As a fellow free-market libertarian, and also a telecom blogger, I ought to be in agreement with Russell Roberts's critique of municipal "socialist" telecom networks.

Yet I'd like to offer, if not a contrary view, at least a different perspective on muni networks. I believe there are good and bad reasons to build them. But dismissal is not an option.

Yes, there's a clear transfer payment in favour of people who are unwilling to pay the current market-clearing price of a broadband connection. This is a net loss of utility to society, since the taxpayers whose money was confiscated had other, more pleasant, ways of spending this money in mind.

But there are also reasons to consider the free flow of information via telecom networks as a special and unique case. Information goods may not always conform to classical economic models, such as rivalrous consumption.

Let's take the market economics first.

The most important and unusual feature of telecom networks is that much of their value is option value — the ability to cater for unknown applications. E-mail, the Web, P2P, podcasting — none were there at the birth of the Internet. (Search for "End to end principle" and "Rise of the Stupid Network" for more data.) Unfortunately, the desire to price discriminate by a commercial supplier also serves to destroy option value, by limiting messages exchanged to those forms for which a tariff has been devised. The act of price discriminating network traffic and billing is also horiffically expensive. However, the desire to price discriminate is overwhelming, because of the billion-fold difference in value of different traffic types.

Telecom networks potentially generate huge consumer suplus. There is also an elemet of merely co-ordinated self-supply in a muni network. You do no harm by fixing your own blocked sink rather than calling a plumber. By engaging a general contractor, a locality effectlvely self-supplies itself ("unplugs its own data sink"), and retains the full consumer suplus. Again, because of the nature of networks, this isn't something users can do unilaterally.

There may be a co-ordination problem that corporations find hard to overcome. Most communications are geographicaly local, but there's little point in buying a videophone if insufficient of your friends and family across town — some of whom are in poorer households and districs — have one first. The natural unit of purchase of a telecom network may be greater than the single household, by the very nature of being a network.

Those poorer households are hard to cross-subsidise via market means, because the applications, where most of the money is spent, are separated from the connectivity. It's so far proven impossible to make the cross-subsidies between different connectivity buyers and application users work in an effective and fair manner, particularly when users are on different networks. The coordination and transaction costs are too high. (Would a universal service type of outcome arise from a pure market operation? It's debateable — at the very least, it's uncertain.)

Another co-ordination problem is a whole district can't easily sign up to a single network provider at once. The network provider has to engage in house-by-house sale. This is extremely expensive. A municipal network eliminates this transaction cost at a stroke.

There is also evidence that communications infrastructure generates new business and economic activity. This value is not captured by a traditional network operator (e.g. Google's billions of revenue do nothing for SBC or Comcast). But if the general public is the network owner, their interests do align. There is real evidence that the net gains from any form of increased data connectivity are large (see this for example). It is unproven, although not unreasonable, to hope that a forced municipal network will indeed have this effect. Unlike most infrastructure white-elephants, the almost infinite flexibility of a telecom network significantly lowers the risk of such a public enterprise.

Next, there is the practical issue of a market forming over rights of way for new facilities-based entrants. Early phone networks involved massive looms of wire down the street from multiple providers. Planning laws, zoning and access restrictions prevent a proper market forming in this bottleneck.

Finally, there are internal public costs of service delivery. A municiapl network enables those local services to be delivered at potentially lower cost. This is particularly true when access is universal and non-digital service delivery channels can be eliminated.

There is also a non-market side to this. The operation of our democracy requires ideas to flow freely. The increasing dominance of digital media may displace traditional analog alternatives. (When did you last go on a protest march? Read a political pamphlet?) The current experience in Canada with Telus, cutting off access to union web sites that offended it, suggests that allowing a tiny handful of corporations to control this access is potentially dangerous.

Lastly, the invisible nature of information goods and rights of way emprically seems to lead to markets in them more easily being manipulated via political means and regulatory capture. The public don't feel the pain of loss of the services they never got the experience, or the low prices they never saw advertised. The endless pleading and lobbying of telcos worldwide should be a warning signal of a market out of balance.

The trick is, of course, in balancing all these softer issues against the harder deadweight loss of a tax-backed transfer payment. But just because competition is good, it doesn't mean that co-operation (if sometimes coercive) is automatically worse.

Many of you will have heard of the famous Black-Scholes equation for the pricing of financial options. The basic idea is that an option contract to buy a thing at some future date at a fixed price has some potential premium over the current price of the asset because it may allow you to buy at a discount some day (i.e. the strike price is below the market price on the day the option is exercised). Black-Scholes lets you calculate this premium.

A stupid network is like an infinite sequence of discrete options to communicate between connected parties. At the finest granularity, every frame or packet contains a message from a pool of possible messages. A stupid network makes that pool of possible messages larger by deferring optimisation for any particular message type. (For example, IMS networks pre-optimise on SIP, which would exclude a more bandwidth-efficient signalling protocol such as IAX — which could be a problem for wireless devices.) The assumptions for a "stupid" network is that networks have considerable longevity, there is high uncertainty about the future returns of different "messages," and high initial cost — otherwise we could just build more and more application-optimised networks. These respectively map onto "low interest rate," "big standard deviation in returns" and "high current stock price" in Black-Scholes, maybe perhaps, if I interpret things correctly.

I'm wondering if any network economists or general econometricians out there have done the work to adapt and extend Black-Scholes to model communications networks from an option theory viewpoint. If any of you know the answer, I'm curious to find out what they discovered.

As someone with a degree in mathematics, I would regard the problem as "non-trivial", and definitely well beyond my aptitude.

Naturally, the option price may be somewhat different for the owner and user of the network, depending on how good the owner is at price discrimination (i.e. manipulating the strike price) — something the "stupid" network is notoriously poor at. There could be some interesting consequences for muni network advocates. Spreading superfast broadband to people who won't today pay for it at market-clearing prices is a way of foisting options to communicate onto the economy, with the hope that someone will invent crazy new applications to redeem them. Perhaps there's a sounder theoretical basis for muni networks yet to be explored?