A commons for spectrum? Larry Staton pointed me to this legal article on replacing spectrum auctions with a "spectrum commons." It begins with a rather nice allegory, comparing the FCC's regulation of spectrum with the government regulating how people speak at a party. Instead of this intrusive auction business, the paper argues, why not just ask everyone to lower their volume, or have local law enforcement get people to quieten down if things get too rowdy? Ultimately, the paper argues, its better to offer spectrum under a "Common Property Regime."

I still don't find the arguments convincing. There are two central issues here: 1) is spectrum rival/scarce and 2) how easy is it for parties to coordinate access.

The scarcity angle is tricky -- "bandwidth" is a mix of transmission, processing, and local storage, and electromagnetic spectrum adds the fourth dimension of attenuation. The various new efficient spectrum technologies ("spread spectrum," "ultra-wideband") essentially shift load from the transmission component of bandwidth to the processing component. This is as it should be given how available spectrum is constant but processing cost continues to fall as Moore's Law grinds on. But this doesn't say spectrum is suddenly "non-rival," it just means that $5 worth of terminal circuitry can now do the work of $1,000,000 worth of 1950s technology, so many applications no longer need as much spectrum as they ought to. And optimizing between transmission, processing, and local storage is best done through a market.

The FCC's spectrum auctions are a step in the right direction--if a party bought spectrum to do job X, and now needs less spectrum to do that same job, they now have incentive to sell that excess to someone who can better use. By contrast, if it was allocated to them, they would simply not bother to petition the FCC for a reallocation.

Sadly, the FCC remains in the business of allocation and restricting the aftermarket. It turns out that spectrum auction winners cannot resell their spectrum to others (except under particular circumstances) and the FCC limits what use a particular slice of spectrum can be put to, limiting innovating spectrum re purposing.

I would argue that the solution to this is more complete spectrum ownership, not less. But the paper argues

If the value of the spectrum cannot be measured ex ante, then there seems to be no basis for saying that the final winners in a particular auction are necessarily those who valued the spectrum most highly. One suspects that the definition of "value" being used here is circular--the highest-value-user is defined as whoever happens to win a particular auction, and then the auction is praised for its power of discovering the highest-value-user...[Auctions] functions as a barrier to entry for those mid-level companies that might have the most innovative ideas about spectrum usage [but can't afford the spectrum].

These assertions are false. Firstly, no one "happens to win" an auction the same way one "happens to win" at Bingo--auctions winners are those who stump up the most cash, and if anything tend to most wildly overestimate the value of the lot. If cold hard cash isn't a good way to uncover what someone thinks something is worth, I don't know what is. Secondly, mid-level (and even small!) companies can borrow against predicted future earnings to buy spectrum. US capital markets have done a wonderful job of democratizing cash and arguing that small companies can't make big investments stands in ignorance of venture capital, junk bonds, warrants, and the other wealth of financial infrastructure that does just that.

Having spectrum restricted to particular uses and forbidding people to resell is particularly harmful to consumers and innovation. For this (like so many other things) we have to thank the content industry, particularly early radio and television that, through appalling cronyism with Congress, took large swatches of public airwaves for their own private use.

But back to the paper. I really struggled to understand what a "Common Property Regime" was, and the best I could figure out was that it was some kind of group sharing agreement between incumbents which collectively blocked entry and apportioned economic rents between themselves, all bound by a variety of measures that enforced collusion and punished cheating. In my book this is called a "cartel" and I don't associate them with innovation. The paper sees this as a feature, not a bug, to wit:

In the lobster fisheries of Maine, for example, people are restricted from fishing at all unless they are members of a harbor gang, while each harbor gang has its own territory. In the Swiss alpine common pastures, entry is usually restricted by community residency or by family lineage.

Gangs keeping out new guys and access by privilege of territory or ancestry reminds me of the Mafia.

Earlier I raised attenuation, and I'm going to close on that. Through the attenuation profile of the 802.11x spectrum, it has the virtues of being 1) narrow enough to be local while 2) broad enough to be useful. Oh, and it's 3) connected to ridiculously powerful processors. These characteristics, along with its dramatic installed base and grassroots support, means its an ideal exception to the rule where a "Common Property Regime" may work. I've been reading about how Starbucks, who wants to sell neighborhood WiFi is trying to shut down an existing (free) community network, and this strikes me as the sort of disputes clear ownership would beg. Let's see how the polis sort it out amongst themselves.