Friday, December 28, 2007

Open Systems, Muni systems, and Lessons from Singapore

A Problem
Advocates of muni telecomm are often met with the blanket, essentially ideological, claim that municipal plans will fail because "everyone knows" that government-run enterprises will always lack the competitive advantages of private businesses. It's hard to greet such claims with anything other than exasperation: anyone who thinks that the duopoly represented by corporations like AT&T and Cox has produced efficient pricing or any sign of innovation just hasn't been paying attention.

Customers of telecommunications companies simply haven't seen the benefits of "free enterprise" that competition is supposed to bring. The telecomm market looks like market-segmented, minimally competitive duopoly and produces results that look a whole lot more like staid, expensive monopolies than anything that might result from a real competitive marketplace.

Lots of folks have noticed this painfully obvious fact about the current telecomm market and in some places are even trying to do something about it. Lafayette has one solution. Singapore is trying another.

Singapore Tries Honest Problem Solving
Singapore is about to invest in a truly radical plan to build a world-class, high-speed network and to do it by encouraging real competition in the telecommunications market. (See 1, 2) Naturally they start by mandating and subsidizing the construction of a fiber to the home network. Beyond that it gets really interesting. Their plan takes yet another stab at inducing competition in the fundamentally natural monopoly wireline broadband market. Competition—when it works—provides cheaper prices and drives innovation. Lot's of country's have tried for that golden ring—and failed. (The American FCC's attempts have been particularly laughable.) What is interesting about Singapore's design is that it might work.

It is worth noticing how far they had to go to have a hope of developing real competition. Consider the starting point: Most networks world-wide are fundamentally vertical monopolies. One company owns the physical network, manages it, and sells retail services to end users. Think about your phone or cable company and you'll get the basic idea. The minimal competition between phone and cable companies over the new internet services should not be allowed to obscure the fact that they are both basically monopolies with only a sideline internet business that has, at best, only one competitor—not nearly enough to develop a competitive marketplace that would yield the benefits of innovation and low prices. As digital services converge over integrated data networks it remains to be seen whether even the current inadequate level of duopoly competition will be maintained...and a lot of history that argues that it will collapse back into a simple monopoly.

But everyone wants competition and its benefits. Singapore wants competition. But Singapore wants it badly enough to try and get it realistically.

Being realistic involves admitting that the basic fiber, the physical network, is a classic natural monopoly. But beyond that evidence of clear-headedness Singapore also seems to recognize that operational layers of the network determine what sorts of application services can be offered at retail and that retail providers need to be able to count on a responsive middle layer provider.

<skipable bit>
<time out for a bit of background>
A typical large-scale network is built up of multiple, but integrated, levels. One way of looking at that is to see at the "bottom" a hardware base built up of the actual fiber and low-level switching. Up from that you have protocols and translation devices/routines that knit together the data from the low-level physical layer. Both of these are pretty much invisible to any end-user. On top of that you have applications that show their face to users of the network. Let's call that 1) the physical layer, 2) the network operations layer, and 3) the applications layer. (This 3-layer description, as forbidding as it might seem, hides an awful lot of complexity. The canonical way of looking at network design is the 7-layer OSI description. That hides less of the complexity. Sophisticated readers should feel free to substitute OSI layer 1 for "physical;" layers 2-3 for "operational" and 4-7 for "applications.")
</time out for a bit of background>
</skipable bit>

Singapore is separating the physical and operational level into two different, unrelated monopolies committed to selling the same services to all retail providers at the same price. The retailers would then be in a position of making all their profit from the quality and the quantity of services they could convince consumers to buy.

Structural Separation: Keeping the Monopoly Owner Honest
Singapore is structurally separating the physical, lower level from the upper operational and application levels by creating a completely independent network company to build and manage the physical network (cleverly called NetCo). That sets things up so the only way the owners can make more money is by providing more value to the wholesale renters of their physical capacity. If you can offer more value more efficiently you can sell more capacity for a better price. And that, to repeat, is the only way to increase your take. This is a simple, reliable, structural solution to the problem of a monopoly owner using their control of the medium to eliminate or forbid competitors. The physical network owner cannot be motivated to manipulate the network to benefit its particular set of retail services if it doesn't own any such will not be allowed, for instance, to sell phone or video services to end users and so has no motivation to structure its network to favor, for instance, cable TV at the expense of DV (Downloadable Video). By making the monopoly network owner's profits depend solely on motives that are aligned with the public's interest the task of regulation is much easier. All you have to worry about is enforcing rules that require everyone to be charged the same for the same service. (This is much of what lawyers mean when they talk about Common Carriage rules.)

Operational Separation: A Balance of Powers
The most unusual (and least clearly specified) part of the plan is separating out the operational division of the network into its own independent company. Most structural separation schemes make this the property of the network owner or allow retailers to install their own equipment at the operational level. The problem with the first solution is that investing all the control in the conservative utility would make it less likely that unproven but potentially innovative middle level equipment would be installed, lessening the hoped-for benefit from innovation. On the other hand letting the retailer install whatever equipment they want on fiber strands they have rented virtually ensures that incompatibility will emerge on the network and pretty much ensures that some classes of equipment will be wastefully duplicated—lessening the hoped-for benefit of lower prices.

Singapore's solution is to provide for a monopoly operational company (cleverly called OpCo) that must maintain a separate existence, board, and identity but which retail owners can own pieces of. Presumeably the Singaporeans, being committed structuralists, think that such an ogranizational schema will eliminate wasteful duplication and will tie OpCo to the more innovative retailers. Now this isn't nearly perfect: it would let powerful incumbents on the network control the provision of new middleware and help them keep out smaller new competitors that would threaten their developed markets....but while imperfect, this is a solution that at least makes a stab at controlling the worst defects of previous attempts to foster competition and encourage both lower prices and innovation at the middle level.

Retail: The Evolutionary Melee
The hope, of course, is that by minimizing costs at the physical layer by putting a free-to-be-careful and conservative utility at the physical level, and by structurally maximizing low pricing and innovation at the middle level the crucial retail applications level will attract many competitors who will have no choice but engage in a ruthless evolutionary melee in order to survive. Consumers would reap the benefits of low prices and innovative, powerful services.

It Might Even Work—At a Price
It is clever. It might even work.

In Singapore. As a National policy. And anywhere that the national government is willing to subsidize a full new fiber network to the tune of 25% of its total costs. Anywhere where it can dictate the terms of the new networks operation in order to ensure the incumbents don't kill competition in its cradle. (The incumbent phone and cable companies are among the bidders for the new network.)

Notice that this plan involves the people paying a substantial subsidy for the development of a system that private corporations will end up owning. And those corporations will reap all the eventual profit.

That's a deal only a authoritarian, corporate state like Singapore could love. It's a high price to pay.

What people are seeking when they try something so draconian is to realize the promise of competition in a framework that has been fundamentally hostile toward competition. (And well, maybe, to provide a little grease for their friends...but let's try to be generous). The hoped-for benefits are lower prices and a high level of innovation. Both are presumed to emerge "naturally" when you structure a natural monopoly so that the owners' self interest is deployed in the service of the eventual consumers.

But there is another, simpler, surer, way to align the owners' self-interest with that of consumers.

Lafayette's way

You could make the consumers the owners, by the simple and time-honored device of making the natural monopoly a public utility. Then the owner-citizens would have no motivation at all to exploit the consumer-citizens...since they'd be one and the same. They could ask themselves for, and expect to get, lower prices and the sorts of services that appeal to them.

I can't fathom why that can't be a national policy as easily as giving away the farm.

No comments: