New York — Senator Robert Bulkley, of Ohio, has made a proposal which is certainly worth considering.
It is as clear as daylight that, to bring about any sort of recovery, somebody must start some new sort of business or some extension of an old business.
It is also clear that nobody is in sight right now who has any notion of doing that — at least not in time to do this country any good as a depression cure.
There is one business which is a public business but is also a private one. This is the road-building business. The Government pays for the roads and hires the contractors. But the roads are built usually by private contractors and with materials furnished by private manufacturers.
If there is one thing needed in this country now, in view of the development of the automobile, it is express highways running east and west and north and south. Why, therefore, cannot the Government go into the business of building these highways?
—Washington News, February 9, 1938
Tough Times for Local Exchange Carriers
This week, the headlines seem to be full of fresh doom and gloom for wireline carriers, who employ people in every congressional district across America. Sooner or later, someone is going to call for Congress to tap some of the hundreds of billions in 2009 economic stimulus to help the LECs through troubled times, save lots of jobs, and preserve the way we do business in our critical last-mile communications infrastructure.
Is this wise? Is there a better way?
Customer Owned Networks
I stumbled across a really interesting paper this week, written by Derek Slater and Tim Wu, and it set me to thinking. Slater is a policy analyst for Google, but not writing on their behalf; Wu is a professor at Columbia Law School.
They propose an interesting thought experiment. What if you could own your internet connection, instead of leasing it from a service provider? In their "customer owned network" scenario, homeowners would literally purchase a strand of glass, running from their house to a common point of presence (PoP), where multiple providers would compete for the right to sell services over the fiber. The property rights you'd acquire in the physical network would be quite literal — when you sell the house, your strand of fiber goes with it. It's an improvement to your house.
The optical trunk, or bundle of fibers, that winds its way through your housing development, town, or county would also be yours, owned as a cooperative by all of the individual strand owners, who would pay the equivalent of "condo maintenance fees" to cover repair costs (backhoes do get hungry, you know).
Slater and Wu call this the "Homes with Tails" model, and a 400-home customer owned network is already being built out in Ottawa, in a trial run by CANARIE. In that experiment, the electric company brings your strand to the house, and charges you an additional $0.02 per kilowatt-hour over five years to pay for it (thus encouraging you to save energy in the process).
Have it Your Way
Their key observation about a customer-owned last mile network is that you, the homeowner, would have complete freedom to light your fiber however you like. You can make whatever deal you can make, with whatever provider you can meet at the PoP, who is willing to offer you service over your fiber. You'd strike a deal, they'd set a price, they'd lease you some link-layer equipment for your house, they'd light the fiber and offer you services under contract. But they wouldn't have anywhere near the same kind of leverage that they would if they owned the glass.
Don't like their pricing? Don't like their lack of network neutrality? Don't like the cut of their jib for whatever reason? Just cut 'em loose and get another service provider. It's your fiber, all the way from your house to the exchange point. They meet you where you have the maximum number of choices (the PoP), rather then the fewest (your doorstep).
How much bandwidth would you have on tap? Well, using current laser technology and cheap commodity equipment, it's trivial to push one to ten gigabits per second over that strand. And it gets you to a PoP where you could be paying a provider of your choice tens of dollars a month for the right to sustain those gigabits, and burst to tens of gigabits, across the public internet. Yes, in a competitive environment, where you buy the data service and own the fiber to the exchange point, that's how little you might pay for retail gigabit connectivity, if declines in wholesale internet transit prices continue on their current trendline. And as the switching and transmission technology curve matures over the 15- or 30-year life of your home fiber investment, maybe you end up with terabits per second of capacity. Futureproof, baby.
Getting Off the Launchpad
So this led me to think about the challenges of launching customer-owned networks, most of which focus around the "first mover disadvantage" — if someone starts a program like this in your neighborhood, your best play is to wait until they (the rich early adopters) have paid the freight. Then you jump in a few years later and pay only incremental connection costs. But because the startup costs are really significant, and because nobody wants to look like a sucker, this in essence means that such projects never get built, except in the easiest-to-reach neighborhoods, and then only by a service provider who hopes to extract a monopoly rent over the fiber. Not what we're looking for!
What's worse, the current economic climate also makes it hard to envision anyone starting a new fiber buildout of any scale. Credit has all but dried up, making it increasingly difficult for anyone to justify sinking huge capital investments into projects that might not return dividends for years, if ever, and then only in the densest markets, and then only depending on dodgy assumptions about penetration rates and revenue per user. Verizon is bravely pressing ahead with its FIOS buildout, estimated to cost $20B over 2004-2010 to reach just 18M households in five east coast markets. But then, Verizon just laid off thousands of people, a few weeks before Christmas, as business softens and their competition strengthens. And Verizon is probably among the strongest of the wireline carriers, most of whom are too focused on their collapsing margins and inability to roll commercial paper in this credit environment to think about undertaking significant new fiber-to-the-home projects. Especially if the model (customer-owned fiber) prevents them from exacting monopoly rents when it's all built out.
Time for a New Deal?
You have to ask: wouldn't this be an interesting use for a few hundred billion dollars of government investment? If the new Obama administration is looking for an investment project that would stimulate the economy in the short term, and build a completely new set of national capabilities in the long term, a national-scale customer-owned physical network would be an intriguing possibility. The government would provide the cash and the muscle to build the network, and then transfer it, strand by strand, to its long-term owners: those who own the parcels of residential and commercial property where the fiber terminates.
Start with a mixture of rural areas and midwestern cities: economically depressed areas where carriers have arguably underinvested in the residential fiber plant relative to what might be in the long-term public interest. Send a surge of engineering companies into the field, working in cooperation with municipalities and the utilities to secure the rights of way, open the streets, install the ducts, climb the poles, string the fiber, reach the houses.
You'd have to cover a few million miles of roads, at an average cost of tens of thousands of dollars per mile. You'd have to build some reasonable number of regional PoPs where service providers can meet customers, and maybe subsidize more fiber to help the independent providers carry aggregated customer traffic back to traditional internet exchange points in the big cities. But because it's "just fiber" (no customer premises equipment, no switching, no nothing), and because you're the government, able to lean on regulated industries and secure rights of way fairly cheaply, you're still talking tens of billions of dollars, maybe a couple hundreds of billions, to cover really large parts of the country.
And because we're still focused on leaving this a customer-owned network, ownership of the fiber network would remain with the individual citizens who own the pieces of property where the strands terminate. At the end of the project, the government doesn't own the network (individual property owners do), and they don't operate it (individual service providers do, competing for individual property owners' business). The property rights would be quitclaimed to the individual property owners. The costs could even be recovered from property owners over the next 30 years, perhaps in the form of a few dollars per month surcharge built into new residential mortgages, or as a federal tax on services provided over the new network.
Limited Government Intervention, Private Ownership
None of this is new, of course. Bill St Arnaud has been advocating customer-owned network projects in Canada for a long time (he's the fellow responsible for CANARIE's Ottawa trial I mentioned earlier). And compare Fred Goldstein's vision of divesting the CO and local loops from the switching and service platform to create a set of highly-regulated "LoopCos" for truly competitive universal access. In our customer-owned last mile fiber network, some Loopco-like entities would have to bid the local contracts to maintain the fiber plant, run the fiber exchange points, and so forth, all complications that I've glossed over here.
And the general concept of a national information grid and fiber to the home has been floating around for a long, long time, at least since the PITAC studies from more than ten years ago. To my knowledge, however, the concept of a customer-owned national last mile fiber plant has never received serious consideration. What works in favor of building it this time around is the economic climate, and the size of the effort that most people are now willing to envision for the government's role in rebooting the national economy.
The hundred-billion-dollar costs for building this project are not out of line with the other flavors of economic stimulus that have been proposed. The strategic returns on the investment are potentially far higher. The technology for stringing and laying fiber is well-established, the materials readily available, the expertise distributed among thousands of local small businesses nationwide. And at the end, instead of having poured that money down a hole fighting a war, or buying every American a new Chinese-made plasma TV, you've actually changed the ground rules of the domestic internet infrastructure.. and the American economy itself.
How many gallons of gas would be saved by eliminating half the miles commuted to school and work every day? How many pensioners could live an extra year or two at home, watched over by telepresent family, instead of checking into nursing homes? How much safer would our economic infrastructure be from attack, if the infrastructure of our information economy could be spread thinly around the country? What kind of entrepreneurial innovations might emerge, what new industries might flourish, if any two American street addresses could exchange nearly limitless amounts of internet traffic on demand?
I don't know, but I'd like to find out. It would be a highly nontrivial project, and I have glossed over many of the trickiest parts, such as building and managing the exchange points, and determining who gets to colocate and cross-connect there, on what terms.. In fact, building a nationwide customer-owned last mile network would be so nontrivial in scale, and so disruptive of the existing economics of Internet edge access, that it could only be envisioned during a time of national crisis. Would the payoff to future generations of property owners be worth it? We have a rare opportunity to consider the question.
Disclaimer: These opinions are mine alone, and not those of Renesys Corporation.
|Cybersquatting||Policy & Regulation|
|DNS Security||Registry Services|
|IP Addressing||White Space|
Neustar DNS Services
Minds + Machines
Neustar DDoS Protection