Competition Can Protect the Open Internet

Thanks to our hosts for inviting me to contribute to this symposium. Like Adam, I found Infrastructure to be a thorough, in-depth examination of the economics of infrastructure. There’s a lot in the book I agreed with, but discussions are more interesting with some healthy disagreement. So I want to focus on an area of disagreement: chapter 13, which focuses on network neutrality. Brett Frischmann makes the strong claim, likely to be controversial even among supporters of network neutrality regulation, that regulations would be needed even in a competitive market for Internet service. I disagree.

The core of Frischmann’s argument is that neither Internet users or their ISPs (for example Click Broadband) can fully know or capture the social value created by public participation on the Internet. He’s clearly right about this. For example, if I contribute to a Wikipedia article, I thereby increase the welfare of everyone who subsequently reads that article. Yet there’s no way for my ISP to capture a share of this “spillover” benefit of my Internet use to other Internet users. So theoretically, there’s a risk that my ISP’s zeal to maximize its private profits could cause it to manage its network in ways that discourage me from engaging in activities, such as contributing to Wikipedia, whose total social benefits exceed its total social costs.

This is a theoretical possibility, but in most cases it’s not a realistic one. Certainly, it’s not a realistic danger in my Wikipedia example. For example, it would be difficult for my ISP to configure its network so that I could read Wikipedia (allowing me to reap the benefits of others’ contributions and thereby increasing my willingness to pay) but not contribute to it. The network traffic that constitutes me reading Wikipedia doesn’t look very different, from an ISP’s perspective, than network traffic that constitutes contributing to Wikipedia. And Wikipedia has every incentive to discourage such discrimination by, for example, encrypting its website.

This is deliberately fanciful example; it’s hard to see why an ISP would try to discourage contributions to Wikipedia. But the point is a general one. Frischmann treats the private and social values of the Internet are separate entities and assumes that ISPs have wide latitude to pick and choose between them. But in the real world, private benefits and public spillovers are deeply connected, and the network management tools ISPs have available are far too crude to effectively capture one without the other.

Frischmann seems to assume that consumers will only be willing to pay for an open network if they understand its benefits in explicit terms. Since few are likely to have that understanding, he assumes users will under-pay for Internet openness. But even users who have no explicitly conceptual understanding of network neutrality can still value it at an implicit, practical level. Users can notice that the open Internet has a lot more cool stuff—Wikipedia pages, YouTube videos, niche news sites, exotic pornography—than the various walled garden on offer. A big part of the reason users like the Internet so much is that Internet users enjoy all those positive spillovers being generated by other Internet users. Given the crudeness of network filtering technologies, restricting Internet freedom inevitably means depriving users of access to compelling content and applications; and that, in turn, will reduce consumers’ willingness to pay for access in the first place.

And it’s important to remember that this isn’t just a theoretical discussion. We’ve run this experiment before. In the 1990s, consumers enjoyed robust competition between proprietary networks like AOL and Compuserve and open alternatives like the Internet. Initially, the proprietary services had all the advantages—stronger brands, larger user bases, familiar features like telephone support, capital to invest in content and applications. Yet users gradually discovered that the open Internet was way better than the proprietary alternatives. Users didn’t switch to the open Internet because they read Larry Lessig or Brett Frischmann, they switched because the Internet had more cool stuff on it. But that was, implicitly, a vote for network neutrality.

If it was difficult for a closed network to compete with the Internet in the 1990s, it would be far more difficult for it to do so today. So the only question is whether there will be enough competition that users will have the opportunity to choose. I’ve become more concerned about the dangers industry consolidation threatens to the open Internet. But we should address that danger head-on, by pursuing policies that preserve and enhance competition. Direct regulation of network management practices should be undertaken only as a last resort—and we’re still a long way from that point.

You may also like...

1 Response

  1. Brett Frischmann says:


    Thanks for participating. I expected you’d disagree on these grounds. I refer to your 2008 paper that makes a related point. I do appreciate your argument. It seems to me that your confidence in competition hinges on at least two premises that I am not so confident about: first, the crudeness of network filtering/prioritization/discrimination technologies, and second, a belief that consumers will continue to choose open over closed infrastructure. While I discuss the private incentives of networks to discriminate and the argument that consumers/markets will discipline networks that choose to discriminate, I do not fully address these two premises. So I’ll briefly do so here.

    Regarding the first, my understanding is that the suite of technologies are not so crude and that they have evolved considerably over the past decade and are continuing to do so; moreover, crude technologies can be a significant problem, unless consumers notice and act as you predict they will, which brings me to the second premise. I do believe that in this particular context, consumers are more appreciative of the benefits from openness. But that appreciation may not be sufficient to discipline discriminating networks for two reasons. First are the demand-side failures discussed at length in the book. I actually agree with you that private benefits and public spillovers are often deeply connected, but one of the points of the book is to show how many of the relevant spillovers from public and social goods spill off-network and affect people besides consumers. Even in an environment where infrastructure consumers (users) appreciate spillovers from public and social goods produced by other infrastructure consumers (users), those consumers’ willingness to pay in the infrastructure market still may fall well short of social demand because of the many spillovers that flow off-network. Basically, consumer demand for “open infrastructure” is still derived demand and resistance to network discrimination reflected in consumers’ willingness to pay for open infrastructure may not be enough. It might have been in the 1990s, might be (for some time), or it might not be, but as I argue in the book, “there is too much at stake to bet on private strategy coinciding with public strategy” or for purposes of this discussion we might rephrase it as “there is too much at stake to bet on private demand coinciding with social demand.”

    Second, I am simply not as confident as you that consumers will behave as you predict. They may notice and react to certain losses of “compelling content and applications” but not others; or they might not react at all; plus, those applications most likely to garner a reaction are probably less likely to be deprioritized. [I feel like there is more to be said here, but the comment is getting unwieldy.]

    Finally, I should note that I also support pursuing policies that preserve and enhance competition. The argument I make in the book is that competition does not alleviate the demand side market failures and consequently, the case for network neutrality remains strong even if we (heroically) assume robust competitive markets.