net Neutrality Violations That Aren’t

I see ACLU's Jay Stanley has penned a reply to my post from a couple weeks back on the civil liberties group's report arguing for the urgency of net neutrality regulation. The main thrust of my post was that many of the examples advanced to show there's an imminent threat to the open Internet, requiring regulatory action on the double, don't really show anything of the sort. Stanley allows that some of their examples are "not violations of Internet network neutrality in the strictest sense" but that they "speak to the motives, intent, and trustworthiness of major telecommunications firms in treating the speech of their customers fairly." But I'm not sure they really show that either. In fact, if I can be forgiven a little digression, two more egregious corporate offenses against net neutrality that turn out not to be.

First, one I'd missed from the ACLU report: Vague terms of service agreements. Apparently, AT&T's terms of service had a list of grounds for suspension of service that ended with the rather nebulous provision bolded below:

AT&T may immediately terminate or suspend all or a portion of your Service, any Member ID, electronic mail address, IP address, Universal Resource Locator or domain name used by you, without notice, for conduct that AT&T believes (a) violates the Acceptable Use Policy; (b) constitutes a violation of any law, regulation or tariff (including, without limitation, copyright and intellectual property laws) or a violation of these TOS, or any applicable policies or guidelines, or (c) tends to damage the name or reputation of AT&T, or its parents, affiliates and subsidiaries.

Based on the company's explanation, it sounds like they intended this as a sort of catch-all for behavior that wasn't covered by their policy or the law, but was sufficiently clearly abusive to damage the reputation of a provider who allowed it. But you can certainly understand why people read it as reserving the right to disconnect people who criticize the company, and in any event, it does seem way too vague: Who wants to risk losing their service based on such ill-defined criteria? Significantly, though, I don't see anybody claiming that AT&T or Verizon (which had similar language) ever actually did suspend a user's account for this reason. It appears to have been one more overbroad bit of legal boilerplate drafted by a lawyer paid to shield the company from liability in as many contingencies as possible, and promptly changed when users complained. More importantly, and at the risk of stating the obvious, this isn't really a question of network architecture. Such a broad provision could surely be enforced in a way that was contrary to the spirit of the open Internet, but it's ultimately a provision about how AT&T treats its customers, not about how routers treat packets. Many things might be wrong with it, but violating the end-to-end principle embodied in the TCP/IP protocol isn't one of them. Indeed, there's nothing really Internet specific about this at all: An offline business could attempt to refuse service to people who publicly criticize the company in the newspapers. Mercifully, such behavior seems rare, but if you're worried about the potential for a certain class of abusive contracts aimed at squelching speech isn't that where the remedy should aim?

Second (via Seton Motley), there's the ongoing scuffle between Cablevision and Fox. Presumably in hopes that Cablevision would be under more pressure to cut a deal for Fox cable channels if their subscribers couldn't just get Fox content online, Fox blocked access to their Internet video content for Cablevision subscribers, prompting Art Brodsky of Public Knowledge to fret about the danger to the open Internet. He acknowledges that normally, folks worried about neutrality have focused on the threat of ISPs leveraging access over the pipes to control content, but asserts that "it shouldn’t matter who is keeping consumers away from the lawful content."

This is just weird. Media companies "keep consumers away from lawful content" all the time! Netflix won't let me stream their movies unless my subscription is paid up. If I try to access academic articles on JSTOR from home, whoops, I'm blocked! I have to be visiting from an IP address at Cato or some other academic institution that's made a deal with JSTOR for access. BBC won't let me watch Sherlock or Doctor Who on their Web site, because they've sold the U.S. rights to PBS and SyFy, respectively. "Net Neutrality" and "Open Internet" have a dizzying array of different definitions, but even so, the idea that either obligates content providers to make their content equally available, for free, to every user is... novel.

I harp on this because I think it indicates how muddled a lot of the debate over "neutrality" has gotten. People have a whole welter of heterogeneous concerns about the future of the Internet that increasingly seem to be lumped under the rubric of "non-neutrality" or "network discrimination," which both obscures the plurality of potential problems and begs the question of whether, assuming a policy remedy is necessary, "neutrality" regulation is actually the ideal silver bullet response to all these diverse concerns. If there were no downside to mandated neutrality—if there were no risk of opening the door to regulatory gamesmanship, and if every imaginable deviation from neutrality were plainly harmful—then this might not be such a big deal. If there are potential downsides, though, it behooves us to get a little more granular and look specifically at what we're concerned about, and whether there are less sweeping mechanisms that would work to address the problem.

The ACLU puts the threat of content-based restriction of expression at the forefront of their argument, but this also seems like the concern with the weakest empirical basis, even in a relatively oligopolistic broadband market. First, to the extent that content-based filtering would be executed by means of Deep Packet Inspection, it would almost certainly run afoul of the Electronic Communications Privacy Act, which permits carriers to "intercept" the contents of a communication only when this is a "necessary incident" to the provision of their service. As my colleague Tim Lee lays out at greater length in his excellent paper "The Durable Internet," there is ample evidence that consumers will react with enormous hostility to efforts to literally cut off their access to the sites they want to visit.

If we're worried about wholesale blocking of domains, then, I think transparency-based regulation should be sufficient. That is, an ISP claiming to offer "Internet access" shouldn't be able to restrict access to a site while making it look as though it's the result of some kind of technical problem—perhaps even the blocked site's fault. On the other hand, if Comcast wants to openly and transparently offer the option of a whitelisted "family plan" to conservative parents who don't feel up to fussing with client-based blocking software, that strikes me as the sort of limitation on "expression" that is neither a serious threat to the larger Internet architecture—the effect is only to substitute for filtering the parents would do client-side were they more tech savvy—nor a proper civil liberties concern. Again, I expect a transparency requirement would be sufficient to preclude misbehavior on this front precisely because most consumers don't want their carrier deciding what sites they're allowed to access, and this, more than the fear of pressure from advocacy groups or even the FCC, will tend to make ISPs hesitant to do so if they can't do it covertly. At the very least, again, if there are potential downsides to neutrality regulation, I can't fathom why you wouldn't try this more modest step first and watch to see if some more radical remedy is necessary.

Of course, consumer pressure is more effective in competitive markets, and as Stanley notes, if you focus on wireline broadband, the picture is not that encouraging in much of the United States. But the fact that wireline may have the characteristics of a natural monopoly doesn't mean that last-mile broadband necessarily does: What sectors are "natural" monopolies turns out to be highly contingent on the available technology. As 4G wireless networks roll out, and as users consider the appeal of cutting the cord, the stranglehold of the incumbent monopolists and duopolists is attenuated. Wireless broadband, of course, is not a perfect substitute—fiber will probably always have a significant speed advantage—but imperfect substitutes can exercise competitive pressure too. Rail is a natural monopoly, but Amtrak still has to worry that dissatisfied consumers will drive, fly, or take Bolt Bus—even though these alternatives differ from train travel along multiple dimensions.

Moreover, specific deviations from neutrality that respond to consumer demand may themselves help secure the very competition Stanley and I both agree will help discipline carriers and keep deviations from neutrality limited to those that serve genuine consumer interests. So—and consider this a strictly illustrative hypothetical, please—Netflix now accounts for something like 20 percent of downstream bandwidth at peak home use times. Probably there are no small number of people who'd find it appealing to cut the cord if they were assured they could come home to a movie or an episode of Firefly streaming smoothly in HD. Their cable provider, of course, can guarantee this by bundling your Internet with a dedicated video service running over the same pipes—and, of course, no pretense that there's any parity of treatment between those two types of "traffic." It's at least conceivable that permitting similar bundling and cross-subsidy between wireless broadband and Netflix could hasten the demise of the effective wireline duopoly that exists in many markets, eroding the very conditions that undergird the argument for fearing non-neutral routing could be anti-consumer.

Now, to be sure, you can paint a doomsday scenario based on extrapolation from this model that I find every bit as unappealing as Stanley does: A Balkanized Internet on which every ISP has exclusive deals within one player in each online service category to provide high-bandwidth routing, while the rest of the Net limps along at speeds too slow to make innovative services viable unless backed by big corporate money. (Though this would really be a concern about innovation, not free expression: There's actually very little reason to fear that deliberate viewpoint discrimination by ISPs under transparency rules is either likely or, more to the point, feasible.) If this were to start to happen on a larger scale—despite the demonstrable preference of most consumers for an open Internet over such a curated walled-garden model, it would be worth revisiting the question. But to impose architectural mandates in advance of such experimentation—to assume a priori that any and all deviations from neutrality would impose such great costs to expression and innovation as to trump any possible consumer gains in price or quality of service—seems very much contrary to the spirit of end-to-end.