Nondiscrimination: Not as Easy as It Seems at First Blush

The highlight of Chairman Genachowski’s speech launching the FCC’s new network neutrality initiative is the proposal to add two new principles to the four initial principles announced in the FCC’s 2005 Internet Policy Statement.

The latter of the two new principles is the less controversial. It states that providers of broadband Internet access must be transparent about their network management practices. Clear disclosure is a principle with which essentially everyone agrees. Even the most ardent supporter of deregulation recognizes that markets cannot work unless consumers have clear information about the services they are buying. To say that it is less controversial is not to say that it will be easy. Network providers have long cautioned that too much disclosure about their precise network management practices may simply provide hackers with a roadmap of how to short circuit safeguards designed to prevent a small group of users from overloading the system. I suspect that the details of exactly how much disclosure is required will prove more difficult to work out than many expect.

The other new principle should make Tim happy, as it is one for which he has been advocating for a long time. It states that broadband providers cannot discriminate against particular content or applications.

Although principles such as nondiscrimination appear simple and possess a broad superficial appeal, a closer analysis that bears in mind the history of previous attempts to enforce nondiscrimination suggests that it will prove extremely difficult to define and enforce.

As an initial matter, scholars and policymakers universally accept that charging different prices is not discriminatory when those prices reflect real differences in cost. A classic example is transportation costs. If customer A is located farther from the manufacturer than customer B, we would expect the total delivered price for these customers to vary, with customer A also having to pay the difference in shipping costs. Similarly, we expect that the price paid by a customer who buys in truckload quantities to reflect lower per-unit transportation costs than would a a customer who buys smaller amounts.

Congestion represents a more subtle source of cost differences than transportation. Consider the congestion costs that arise in a restaurant. If an additional customer arrives at a restaurant that is empty or nearly empty, the impact is negligible. The restaurant has plenty of extra tables and staff to accommodate the additional load. The situation is quite different if the customer arrives at a time when the restaurant is already quite crowded. The customer may well have to wait for a table to open up. In addition, the customers who are already in the restaurant may have to wait longer for their server to check on their table, put up with greater proximity and noise from other diners, wait in a line for the restroom, and endure other costs. In short, the arrival of the customer imposes congestion costs on everyone in the restaurant.

Restaurants have an easy solution to this problem: they offer early bird discounts, thereby inducing customers to dine at times when they would create fewer congestion costs. Although it may not be as intuitive as transportation costs, this is a real cost-based differential that nondiscrimination regimes should find unproblematic.

In addition, scholars and policymakers also recognize that charging different prices is not discriminatory when the products that the customers receive are of different quality. This too is quite intuitive. Anyone who receives a superior quality meal expects to pay more for it.

Both of these considerations already render the nondiscrimination analysis quite murky. As anyone who has studied the history of telephone regulation knows, determining how much a service costs has long proven incredibly problematic, requiring difficult assessments in changes in production technologies and business conditions (including classically the choice between historical and replacement cost) and the analytically thorny problem of allocating of the cost of assets that are shared by more than one service. Moreover, determining which prices represent valid cost differentials is particularly difficult when production technologies vary from firm to firm and when industries are undergoing rapid technological change. In addition, as the Supreme Court recognized in Trinko, the number of ways in which the quality of access to Internet service can vary are myriad, which will further complicate any attempt to enforce a nondiscrimination mandate.

The restaurant example also suggests another type of discrimination that I suspect that the FCC will have trouble addressing. Many restaurants offer senior citizen discounts. These discounts are not cost based. Senior citizens don’t cost any less to serve than other diners eating at the same time. Nor are these discounts quality based. The senior citizens are ordering off the same menu and receiving the same food as other diners. The real difference is that senior citizens as a group are typically more price sensitive than other types of diners. Therefore, rather than charging everyone the same price, which would lead many senior citizens not to dine at all, restaurants instead charges a lower price to the class of customers it knows to be more price sensitive. This is not based on a difference in cost, but rather in a difference in the elasticity of different groups of customers’ demand. Those familiar with regulated industries will recognize this as the logic underlying Ramsey pricing. If executed properly, this form of discrimination can permit more people to enjoy the benefits of the Internet and allow more last-mile competitors to survive than would forcing every network to offer only a single class of service.

There is much to like in the Chairman’s speech. It recognizes that important innovation occurs in the core of the network as well as the edge. It accepts the propriety of curbing the bandwidth consumed by heavy users during times of congestion. It acknowledges that any nondiscrimination should not preclude filtering content for spam or illegal content. Most importantly, it notes that managed services can play an important role that benefits consumers.

That said, the devil will be in the details. In particular, it remains to be seen how the FCC will solve the problem of defining what represents justifiable and nonjustifiable discrimination in an industry in which cost and quality vary widely. In addition, the FCC will have to create rules that preserve the ability of network providers to engage in demand-side discrimination of the type suggested by the senior citizen discounts offered by restaurants. The extent to which the FCC’s proposed rules take these considerations into account will go a long way to determining whether they will ultimately benefit consumers or harm them.