I'll start by making the point that several other people already have — that there are quite a few areas of agreement between the two camps here — and just try to quickly set out what I think those are.
The first one is, we all agree that the current system is horribly flawed.
We agree that property and commons are not necessarily mutually exclusive. Gerry [Faulhaber] makes an interesting move, saying, "that's true, but because a perfect universal property system can somehow accommodate commons, and a perfect universal common system everywhere can't accommodate enough property, therefore we should choose property."
But I think that's something of a distortion. We're not here talking about whether any of us finds the writings of Ayn Rand or Karl Marx more convincing. We're talking about the real world. And in the real world, systems are mixed. The question is whether we need to choose one of them as the overarching answer.
The third thing I think we agree on, and it's come up several times, is that there is a great deal of uncertainty here. We don't know with certainty in a property regime who is going to participate in an auction, how the auction is going to go, or what's going to happen. And though we have many speculations and intuitions and examples on the commons side, we don't know with certainty how people are going to behave, or certainly how technology is going to develop in a commons.
So that should lead us to be humble, to build room for experimentation and reconfiguration into any system that we put into place.
And finally, I think we generally agree that scarcity and transaction costs are two crucial variables in trying to make choices between these systems. Those points are implicit in both the papers [presented here]. The Farber/Faulhaber paper talks about scarcity and transaction costs and the coexistence of the two models more explicitly.
So where do we disagree? Well, both of the papers, again Farber and Faulhaber more explicitly, talk about this idea of property as the overarching default regime, and suggest we can accommodate commons within it, essentially the way we have public parks.
That's a nice thing. I really hope it will happen. I think it might. But I don't think it is good enough to base a choice of regime on, for a number of reasons.
First of all, if that makes sense, if a company, a manufacturer, an Intel or a Microsoft, would find it valuable to buy up spectrum and make it available as a commons, why haven't we seen it already? Why hasn't a manufacturer licensed spectrum under the FCC's flexible use auctions that have been going on the last few years, and done this? Why in fact have the property proponents not come up with any real example or analogies for why this might happen in the real world, as opposed to in the realm of theory?
I think the reason is that there is an incentive problem here. And again this has come up through some of the discussions, in David Reed's comments and in particular Yochai [Benkler]'s comments. But the problem is really a matter of ex ante vs. real-time decision making. The notion in the property regime is that you need to, up-front, pay the full value of the spectrum to the person who is selling it... or what the person who is selling it will charge, which if there's a holdout situation may even be more than what its value to them is. In a property/exclusive rights situation, you can make that determination. I want to do broadcast television on a frequency that someone else is using for cellular [phone service], or on a frequency that is now Channel 57 UHF TV, I can decide what the value is to me to offer some other service. But in a commons it's hard to decide ahead of time.
For example, what's the value of the 2.4 GHz band? I'll give you a hint. Before Wi-Fi came along, at the FCC we called it the junk band. It had very little value. Suddenly, in the last three or four years, which has been largely a period of terrible destruction of value and lack of investment in telecom, it's exploded. It has become incredibly valuable; probably much more valuable than it would have been to someone had they had that band for exclusive use. How will we know that ahead of time?
The reason that happened is because there was an open type of environment. Not a perfect commons, but something along the lines of a commons, that let people find it and go the innovator route; let companies deploy technology, do experiments, and come up with improvements. And we've just seen the beginning of what's going on in that band so far.
So the problem with this notion that we can have public parks, or private parks, or some other kinds of combinations, is that it assumes too much about the information that someone would have ahead of time, to get to a policy. And really at a deeper level this idea of parks is the wrong analogy. Parks compete with private ownership for land. They don't compete for use. You don't do the same thing in Central Park as you do in the rest of Midtown Manhattan. That's the point. But with spectrum we're competing to offer potentially the same services. And any allocation to one model or the other, and this is certainly true on both sides, facilitates for certain uses but it precludes certain business models, certain market structures. For example, a market that's based around competition for different kinds of equipment, versus a market that's based on competition between different exclusive service providers. Each choice facilitates certain models and each choice precludes certain models.
I'll talk about this a little bit more in a minute in terms of what it means for scarcity and transaction costs. But think about some other analogies that have that property where there is contention for the actual use. Not parks vs. real estate, but public parks vs. private parks — things like campgrounds and nature preserves. Public highways vs. toll roads. Maybe Linux vs. Windows. I don't have time to go into all of those. But if you think about it, the aspect that all of those examples share is that either the commons regime is the dominant one — we have a lot more public roads than we have toll roads — or there is no choice made ahead of time by government. There is room for both systems to coexist and operate and develop.
Scarcity. The trouble with the way both of these papers and many of the property arguments look at scarcity is that it's done in a very static way. The examples are all about something like the UHF television situation — a defined frequency block where spectrum is either not being used, or it's being used for some sort of low-value activity, and the idea is to transfer it to someone who can use it for something of more value. Or maybe we can aggregate and change the boundaries of those blocks a little bit. But the value of the spectrum — and again this was mentioned in the initial discussions by David Reed and Yochai Benkler — the value depends on the technical architecture of the systems that operate in that spectrum. The availability of the spectrum, how much there is, how much scarcity there is, is a function of those architectural decisions. And the choice of regime influences those architectures.
We don't have unlimited bandwidth under any regime tomorrow in the sense that you can snap your fingers and suddenly have as much spectrum as you want, at all times, for anything, for no cost. But we might have unlimited bandwidth in a commons if you look at it differently.
How long is the U.S. coastline? This is a famous example from the field of fractals. Well, if you measure the coastline from a satellite picture, you get one answer. If you measure it from airplane reconnaissance photos, you'll see a lot more of the twists and turns in the coastline, you'll get a much bigger number. If you walk along the coastline with a ruler, and measure every single twist and turn, and in and out, you'll get a much bigger number. The coastline is infinite, depending on what scale you look at it from.
Now, of course, the coastline isn't infinite. It's not infinite in any practical sense. But how big it is depends on how you look at it. How big spectrum is depends on what kinds of systems are being used in that spectrum. The advantage of a commons system is that it creates incentives for the users of that spectrum to use it more efficiently. To design their existing technologies to be robust for interference that's there, and to design new kinds of technologies that squeeze more capacity out of the system. The property regime forecloses some of those alternatives by putting into place an exclusive use system. The most obvious one is ultra wideband types of uses, where ultra wideband needs a wide band, and therefore there is the anticommons problem. There are a bunch of situations under a property regime where that kind of technology may not exist. So we get, for example, the easements alternative that's put forth in the Farber/Faulhaber paper. And again, this is a good idea: Something that would be worth trying and I hope will work. But how will we know that easements are going to work?
First of all, again, property is the overarching regime. So the burden of proof is all on the easement user to show that it's not interfering. And how are we going to know that the easements are wide enough, or robust enough, that someone can actually build something realistic in the easement? It's sort of the excess capacity that we throw out there and say "okay commons, go at it, you have it." That may actually foreclose any practical use of the commons.
Similarly, in terms of transaction costs. The transaction costs of a property regime increase as the system gets more dynamic. As you have more degrees of flexibility, [there are] different ways to use the system, different mechanisms where you have people relaying each other's traffic, or cooperating and sharing — some of the kinds of technologies that David Reed talked about. And in a property system, every time you do more flexibility, you need more transactions. You need to price, you need to decide, you need to have someone buying and selling. So that transaction cost grows.
It doesn't scale. And actually the Kwerel and Williams paper inadvertently shows some of these transaction cost difficulties when they go through this laundry list of aspects that they would want to put into an auction. Things like combinatorial bidding. But they probably can't be done, just because the auction software isn't complex and sophisticated enough to support it today. And that's just in the initial allocations.
On the other hand, commons creates this incentive for real-time decision-making, which was brought up a little bit before. Think about what happens in the Internet. Many experts like, for example, Bob Metcalfe, the inventor of Ethernet, thought the Internet would collapse. Why does the Internet not collapse? Because everyone has an incentive not to go to a court, and not to go to a government when someone is preventing their packet from getting from A to B. If their packet doesn't get from A to B, they have an incentive to route around it. The Internet treats these things as damage and routes around them. If we decentralize the decision-making process to the participants in the market in real time, as opposed to assuming that ahead of time they can decide what the value is and transact.
So a commons that is failing one day may be successful tomorrow, as people route around problems. A commons doesn't just address the issues that we have today. It puts in place the facility to address the issues of tomorrow. And that's the kind of regime we should be looking for.
[Applause.]