I am reading Lessig’s article in Wired on Network Neutrality. I want to translate some of his points, accurately I hope, into more economic terms with which I am familiar. Here is a short recent note by Lessig with relevant comments on McCain’s technology Platform.

I think he has made the point that the last mile providers are in a position to segment their market, at least if you view content providers as their customers. Segmentation allows price discrimination (where content providers are customers) which can increase profits. The providers would like to charge some more than others. Monopolists, legal and otherwise, are in a position to increase their profits by selling the same thing to different customers at different prices. Natural monopolies are often regulated to prevent this. Last mile providers argue that unlike the phone companies of old, they are no longer natural monopolies.

There are several categories of network providers. Internet Service Providers connect the average consumer to the internet. A large internet player, such as Google, attaches to industrial grade access suppliers.

Technologically I think that networks, and in particular the last mile component thereof, can deliver a beneficial variety of services at a variety of costs, especially congestion costs. Things that cost differently will be priced accordingly when those things are a commodity—that is when there is competition to provide them. (This is murky! When a network component handles traffic X instead of traffic Y then the opportunity cost is the value of Y, or at least the cost born by the network operator in failing to have delivered Y, at least on time.) It is as if several sorts of water could be delivered in the same pipes. (Suppose indeed that digital TV were delivered over the phone line as DSL commonly is, but in ATM format. ATM is designed to control latency and bandwidth; IP is not. IP can ride along on ATM as well as it does in many places today. The costs of these two sorts of data delivery over the self-same wire are different because of different upstream hardware requirements.) It would be a pity to regulate against such economic possibilities. Already cable companies deliver video information over the last mile with different latency properties (less jitter) than they deliver internet packets.

One theory of regulating natural monopolies is to attempt to set prices to what they would be if there were competition. Putting these ideas together argues for allowing, perhaps mandating, different services at different prices. It is a difficult task to define these services and indeed a technological challenge to define them; they must emerge thru co-evolution of providers and their customers. Regulation and evolution are at odds with each other.

These different services are an opportunity to do price discrimination, in effect, if the services are tailored narrowly to suit only specific customers. In any case it would seem desirable to have the benefits of competition in both the definition of these services and the provisions of them. This sounds like a very difficult regulation task.

A meagre proposal: regulate to require prices to reflect costs. Of course allocating costs is notoriously difficult even when there are no conflicts of interest.

I approve of network neutrality as described in this Google note. I take issue with network neutrality proposals that ignore quality of service parameters.


Some Tymnet history may be relevant here. Tymnet got started when AT&T’s communications monopoly was deteriorating under pressure of deregulatory winds. AT&T had made many regulations preventing reselling repackaged services from AT&T. These regulations were falling in legal battles. Tymshare never was quite at the forefront of these battles but there had been prohibitions on sharing leased phone lines and it could have been argued that Tymnet leased lines from AT&T and allowed parties, who were unknown to each other to share them. AT&T still held a monopoly on basic movement of data but had planned to serve the interactive Teletype market with non-multiplexed lines at $40/hour. (See ‘TWX’ here.) I don’t know what their profit would have been but Tymnet provided a superior service, based on reselling multiplexed circuits from AT&T at only a few dollars an hour.

Tymnet was a service built upon raw service from AT&T and we substantially under-priced them. CCITT produced the X.25 standard (too) so that phone companies could serve this market efficiently and with standard protocols. Tymnet provided X.25 service as well.


Important neutrality note
NPR on politics of last mile and realities in other countries.