I'm moving house, business is booming, we're running a complex project, and I'm travelling a lot. Oh, and two kids and a wife who are hinting they'd like to see more of me. Plus I can't live without some occasional time off and fun too. I don't have any spare time for much right now, but this one is too important not to comment on, so I'll break my long radio silence.
Susan Crawford, one of the sharpest knives in the telecoms policy drawer, reports on the 700MHz spectrum auctions in the US. (Susan, for some reason Blogware refuses to display pages when the referrer is Bloglines -- get 'em to fix it :) ) -- she has several follow-up posts worth reading.
The nub of the issue is whether the auction should mandate some kind of open access regime where users can attach any device of their choosing to the wireless network, not just ones approved by the carrier who supplies the retail connectivity. This is (misleadingly, in my opinion) referred to as a wireless equivalent of the famous Carterphone decision that heralded the break-up of the vertically integrated AT&T landline monopoly (rev 1.0). (Susan's just reporting the proceedings, not advocating the terminology, so I'm just pointing you that way for the succinct background material.)
I don't think the Carterphone precedent is an apt one for wireless IP networks. The fixed network provided an end point with metered access to a (then) noticeably capacity-limited circuit-switched network. You could attach a device with a radically different usage profile (e.g. a dial-up modem, fax machine) and you'd automatically carry the cost of that usage yourself. The network also offered a single line speed at the edge -- you couldn't demand more (at least not without a massive price leap to a business-class T1 line or more). Metering based on time alone works well.
Furthermore, the competition between users for scarce capacity was in the switching fabric and the long-distance network, not in the local loop. Yet when you move to an unbundled regime these cease to be bottlenecks as competitive carriers can simply install their own switches and backhaul -- at least, it works here in Europe, even if the FCC can't figure out how to enforce its own rules in a timely manner.
Wireless doesn't work that way. When you buy an "unlimited" Internet access plan from Sprint or Verizon, they're calculating the likely usage profile based on the capabilities and form factor of the device and pricing accordingly. Yes, in some cases they even nobble these features to dampen demand. They also use contractual terms to say you can't use the device as a modem for a PC, for example.
If you can come along with any data-hungry device and expect to demand the same retail pricing plan, you're going to blow up the business model. The network becomes over-congested, often with low-value file sharing or media download traffic for which there is low user willingness to pay.
Destroying the vertical silo might sound like a good idea to those who feel the telco business deserves some radical change. However, you need to come up with a better idea of creating a market around access to a finite spectrum resource. (And you mesh folk have a lot of technology, economic, policy and usability problems to solve before that changes.) Unlike wireline, there's contention on the access layer. The scarcity is at the edge, not in the core.
The outcome of a "retail Carterphone" will be a shift to metered or congestion-based pricing. This may result in a loss of consumer welfare, as users highly value flat-rate price plans. Flat rate only works as long as the usage curve has a reasonably large and predictable spike in the middle, and you can manage the fat tail via traffic shaping, fair use terms, and contract enforcement. Allowing any device, software or service drives the network capex tail of heavy users without raising compensating revenue.
We've already seen in Korea on their fixed network a move from flat rate to metered, with much consumer resistance. The same has pretty much happened in the UK with BT's wholesale pricing regime on fixed. Due to over capacity of 3G spectrum and networks from the build-out mania, it's too early to say if the huge buckets of data offered by folk like T-Mobile with Web'n'Walk will persist. If they got too popular, and people actually started to use what they've bought, say to watch unicast video on their iPhones, the networks would clog up quickly.
You could instead opt of a "wholesale Carterphone", where anyone can come along and buy wholesale connectivity, and then offer retail pricing for locked-down and non-interchangeable devices. Those wholesale contracts can then be extremely complex if need be, with mixes of usage, time, congestion, device class and other factors such as traffic shaping in the backhaul or content caching services. The users never get to see that complexity. However, we're getting a very long way from the original Carterphone deal where the retail market was opened up to competing device suppliers, and we're a long way from designing a dynamic marketplace for wireless spectrum.
And it's all because of the physics. Spray photons in every direction, and you get a different market structure than guiding them down a strand of glass to a concentration point, because the scarce economic resource is in a different place. Hey, go ask KPN, who cunningly are creating a scarce resource by putting all the electronics into street cabinets (with limited physical space) and selling off their exchanges (where it's easy to put in competing unbundled gear).
My policy recommendation? The incumbents own too much of the backhaul and on-net traffic, which gives them an unfair advantage over new entrants. Do what we did in the UK, and reserve some slices for new entrants (Hutchinson 3G won it), and tilt the field a little in their favour with the interconnect and termination rules. Keep the network auction national so the initial starting condition isn't fragmented with "missing patch" owners extorting everyone else. Don't place any rules or license conditions on how the spectrum is used, except to mandate that squatting isn't allowed. Assume every rule you campaign for will be outweighed by two bought by lobbyists. Allow sub-leasing and resale. Make public safety users pay market price, just like they do for office chairs and other inputs. Even better, make them buy the output safety communications service in the open market, not the input spectrum.
If there's more value in creating an open wholesale network than vertical integration, someone will conduct that experiment without the need for bureaucratic seers predicting the right market outcome. The future is uncertain, none of us are smarter than the market. There's no need to mandate any kind of wholesale structure.
Another fallacy is that raw and pure Internet access is the end-user service. It's not, it's the things you can do with the device. So if I can use a locked-down browser to access any kind of HTTP-sent site, that's not the same as a "go anywhere, do anything" Internet ISP plan. Just ask Steve Jobs. Users don't want to hear "megabyte" once in the store. The YouTube app on the iPhone isn't a bug in the economic model, it's a feature. Users will buy the degree of openness and flexibility they need, not pay for option value of the network capacity that they don't need and others redeem at their expense.
We've already seen the first step with MVNOs and alliances like Sprint-Clearwire-Google and AT&T-Apple. Wholesale and alliances are the future. Fine-grained wholesale for mom'n'pop entrepreneurs will arrive some day without external intervention as long as there are enough competing nationwide spectrum owners (my guess is 5-6) and a vibrant backhaul/backbone market. And the users will get the right devices, plans, connectivity and apps packaged up in convenient form.
Less regulation is better regulation. Particularly in Washington DC. Stop trying to make rules to shape the future, let it sort itself out.
Posted by Martin Geddes at 2:43 PMTrackBack URL for this entry:
http://www.telepocalypse.net/cgi-sys/cgiwrap/mgeddes/MT/mt-tb.cgi/874