Why Internet pricing is on ongoing tragi-comedy Andrew Odlyzko The latest controversy over telecom pricing is just another episode in the never-ending tug-of-war between conflicting incentives. That the same questions arise repeatedly, going back centuries to the days when ubiquitous and affordable snail mail was a revolutionary innovation, is a source of either amusement or dismay, depending on one's point of view. It is especially entertaining to watch the same organization battling strenuously on opposite sides of the same issue. Thus, for example, the cable company Time Warner Inc. has argued that it needs to charge by the byte for Internet access (a form of unbundling, as we'll see below), while at the same time fighting tooth and nail against unbundling its packages of video channels. The repetitive nature of the fight over pricing owes much to the inherent intractability of the problem, as well as lack of a clear discussion of the factors underlying the controversy. Metered rates have obvious attractions to service providers. Their basic incentive is to charge according to value, ideally collecting $5 for the 45-character text to your significant other, warning that you are stuck in traffic and won't make it back in time for dinner, while letting you watch a crummy 2-GB movie for 5 cents. There is nothing immoral or unethical about such moves, but they are constrained by technical feasibility, as well as by competition. Furthermore, they have almost always been limited by legal and regulatory measures (such as common carrier rules). The reasons for such limitations are grounded in public concerns about fairness, as well as about not introducing distortions in the rest of the economy. The problem is that untrammeled pricing power (especially by monopoly or oligopoly telecom or transportation providers) has effects similar to that of a lack of secure property rights in limiting innovation. Fine-grained pricing practices have also been limited by public preferences, in ways that have almost always been underappreciated by the telecom industry. Simple, ideally flat-rate, pricing has been preferred by customers. There is abundant evidence for this preference from decades of telecom experience, as well as from more recent behavioral economics research. Quite often, and almost invariably to the surprise of telecom managers as well as policy makers and economists, there are benefits from such pricing for service providers as well. People are often willing to pay more for flat rates than they would for the same consumption under metered ones, costs of monitoring and billing are eliminated, and so on. Much of this is well documented; see my decade-old paper on the topic. Usage almost invariably rises under flat rates, but that is often of substantial benefit to service providers as well. The one advantage of flat-rate pricing that appears to be understood the least is that it is a form of bundling. Bundling, selling several goods or services for a single price, as in many cable TV channels in a single package, has been practiced since times immemorial. It can often be shown to be of advantage to both sellers and buyers, even in the conventional economic point of view. As an example, suppose that Alice is interested in just two sites on the Internet, say, Facebook and YouTube, and her monthly traffic with Facebook might come to 100MB, while that with YouTube might be 5GB. And suppose that she is willing to pay $20 per month for access to Facebook and $10 for YouTube. If she is to pay on a per-byte basis, the optimal pricing for her service provider in terms of maximizing revenues is to charge 20 cents per MByte, in which case Alice will only use Facebook, and will pay $20 a month. On the other hand, if Facebook and YouTube are bundled together (in the form of flat-rate access to the entire Internet), Alice should be willing to pay $30 a month, which should make both her and her service provider happy (provided the latter has low transport costs). Both will get more than they could obtain with metered rates. The optimal pricing, even when the service provider has complete control, is not a simple matter. There are conflicting incentives, and the cost structure varies, so that there is no /a priori/ reason to say that what should dominate in wireline will be the best solution in wireless. But all precedents argue that the industry will continue floundering around, blissfully unaware of history and basic economics. Hence we are destined to see many more acts of this tragi-comedy.