August 29, 2003 DRAFT
(To be presented at
the 31st TPRC Research Conference on
Since 1927, the federal government has controlled allocation and assignment of the electromagnetic spectrum, considered the lifeblood of wireless communications. A longstanding critique advocating exclusive property rights in spectrum has recently been joined by an argument for "spectrum commons." Commons proponents claim that emerging technologies call into question the basic rationale for exclusive control of wireless frequencies.
Yet both sides fail to come to grips with an essential point: there is no such thing as spectrum. It is an intellectual construct whose utility is rapidly decreasing as new avenues for communication develop. Because spectrum is not a concrete thing, oft-used analogies to land or to natural resources break down. Proposals based on spectrum as a physical asset denominated by frequencies artificially constrain transmission mechanisms, producing inefficient outcomes. A better approach is to draw analogies to legal domains that do not presuppose ownership.
The proper focus of wireless regulation is not the spectrum, but the devices that use it for communication. The property and commons regimes are just different configurations of usage rights associated with wireless equipment. Rethinking the spectrum debate in these terms allows a better assessment of the two proposals. It also shows that both are incomplete. There is a vast new communications space emerging, whose full extent is unknown. The way to exploit this “supercommons” is to begin with a fundamental privilege for anyone to transmit anywhere, any time, for any purpose, in any way. Tort and intellectual property principles can effectively manage interference, while efficiently resolving disputes.
A specter is haunting spectrum policy – the specter of commons.
Spectrum policy is fundamental to traditional mass communications and to the emerging digital information infrastructure. All wireless communications devices, from analog television transmitters to Internet-enabled smart mobile handsets, transmit radio waves through the air. The federal government tightly limits those devices based on its control of the electromagnetic spectrum. Yet the assumptions underlying that control are under siege.
Seventy years after the birth of governmental spectrum management, and forty years after Ronald Coase and his colleagues began a campaign to kill it, the end of history for spectrum regulation seemed close at hand. By the mid-1990s, advocating extensive propertization of the electromagnetic spectrum had become, in Eli Noam’s words, the “new orthodoxy.” Even the Federal Communications Commission (FCC), which would lose much of its power if spectrum were privately owned, seemed to agree. The FCC enthusiastically adopted auctions as its preferred method for assigning spectrum licenses, proposed secondary markets for licensees to lease the spectrum they controlled, and issued statements endorsing further expansion of “market-based” spectrum reform.
In recent years, however, a new perspective on spectrum policy has emerged. The “commons” position holds that private property rights in spectrum are as unnecessary as government-issued licenses. Commons advocates claim that, thanks to advances in technology, collections of wireless devices can share spectrum effectively without exclusive rights. They therefore support expansion of “unlicensed” frequency bands, and oppose calls to turn spectrum rapidly and exhaustively into private property. Commons advocates offer two lines of support for their claims: the theoretical benefits of unlicensed operation, and the empirical success of unlicensed spread-spectrum devices.
Despite its novelty, the commons position has quickly become a significant force. The FCC’s latest comprehensive spectrum reform report endorsed greater use of commons mechanisms, along with expansion of property rights. Commons and property advocates debate each other energetically in both academic and policy circles. Recently, some scholars have claimed that the two camps are not so far apart, and have proposed approaches that encompass both mechanisms. Commons advocates have so far rejected these as essentially the property regime in disguise. However, they have not yet mapped out proposals with the specificity of the more extensive property literature. The debate, while fertile, is at something of an impasse.
Fortunately, there is a way out. The property and commons positions do come together, though not in the ways previously articulated. Both sides use analogies to fixed physical resources such as land that obscure more than they clarify. Consequently, both have wrongly assumed frequency exclusivity as a mandatory element of wireless communication rights. In other words, rights are tied to a band of wireless frequencies, whether those frequencies are subject to ownership or shared use. Frequencies are scarce, it is said, so they must be allocated.
Yet as legendary physicist Richard Feynman once said in a different context, “there’s plenty of room at the bottom.” There are many ways to communicate without disturbing other users of the same frequency band, in what I call the supercommons. The supercommons is hardly exploited today. Neither property nor commons advocates devote much attention to it. Yet it may represent the majority of potential wireless communications capacity. And any spectrum policy framework that does not expressly permit supercommons transmissions will unreasonably preclude them.
The supercommons illuminates the flaws in prior spectrum reform proposals, especially those built on exhaustive property ownership. They make assumptions about interference that may once have been justified but are irrational today. In mistakenly associating property rights with wireless frequencies, they make novel forms of communication impractical. Wireless regulation should focus not on ownership of spectrum, which is a construct, but on rights to use wireless equipment in certain ways.
The basic legal framework for wireless communication should build on bodies of law that resolve usage disputes where ownership is not a salient issue, such as tort. As an initial matter, users of wireless equipment should be permitted to transmit anywhere, any time, in any manner. This universal entry privilege should carry a duty of care backstop and a set of implied legal safe harbors to balance the interests of the transmitter and those affected by its actions. Such a tort-like regime provides a dynamic, distributed mechanism for avoiding and resolving conflicts among wireless users. It combines the deregulatory attributes of the property proposal with the openness of the commons, allowing the full range of communications possibilities to be exploited.
This article seeks to reconceptualize spectrum policy around wireless equipment rights and the supercommons model. Part I outlines the historical stages of the spectrum debate, the current situation, and where we could go from here. Part II attacks the two fallacies, reification of spectrum and assumptions about usage, that prevent a clear understanding of the problem and its solutions. Part III rebuilds wireless regulation on the new foundation of equipment usage rights. It outlines how a universal transmission privilege, limited in practice through tort and other means, provides the best and most flexible framework. Part IV returns to the property vs. commons debate, concluding that, in the near term, the commons position remains potent despite responses from property advocates. Part V offers specific recommendations.
The now-dominant government licensing approach may have been defensible in 1920, but its failings were evident by 1960. The property approach made sense in 1960 but is now questionable. The commons approach is viable today. The supercommons may become real sooner than we think.
The proper legal regime for radio frequency spectrum has been the subject of controversy since the early days of the last century. It is remarkable the debate remains recognizable. The usable spectrum today is five thousand times larger in terms of bandwidth than in 1927, when the federal Radio Act was adopted. Where there were once a handful of commercial services, including broadcast radio and maritime communication, now there are a plethora of industries including television, mobile telephony, satellite communications, radio dispatch services, and wireless local area networks. Few aspects of 21st century communications would be comprehensible to a visitor from the 1920s. Yet when it comes to spectrum, we are still arguing over the same questions: does government need to manage centrally how spectrum is allocated and assigned, and can users of wireless communications devices effectively coordinate their actions to avoid ruinous interference?
Perhaps the debate has endured because spectrum is so very important. Hardly any American is untouched by radio frequency communication. The relevant industries generate billions of dollars in annual revenue. And wireless communication is dominant form of speech in our electronic age. The radio spectrum is the town square of our digital polity. It is a major, if not the major, channel through which we obtain our news, entertainment, social interactions, and business communications. Most participants in the spectrum debate claim the spectrum is woefully under-utilized. If this is true, reforms that foster more efficient use of spectrum would have dramatically beneficial effects on daily life.
There are three major approaches to managing the spectrum. I will refer to them as “government licensing,” “property,” and “commons.”
Guglielmo Marconi first patented the mechanism for radio communications in 1897. Radio waves are manifestations of electromagnetic radiation that oscillate at characteristic rates, called frequencies. The radio frequency (RF) spectrum is nothing more than the series of frequencies usable for communications below the range of visible light, approximately 10 kilohertz (khz) to 100 Gigahertz (GHz).
Marconi’s original “spark gap” transmitters sent signals across a wide range of frequencies simultaneously. Only a single radio could operate in a particular area at a particular time for its signal to be intelligible. Perhaps the single greatest enhancement to Marconi’s original invention was frequency division. A tuning fork vibrating at a characteristic frequency will cause another tuning fork at a distance to vibrate at that same frequency. By impressing a radio signal on a carrier wave of a specific frequency, Marconi was able to transmit that signal to a receiver tuned to the same frequency. Subsequent inventors refined the technique.
Attaching a signal to a frequency allowed other signals associated with different frequencies to be sent at the same time, without preventing mutual reception. In other words, frequency division is a mechanism for subdividing the spectrum to enhance communication. It was a design choice, like the packet-switched architecture of the Internet, rather than something present in nature. This seemingly obscure technical fact will become important in the discussion below. The point is that dividing the radio spectrum into frequencies is just a consequence of a technical approach to interference management adopted in the late 19th century.
At first, anyone could operate a radio transmitter. When the first federal radio legislation passed in 1912, radio was primarily used to communicate with ships, and thus of particular interest to the Navy. Under the 1912 Act, radio stations were required to obtain licenses. By the 1920s, commercial broadcast stations had developed, and disputes about interference began to arise. Secretary of Commerce Herbert Hoover sought to use the government’s licensing authority to regulate the nascent broadcast industry. He was rebuffed by the courts, which held in 1923 and 1926 that the Department of Commerce had authority only to issue licenses, not to deny or restrict them. The result was several months in which radio stations jostled with each other to control the airwaves. This period of “chaos” came to a close with the passage of the Radio Act of 1927. The 1927 Act established federal control over the radio spectrum, and put in place the licensing regime that persists today.
The primary rationale for government control of spectrum is that spectrum is inherently scarce. The Supreme Court has upheld the FCC’s right to determine who can use the spectrum on the grounds that, thanks to scarcity, open entry would prevent anyone from enjoying the benefits of radio communication. Because of scarcity and spectrum’s fundamental importance to the public interest, decisions about who is able to use spectrum are not left to the vicissitudes of the market. At least, this is the argument.
The current dominant licensing regime involves a detailed series of top-down government decisions that determine who can build what kinds of systems, in what portion of the spectrum, for what purpose. The FCC first “allocates” a band of frequencies to put into the marketplace. It designs a set of technical requirements, including subdividing the band into blocks, mandating power limits for systems, and in some cases, determining the specific service to be delivered, such as mobile telephony. The FCC then “assigns” those frequencies to licensees, such as Verizon Wireless or ABC. A licensee is entitled to operate devices that transmit in the specified frequency, usually in a specific geographic area and occasionally during specified times. It is also entitled to protection against other licensees, or against non-licensed transmitters, which cause it “harmful interference.” It is not entitled to sell or subdivide its license without FCC approval, and the license is officially temporary.
The government licensing model for spectrum policy fit the zeitgeist of the first half of the 20th century. This was the high-water point for “scientific management” of economic activity. While the Soviet Union extolled the virtues of central planning, the bureaucrats of Franklin Roosevelt’s New Deal preached that expert managers could efficiently steer economic activity. And indeed, radio, television, and other forms of wireless communication became huge and hugely influential industries under the FCC’s stewardship. The FCC’s status as the benevolent ruler of the airwaves persisted unchallenged for a quarter century. In the 1950s, however, economists began to critique the rationale for government-issued spectrum licenses.
The economists argued that, instead of being managed by government, spectrum rights should be bought and sold like any other commodity. The first to articulate this view was a law student, Leo Herzel, in 1951. The argument was taken up brilliantly by Ronald Coase in 1959, in an article that eventually contributed to his 1991 Nobel Prize in Economics. Coase’s basic point was that markets are the most efficient mechanisms for allocating scarce resources. Spectrum is no different from any other scarce resource, so markets should be used to allocate and assign spectrum. Instead of granting licenses, he asserted, government should issue property rights that companies could then trade, subdivide, combine, or modify through mutual negotiation. Later authors, notably Arthur De Vany et al, Harvey Levin, Jora Minasian, and Milton Mueller took up the challenge of defining just what those initial property rights should look like.
Along with their proposals for what form spectrum rights should take, economists following Coase suggested a mechanism to use in assigning those rights: auctions. The spectrum assignment process has used a variety of mechanisms such as first-come, first-served; comparative hearings; and lotteries. All of these could and did fall victim to inefficiencies, capture by interest groups, or out-and-out corruption. Spectrum auctions are designed to put licenses in the hands of those who value them most highly, and who will therefore make the highest bid. Auctions have become the FCC’s preferred assignment vehicle, because of their perceived efficiency and revenue-generation benefits for the US government.
The economists’ critique of spectrum policy was part of a larger project to demolish the foundations of scientific management. The Austrian School led by Friedrich Hayek, and its American adherents in the Chicago School, have largely succeeded in promoting laissez-faire principles and price-based mechanisms in almost all areas of economic activity. In fact, wireless communication may be the major sector of economic activity where they have been least successful. What the FCC auctions today is still a license, not an alienable property right. In recent years, economists such as Thomas Hazlett and Lawrence White have vigorously pushed the FCC to take the final step and turn spectrum into private property. The FCC, which at first dismissed Coase’s proposal, has moved closer and closer to the economists’ position. Its 2000 Spectrum Policy Statement extolled the virtues of market forces in spectrum policy, a code word for property rights.
Just as advocates of property rights in spectrum seemed headed for their final victory, they faced a new challenge. A novel critique emerged that doesn’t defend the government licensing regime. In fact, it largely grants that property rights were superior techniques for regulating use of the spectrum when Coase proposed them. Its claim is that developments in technology make possible a still-better approach: treatment of spectrum as a commons.
The commons argument recognizes that spectrum can now be shared effectively, without requiring exclusive frequency licensing. Recall that Marconi’s use of frequency division to allow signals to coexist was a particular technical choice; it wasn’t a basic property of radio communication. A variety of techniques, some dating back to the 1940s, allow two or more transmitters to coexist on the same frequency. The best-known of these is spread-spectrum. As demonstrated by Bell Labs researcher Claude Shannon in his seminal 1948 papers on information theory, a signal can either be sent across a narrow channel at high power, or spread across a wide channel at lower power. When the signal is spread, the lower power reduces the degree of interference another signal.
The practical consequence is that no government regulator or property owner need decide which signal is entitled to use the frequency; both of them can use it simultaneously. More generally, the spectrum, or portions of it, can be treated as a commons, in which anyone is free to enter. In such an environment, property rights are at best unnecessary and at worst deleterious. The main real-world manifestations of spectrum commons are the unlicensed bands, where any device certified to meet specified technical criteria may operate. Unlicensed bands are products of the same FCC allocation process as other frequency bands, but instead of being assigned to an exclusive user or users they are left open to any devices certified to meet specified technical criteria.
The commons critique was first voiced in the early 1990s by technology pundit George Gilder and renowned network engineer Paul Baran. It was expanded and formalized by two academics, Eli Noam and Yochai Benkler. Noam used the possibility of spectrum sharing to demonstrate the failings of auctions, and to show that the economists’ critique did not necessarily lead to exclusive property rights to transmit on specified frequencies. Benkler argued that spread-spectrum techniques allowed for institutional arrangements that did away with the need for price signaling in transmission rights entirely. He further claimed that such commons regimes were normatively superior to property regimes, because they allowed more speech and served the preference functions of a wider range of users. Others who have built on the commons critique include myself, cyberlaw scholar Lawrence Lessig, technologist David Reed, and attorney Stuart Buck.
Two elements of the commons critique bear noting. First, it rests on two independent rationales: greater efficiency in optimizing the social welfare gains from wireless communication, and better fidelity to social values such as autonomy, diversity, and innovation. Second, commons advocates accept the economists’ diagnosis of the problem, just not their solution. The commons critique acknowledges that scarcity doe not justify government control of spectrum, and is in fact exacerbated by it. It concurs that spectrum should be managed through market forces rather than government dictates. Instead, it shifts the debate. It highlights the common assumption of exclusivity between government licensing and property rights, and opposes it with lightly controlled forms of shared access.
Despite its relative novelty and the widespread acceptance of the spectrum-as-property position, the commons critique has rapidly gained traction. Advocates of expanded property rights in spectrum have felt the need to critique it, though initially these attacks were dismissive.
The FCC, the object of all this intellectual give-and-take, hasn’t been a passive bystander. Though the Commission initially dismissed the economists’ critique, it gradually came around to the view that a market-based spectrum policy, and particularly spectrum auctions, were preferable to the tools it had previously used. The FCC won authority from Congress in 1993 to issue licenses through auctions. It held its first major auctions, for Personal Communications Service, in 1995. By 1997, auctions had become the FCC’s preferred mechanism for spectrum assignment, and the Commission was well on its way toward adopting the rest of the economists’ proposals: flexibility, secondary markets, and ultimately full property rights. The word “unlicensed” does not appear in the November 1999 FCC press release announcing its comprehensive Spectrum Policy Statement. As recently as November 2000, the FCC’s major spectrum reform initiative was a proceeding to authorize secondary markets.
Given this history, the FCC’s November 2002 Spectrum Policy Task Force Report is surprising. The Task Force worked for several months to develop a detailed comprehensive blueprint for future FCC spectrum decisions. The report endorsed expansion of property rights in spectrum, or as it preferred, “exclusive use.” It also, however, devoted a significant portion of its analysis to the commons model, treating it as a promising approach on par with exclusive use. The report suggested that exclusive use should generally be the primary mechanism for desirable lower-frequency spectrum, while commons should be the primary mechanism above 50 GHz. Following the Spectrum Task Force Report, the Commission launched several proceedings to make available more unlicensed spectrum, including the allocation of an additional 255 MHz in the 5 GHz range, and a proposal to allow unlicensed “underlays” in the broadcast television bands.
There are several reasons for the rapid legitimation of the commons argument, beyond the rhetorical persuasiveness of its proponents: lingering fears about the consequences and irreversibility of spectrum propertization; excitement about unlicensed wireless data networks due to the business success of WiFi; and desire for fresh approaches given the collapse of the telecom sector and the problems with some spectrum auctions in the US and Europe. Regardless, the commons position is now entrenched as a factor in spectrum policy. The debate is now among two rival proposals instead of over whether to change from the status quo.
Unfortunately, the argument is being framed in the wrong way. The common picture of the spectrum debate as a winner-take-all battle over whether to treat frequency bands as private property or unlicensed commons is problematic. Property and commons are not polar opposites. They are different, and the differences matter, but both will almost certainly be part of spectrum policy for the foreseeable future. More important is what the simplistic property vs. commons description leaves out. It ignores an array of new techniques that could transform use of the radio spectrum. Both proposals structure rights too coarsely, creating insurmountable transaction costs for novel communications mechanisms. An expanded formulation of the commons critique reveals not just an alternate way to manage frequency bands, but an entirely different way to look at wireless communication. Understood property, spectrum is more than frequencies, and less than a scarce physical resource.
By challenging the assumption that interference risk necessitates legally-enforced exclusivity, the commons argument opens the door to a fundamental reframing of wireless regulation. We’ve been engaged in the wrong debate about the wrong things. The wrong debate, because both property and commons are configurations of the same matrix: a web of rights, privileges, and duties assigned to certain types of equipment. The wrong things, because as Part II will demonstrate, concepts such as “spectrum,” “interference,” and “frequency bands” are deeply misleading. Removing those veils makes possible a new theory of wireless regulation that best promotes efficiency, equity, and freedom.
Both the property and commons approaches propose that users of wireless transmitters and receivers be subject to special legal conditions not applicable to other forms of private property. For, example, transmitters may only operate on certain frequencies. The government licensing model has the same effect. It differs in the restrictiveness of the conditions (for example, specifying services and protocols) and, most importantly, in forbidding any changes to the property rights without government authorization. The major innovation of the commons mechanism is in what the property rights do not grant. They do not impose duties upon other equipment as a corollary to the transmission rights.
A broadcast license allows the licensee to build a transmission tower, and to summon federal marshals to tear down pirate antennas in the same region. This power extends even to pirate broadcasters operating in adjacent locales or bands, if those cause harmful interference to the licensee. A fee simple ownership right granted to that broadcaster would have the same benefits. The fact that the right is now “private,” and can be traded or altered through the market, does not alter its basic structure. Government isn’t just giving something to the broadcaster; it’s taking something away from all potential pirate radio operators, even though they aren’t party to the agreement. Using a commons approach, however, the property right would still include an entitlement to transmit, but not the corollary ability to exclude other transmitters in the same band. Every WiFi user is both an authorized transmitter and a “pirate” to other authorized transmitters.
Why should rights granted to one user imply obligations on other users not subject to that grant? The reason is that rights are social. They have no value if others take actions that render those rights worthless. Whether and how that should be factored into the legal allocation, however, is a contingent decision. Law offers many configurations for different situations. Ownership of land conveys a right to exclude others from that land (trespass), and rights to regulate actions taken elsewhere (nuisance). Trademarks convey rights to prevent others from engaging in similar uses, but not to prevent different uses or descriptive utterances. The Fifth Amendment protection against self-incrimination conveys a privilege to remain silent, a duty on the government not to interrogate you, but no right to prevent others from incriminating you.
So, where on the spectrum (pardon the pun) does spectrum fit? My claim is that spectrum is at worst like trademark and at best like self-incrimination, yet it is being treated like land. The common metaphor of trespass to spectrum oversimplifies the diverse mechanisms for structuring legal obligations around wireless devices.
Computational technology has enjoyed such huge improvements that today’s wireless devices are qualitatively different from those of Marconi’s day. Even the technology and usage patterns of the 1950s and 1960s, when Coase issued his critique and others elaborated upon how it could be implemented, are barely relevant today. Wireless rights look the way they do because of assumptions about interference. Modern wireless systems, and those just over the horizon, are not just orders of magnitude more efficient at minimizing interference. They turn interference into a different kind of problem. In so doing, they turn the spectrum debate upside down. Instead of strengthening exclusive control of frequencies through perpetual property rights, we should be making it broadly possible to share spectrum in ways we cannot even imagine today.
To rebuild the legal framework for wireless communication, we must first remove the façades that obscure clear thinking. Spectrum policy falls victim to several fallacies. Each is demonstrably false, yet remarkably durable. The most damaging is the notion that there is such a thing as spectrum, and that it behaves as a fixed physical resource like land. Establishing a legal regime under such a misconception is like sailing West from Europe to find a shorter trade route to India. You might find something interesting along the way, but you’ll never achieve your objective.
The fallacy is not confined to any side in the spectrum debate. However, overcoming the confusion provides ammunition for the commons position.
Albert Einstein, asked to explain radio, is reported to have replied:
You see, wire telegraph is a kind of a very, very long cat. You pull his tail in New York and his head is meowing in Los Angeles. Do you understand this? And radio operates exactly the same way: you send signals here, they receive them there. The only difference is that there is no cat.
Einstein’s analogy is accurate, because it says only what spectrum is not. There is no proper way to explain what spectrum is, because there is no such thing as spectrum. It is an illusion we grasp hold of to avoid concepts that trouble our intuitions about how the world works. Radio transmissions are tied to frequencies only because that is the mechanism Marconi developed for multiplexing simultaneous signals in the same physical space. The spectrum as a progression of frequencies tied to services exists nowhere in nature. It is analogous to the periodic table of elements, helpful for understanding but purely an intellectual construct. The reification of that construct into a concrete physical manifestation causes nothing but confusion.
In any wireless communications system, there are only three elements: transmitters, receivers, and electromagnetic radiation passing between them. The waves do not ride on any medium; they are the medium. In information theory and engineering practice what lies between transmitter and receiver is called a channel. A channel is just another convenient way to describe the interaction of transmitters, receivers, and electromagnetic waves. It does not exist outside those interactions.
The popular notion that radio waves travel through the spectrum does not reflect the deep physical structure of reality. It recalls the luminiferous aether, the universal fluid that Isaac Newton postulated to explain how bodies moved through space. The world’s leading scientists accepted Newton’s construct for centuries, until it became clear that it did not accord with experimental results. It took Einstein’s theory of relativity to demonstrate that the aether was a fiction, and to offer a new mechanism to do with that fiction had done.
It is no more rational to talk about rights in the spectrum than rights in the musical scale. What government is assigning are rights to use certain types of equipment. That is true whether the legal regime is licensing, property, commons, or anything else that can be imagined. Government cannot issue rights in radio frequencies themselves, because those frequencies are just properties of electromagnetic waves emitted and received by particular devices. Yet the literature is replete with articles that declare “the spectrum” imaginary and proceed to treat it as a concrete physical asset.
The problem is not that spectrum rights are an administrative creation associated with an intangible asset. So are pollution emission credits. The trouble with assigning rights to the administratively created spectrum resource is that it serves no useful purpose. The equivalent would be to assign rights in masses rather than in physical objects such as cars and books that possess those masses. Standing behind the spectrum construct is frequency, which is just a property of electromagnetic waves, which are just energy radiated by equipment with particular properties. Nothing is gained through this indirection. We can consider the equipment properties directly, and in an age of cheap computation and flexible devices, equipment is the better locus for regulation.
Even worse is the pervasive analogy to real estate. Courts considering the exotic realm of cyberspace frequently grasp at familiar common law doctrines designed for land. Spectrum policy experts make the same connection. Yet land is not only a thing, but a thing with very particular qualities. Comparing wireless communication to grazing sheep in a meadow suggests that a whole series of legal and economic constructs applied to meadows can usefully be applied to spectrum. They cannot. A meadow has a specific amount of grass, and one sheep eats so much of that grass each day. Wireless communication works differently.
Even if one were to grant that interference among wireless communications devices is similar to nuisances that adjacent land owners impose on one another, that would not make spectrum analogous to land. The proper analogy would be between wireless communications rights and certain uses of land. Ownership of private property always includes limitations on how that property can be used. A murderer, for example, cannot claim he was merely exercising his right to use his legitimately owned gun. A hog farm or tannery may be subject to restrictions for the benefit of adjacent homeowners, not because it somehow invades their land, but because its use of its own land is inconsistent with their enjoyment of theirs. The land is still property, with the same physical boundaries, but the bundle of right associated with that land has changed.
Crucially, the contours of the land-owner’s usage rights are defined in a social context, with reference to other owners who may be affected. So too with spectrum. Interference is a function of collective uses and equipment choices, not of the medium involved.
A better, but still misleading, analogy is between spectrum and natural resources. This view appears most prominently in the work of Harvey Levin. Yet even Levin acknowledges that, in precise terms, the spectrum is “a three-dimensional capability for transmitting information with electromagnetic energy.” A capability is not the same as a resource. The spectrum resource Levin imagines still has an independent existence from the devices that engage in transmission. Levin admits that, unlike other resources, spectrum is perfectly and costlessly renewable, but suggests this is only a difference of degree. He argues that spectrum is a common property resource that, like oil or fisheries, must be subject to administrative regulation or exclusive property rights to avoid over-use and depletion. In other words, spectrum is a resource because it is subject to interference. Lawrence White makes the same linkage. This assumption is false, for reasons I explain in the next section.
Not all scholars of spectrum policy treat spectrum as corporeal. Benkler repeatedly emphasizes that spectrum is not a thing. He goes so far as to label his preferred solution as “open wireless networks” to avoid references to a “spectrum commons.” Benkler wants to avoid the spectrum fallacy because he argues for an industry model based around end-user purchases of equipment that operates without licensing. If spectrum is a thing, granting property rights in that thing seems only natural. Better to compare exclusive transmission rights and opportunities for manufacturers to build and sell frequency-sharing equipment.
Coase also clearly understood that spectrum wasn’t a thing. As he explained in his seminal article on the FCC:
Every regular wave motion may be described as a frequency. The various musical notes correspond to frequencies in sound waves. The various colors correspond to frequencies in light waves. But it has not been thought necessary to allocate to different persons or to create property rights in the notes of the musical scale or the colors of the rainbow. ... What does not seem to have been understood is that what is being allocated by the Federal Communications Commission, or, if there was a market, what would be sold, is the right to use a piece of equipment to transmit signals in particular way.
Coase needed this point to counter a different argument than Benkler. The FCC licensing system bases government control not only on interference avoidance, but on the idea that the spectrum is a “public trust.” If there is a thing called spectrum that belongs to the American people, government should regulate access to it in the same way it regulates access to the Grand Canyon. Privatizing the Grand Canyon is abhorrent to most Americans. How could government turn over a public treasure in perpetuity to the rapacious interests of private companies? Hence, Coase first had to replace the public airwaves with a set of private transmission rights. He could then convincingly argue that markets should allocate those transmission rights.
Coase in the 1950s understood that spectrum was an incoherent concept. A related point, the incoherence of interference, would have to wait until technology evolved beyond the analog broadcast systems prevalent at that time. Coase’s property rights solution made sense when he developed it, though his work on transaction cost economics revealed just how contingent the determination was. With what we know today, the same analysis leads to a very different conclusion.
It turns out that interference, like spectrum, is a convenient fiction. As a physical matter, radio waves do not bounce off one another. They continue merrily on their way, propagating through free space forever, though attenuating in strength until they become undetectable. In a sense, therefore, interference is always present. No transmitter on Earth is perfectly immune from other signals. What matters in communications systems, however, is not the waves themselves, but the ability to extract information from them. If two waves are nearby in frequency and location, it may be difficult for a receiver to determine which is which. This is no different than the difficulty the receiver has in distinguishing a single wave from the ever-present background noise produced by everything from electric motors to cosmic radiation.
Interference manifests itself in the receiver, not in the radio transmissions themselves. Moreover, it is a function of the receiver’s computational intelligence. A digital mobile phone handset sold today would pick up crystal-clear conversations where devices built in 1960 hear only static. The issue is not merely sensitivity. Claude Shannon’s capacity theorem, developed in his classic papers that established the foundations of information theory, holds that the capacity of a communications channel is proportional to the width of the channel and the transmission power used. In other words, more bandwidth, all things being equal, means more capacity. Bandwidth is therefore commonly used as a synonym for capacity.
However, the concepts are not equivalent, because the other variables can change. A system may offer more capacity with less bandwidth, so long as it increases power. Or it can keep capacity constant at lower power by increasing bandwidth. This last scenario is important because high transmission power overwhelms receivers and causes what we call interference. Faced with a high-power and a low-power signal, the receiver will detect the high-power signal or some combination of the two. If the first signal is spread across a wider bandwidth and sent with very low power, however, the receiver may be able to pick up the second signal cleanly. The most common technique for trading off bandwidth and power in this way is known as spread-spectrum.
Spread spectrum is not the only method for mitigating interference. The simplest version of Shannon’s theorem provides the capacity of a communications channel between a single transmitter and a single receiver. In the real world, though, we are concerned with radio communications systems, which can involve many transmitters, many receivers, and many intervening factors such as walls that reflect or distort the signals. This might seem to make the interference problem worse. And indeed it does, if we limit ourselves to the primitive radio technology of the 1920s. Fortunately, technology has come a long way. We are nearly three times as far in time from the 1927 Radio Act and the birth of regulated broadcasting as it was from Marconi’s experiments.
Research in multi-user information theory has identified numerous mechanisms to enhance capacity and avoid interference. For example, receivers can be designed to function simultaneously as relay transmitters, allowing messages to hop from node to node like packets across the Internet, an architecture known as meshed networking. This and other techniques exploit what David Reed calls cooperation gain and what multi-user information theory labels diversity gain. They share the property that additional nodes in the network add capacity as well as consuming it.
How far back can the interference frontier be pushed? We don’t know. It is an open research question whether the capacity of a physically-bounded network with an arbitrary number of transmitters and receivers can scale linearly with the number of nodes. If it could, each new user would add as much to the network as it took away. Even if it can’t, interference might become such a minor problem that legal regimes to cope with it are overkill. How close usage comes to some theoretical optimum matters less than whether, in practice, the benefits from more users exceed the costs. The more likely it is that interference will be a practical problem, the more transaction costs we should tolerate to avoid it.
Even the baseline for interference is not where it seems. Virtually every frequency through the 5 GHz range has been assigned either to a licensee, unlicensed operation, scientific activity such as radio astronomy, or government. The fact that there are few if any unassigned spaces on the frequency dial, even as wireless services become more popular and varied, reinforces the popular notion of a spectrum drought.
Examining actual usage reveals a very different picture. Most frequencies are idle in most places most of the time. They may be off-limits to protect against interference with adjacent channels, the licensee may not actually be transmitting (as with many UHF television licensees), or the authorized service may not saturate the channel. A cellular phone tower, for example, is only active when communicating with a handset in its range. A recent survey by Shared Spectrum Inc., sponsored by the New America Foundation, found that two-thirds of the most desirable “beachfront” spectrum was “immediately available for shared, unlicensed use.” And that was during peak hours in a dense urban area.
One final point about interference. Because it is solely a phenomenon of receivers, the receivers are legitimate subjects for allocation of legal rights. Our intuitive notion is that interference results from unauthorized transmissions that “block” other transmissions. However, the same “interfering” transmission may be totally unnoticeable to a more robust receiver.
Say A has a mobile phone license, and B establishes a wireless Internet link nearby over adjacent frequencies. It suddenly becomes difficult for A’s customers to receive calls when they are near B’s transmitter. One interpretation is that B is “causing” the interference and should be shut down. Another interpretation, however, is that A should bear the responsibility. A decided to use receivers that could not distinguish B’s signal. Society could make a choice to protect A rather than B. However, that choice would be based not on causation but on some calculation of the welfare effects of assigning the right to one side or the other.
Coase engaged in exactly this analysis in both his FCC article and his seminal paper, The Problem of Social Cost. For illustration, Coase used a 19th century case involving a confectioner and a doctor who builds an examining room at the edge of an adjacent property. The doctor finds his work impaired amid the vibrations from the confectioner’s machinery. As Coase pointed out, we could say the confectioner caused injury to the doctor, or that the doctor is excessively sensitive to vibrations. We can choose, but to do so is a value decision among two legitimate activities. Any claim about interference can be expressed either in terms of transmitter intrusiveness or receiver sensitivity. We can choose to impose a duty on the transmitter, or we can impose a duty on the receiver, but either way we make a choice.
Surprisingly, the FCC’s rules implicitly acknowledge the contingency of interference. They define interference as “[t]he effect of unwanted energy....” Interference is not an action, or even a state; it is an “effect.” Moreover, it occurs only when energy is “unwanted.” Unwanted by whom? The erstwhile receiver of some other “wanted” energy. The transmissions themselves have no idea whether they are welcome or not.
The collective nature of wireless communications rights is clearly apparent in the FCC’s definition. I can emit the same radio waves a hundred times, but if you decide the next emission is “unwanted” for your simultaneous communications, it suddenly becomes interference. Interference is a social construct arising from collective uses of wireless devices. It depends on the technical capabilities of those devices as well as the applications and services for which they are employed.
The ultimate policy goal is not to eliminate interference. That is hopeless. Some energy will always propagate where it is not desired. More fundamentally, though, interference is not an evil that must be eradicated at all costs. Interference is a by-product of the very phenomenon policy-makers hope to achieve: more value from wireless communication. If there were only two radio stations using the entire spectrum there would be little opportunity for interference. The widespread possibility for interference is a sign of success, not failure. Focusing too hard on eliminating it would be like killing off an annoying animal species, only to have a worse pest which the first species had kept in check multiply.
The proper goal is to optimize interference. A transmission should take place if the marginal value it adds exceeds its marginal cost, with interference counting as a cost. This analysis becomes complicated because interference is neither a localized nor an all-or-nothing phenomenon. The interfering “noise” for any transmission is a combination of intentional and unintentional emissions from many other sources, which affect reception both individually and collectively. A degradation of reception may mean a slight hiss in the background of a phone call or a lost message between an air traffic controller and a jumbo jet pilot. The proper analysis is not whether a regime prevents or tolerates interference, but how it resolves boundary cases and allows for tradeoffs along many dimensions.
3) Frequency blocks
A third aspect of the spectrum fallacy is the emphasis on frequency blocks as the unit of allocation. Frequency is indeed a physical property of radio waves. The relevant legal structures, however, are designed not for science experiments but for communications systems. Frequency, like bandwidth, is but one aspect of those systems. Though every wireless license or property right includes other constraints as well, frequency has been the central delimiter among systems since Marconi’s day.
The commons critique coalesced with the development of spread-spectrum systems in the unlicensed 900 MHz, 2.4 GHz, and 5 GHz bands. Benkler’s touchstone was the FCC’s 1997 decision to allocate spectrum at 5 GHz for unlicensed National Information Infrastructure (U-NII) devices. Subsequently, the rapid growth of the market for WiFi devices, primarily in the 2.4 GHz Industrial, Scientific, and Medical (ISM) band, has proved both the feasibility and dynamism of a commons-like arrangement. In both cases, the commons exists within a “park” designated exclusively for unlicensed operation. Emphasis on these developments has created the misconception that the commons critique relies on dedicated unlicensed frequency bands.
Hence, property advocates claim that governments will fall victim to the same failings in allocating bands for unlicensed use as they do in allocating bands for licensed systems. They assert that if unlicensed parks such as the U-NII band are valuable, they will appear in a property-rights world either through government creation of “public parks” or manufacturers buying rights to create “private parks.” I address these arguments in Part IV. Even if they are true, though, they fail to rebut the commons critique. The wireless commons involves more than unlicensed bands.
There are several technical mechanisms to communicate without assigning dedicated frequency bands to each channel. All of them provide pathways to expand wireless communications capacity other than exploiting higher frequencies or using existing frequencies more intensively, which are the primary techniques property advocates have considered.
o Wideband underlay
Take spread-spectrum to its logical conclusion, and the result is ultra-wideband (UWB). UWB transmissions use such large bandwidth that they can transmit at power levels below the “noise floor” for other devices. In other words, a licensed system operating in a band covered by the UWB system won’t even know it was there. The FCC authorized such “underlay” techniques in its February 2002 UWB order. Most UWB systems are carrierless: in contrast to virtually all other radios since Marconi’s day, they do not operate by impressing messages upon carrier waves of a specified frequency. Instead, they use extremely short electrical pulses.
The FCC’s UWB order effectively created a commons without setting aside a dedicated unlicensed band. It had to. As a matter of physics, the shorter the duration of a wireless signal, the wider it spreads. To achieve its full potential, UWB cannot be confined to traditional frequency blocks of a few Megahertz. The FCC’s order, for example, authorizes UWB across a range of seven Gigahertz, though not every system will use the entire range. There is no way a prospective UWB system manufacturer could possibly negotiate with all the constituent frequency bands for authorization. The problem is not that a particular allocation mechanism involves too-narrow blocks; it is that any limitation on frequency range will constrain some UWB systems which may be optimal to permit.
o Opportunistic sharing or interweaving
Many frequencies, even those ostensibly licensed for established services, are actually empty some of all of the time. For example, television channels 3 and 6 are occupied in Philadelphia, while 2 and 4 are vacant. The reverse is true in New York City. At the time broadcast television was introduced, receivers in Philadelphia couldn’t distinguish between signals on channel 2 and channel 3, or between the local channel 2 and channel 2 in New York. So some channels simply lie fallow. In other cases, such as UHF television and ITFS fixed wireless, licensees may have the right to transmit but are not doing so for economic reasons. Or a system, such as a cellular telephone network, may operate throughout the licensed band, but not transmit in all places at all times.
Current technologies can exploit some of these holes, a process known as opportunistic sharing or interweaving. In particular, software-defined radio, which uses reconfigurable software to tune radios to different frequencies and encoding schemes, holds great promise for facilitating more powerful opportunistic sharing strategies, should the law change to permit them. The Defense Advanced Research Projects Agency (DARPA), which funded much of the basic networking research that led to the Internet, is actively exploring one opportunistic sharing mechanism through its XG research program. In the future, “cognitive radios” may be able to scan the local spectral environment, find an open frequency, transmit there using an efficient encoding mechanism, and move to another frequency so quickly that a coexisting system won’t even know it’s there. Allocation according to frequency blocks would hamstring such devices.
Though the “holes” opportunistically exploited are usually frequency-based, this is not necessarily the case. For example, some meteorological radar systems are in operation only a few minutes per hour. Another system could split use of the frequency purely on a time basis.
o Intelligent coding and smart antennas
Smart digital devices can employ many techniques other than frequency diversity to improve the performance of wireless systems. These mechanisms use factors such as the physical location of transmitters, motion, or the scattering effects of intervening obstacles that portions of the signal bounce off, to better lock onto signals and distinguish them from noise. For example, the BLAST system developed at Bell Laboratories uses multiple antennas on both the transmitter and receiver. By tracking the multiple signal paths between the antenna arrays, BLAST obtains a better understanding of the signal characteristics. This type of approach is known as space-time coding or multiple in, multiple out (MIMO).
These are not just theoretical ideas. Airgo Networks, a Silicon Valley startup, announced MIMO chipsets in Summer 2003 that extend the range and capacity of WiFi systems. Companies such as Northpoint Technology have demonstrated “angle of arrival” systems that allow for terrestrial wireless communications on the same frequencies used for satellite uplink and downlink, with neither service subject to interference. The “new” spectrum, which could in theory be a commons, perfectly overlaps the frequencies of the satellite system, and uses similar power levels.
Intelligence can be built not just into the software that processes signals at the transmitter or receiver, but into the antennas they use. The classic TV aerial on the roof of a house is exceedingly simple. It uses horizontal bars of lengths that match the periodic frequencies of a broadcast channel. Modern electronic antennas can be highly directional and adaptive. They can even be tuned dynamically to lock on and shape a narrow directional beam to a signal, preventing it from spreading widely where it might impinge on other signals.
o Physical space
Wireless systems can also be divided by physical location. A low-power wireless transmitter in a house may not create noticeable interference to any system outside that house. That is true regardless of what frequency band the transmitter uses. Under longstanding doctrines of property law and the Fourth Amendment, people are permitted to engage in many forms of conduct in their own homes that would be impermissible in public.
Michael Chartier of Intel has proposed a rule that wireless transmission rights should be “fixtures” to private property in certain bands. This principle could be adopted more broadly. If a transmission within a house does not radiate outside to the point at which it affects other signals, why should there be any constrains on that transmission? Property owners are entitled to knock down walls of their houses or decorate their bedrooms in a manner others would find garish. If I choose to operate a wireless system in my house whose only negative externality is that I knock out my own broadcast TV reception, perhaps that should be my choice.
o Endless possibilities?
There is no reason to think all possible mechanisms for sharing spectrum on a basis other than frequency division have been invented. Computers continue to become more powerful, opening up new possibilities that were not feasible before. Multi-user information theory is a particularly fertile research area in which several major questions remain unsolved. Many of the intelligent coding mechanisms have the interesting property that they take phenomena that once “caused” interference and use it to improve reception. For example, when portions of a signal bounce off walls or other obstacles, they arrive at a receiver slightly after signals that passed straight through the air. Such “multipath fading” is the bane of wireless systems, because receivers don’t realize the second signal is a copy of the first. If, however, the system is smart enough, it can correlate the two signals and combine them, improving reception. This suggests we may just be seeing the beginning of the post-frequency wireless era.
The spectrum fallacy is pernicious not only in placing too much focus on frequency, but also in directing attention away from characteristics that matter a great deal, such as architecture. Architecture is an essential element of any communications system. In this context, architecture refers to the organizing principles and structure of relationships among the network’s components. Focusing on the spectrum rather than the devices obscures the different ways those devices can be designed and connected. This in turn produces a blind spot about how architecture can enhance wireless capacity and value.
Wireless communications systems are more than isolated transmitter/receiver pairs. Two systems in the same frequency and location may deliver very different services if their architectures are different. For example, a broadcast service such as television sends the same signal from a central transmitter to many passive receivers. A cellular service such as mobile telephony uses many smaller transmitters which connect locally to two-way handsets. Each user may get less capacity in the cellular model, but the total capacity of the system is much greater because so many different transmissions can occur simultaneously. The broadcast and cellular networks may be alternative uses for the same location, time, and frequency, but they are not interchangeable. Each produces a different utilization pattern. And each produces a different boundary along which other communications systems could theoretically coexist.
The development of cellular systems was a key innovation in wireless technology, because it allowed many small networks to operate as one big network. Meshed networking takes that concept even further, turning receivers into repeaters that add capacity as they consume it. Another new architecture is ad hoc networking, in which new nodes anywhere automatically become part of the network, compared to the planned expansion of traditional systems. How spectrum is made available influences network architectures, which in turn affect how spectrum can be used. A commons, which substitutes open entry for exclusive control, tends to foster decentralized networks of many transmitters, with capital expenditures centered on user purchases of commodity equipment. Exclusive licensing or property rights favor centralized infrastructure investment by an operator.
5) Implications for the property vs. commons debate
The spectrum fallacy puts the debate on the wrong terms. Analyzing property and commons models for rights in a spectrum resource is an interesting intellectual exercise, but it’s not a discussion about the real world. The only thing that matters is the effects of the two regimes when seen for what they really are: different configurations of the rights in wireless transmitters and receivers. And in that context, many arguments for property rights are only valid under particular factual assumptions which are increasingly questionable.
The property critique falls headlong into the spectrum fallacy. Granted, property advocates understand the basic physics of wireless communication. Their arguments, however, ineluctably lead to a model of spectrum as land. It is difficult to advocate ownership without a tangible resource to be owned. The land metaphor allows property advocates to fit extensions such as easements and subdivisions into a consistent cognitive map. However, as I discuss below in Part IV, there are reasons to be skeptical the property system can accommodate the wealth of additional possibilities that are now becoming real for wireless communication. The only way to treat frequency blocks like land is to ignore the mechanisms under which spectrum can be used differently.
The debate between property and commons is not a fight over spectrum; it’s a fight about different configurations of rights. The policy question, therefore, is which constellation of rights is most efficient and socially desirable. As Coase demonstrated, there is no “correct” place to assign rights. Once the right is assigned the parties may bargain to reassign it. The assignment does, however, affect the likelihood and transaction costs in getting to that equilibrium point. The property and commons regimes for wireless communication involve different kinds of transaction costs in modifying transmission rights. By placing all the burden on the potential entrant to aggregate information and negotiate the purchase of the necessary rights, the exclusive property rights model imposes a bias toward established uses and techniques. By making boundary definition between systems necessarily a market-based transaction, it adds rigidity and cost to the evolving process of determining the most efficient configuration of devices.
Exclusive property rights are superior to unconstrained entry for most physical resources. Wireless is different. It is different because spectrum is not a physical resource, and because users can add capacity or avoid conflicts dynamically. If spectrum were a thing, the transaction cost analysis for a property regime would be relatively simple. The FCC-defined constraints on licenses could be converted into private property rights to transmit in certain frequency bands. If, as seems likely, the existing boundaries were not completely efficient, rights-holders could buy or sell them. The spectrum market would reshuffle the ownership of frequency blocks, much as the real estate market reshuffles title to land.
Eliminate the spectrum fallacy, and the picture becomes more complicated. Frequencies are not the only dimension for transactions, because they are far from the only variable that determines interference. Underlay mechanisms, to take one example, depend on non-exclusivity of frequency blocks. So does opportunistic sharing through cognitive radios, but in a completely different manner. The more different ways there are to configure wireless communications systems in order to increase capacity, the more complex the transactional regime to implement those mechanisms becomes, assuming each time there must be a financial transaction. A property regime may still be a good answer, but the choice is not so clear ex ante.
1) What we don’t know could hurt us
Closely related to the false vision of a spectrum resource is an epistemological fallacy. We think we know how wireless communications systems will be used. Though rules are described in general terms, we typically have a particular service, or group of services, in mind when we talk about them.
Historically, wireless policy was about broadcast. Broadcast radio was the primary commercial application that drove the government to assert control over the airwaves in the 1920s. Broadcast television, which supplanted radio as the most lucrative method of wireless communication, was the initial animating service for the economists’ critique. In the 1980s and 1990s, cellular telephony became the focus, and it was the testing ground for the FCC’s auction movement. The commons critique takes wireless local area networks and other data connections as its examples.
The emphasis on specific services is partly a legacy of the FCC’s practice of designating frequencies for particular applications. However, that does not completely explain the extent to which analysis has been tied to uses. Wireless communication is mysterious and ethereal. Just as it is easier to talk of frequency blocks than incorporeal phenomena, it is easier to think about concrete services than pure radiation. Easier, but misleading. In the current digital world, there is no fundamental difference among the content delivered through any communications service. Bits are bits. Therefore, many service-specific policy decisions no longer have the same force. At the same time, services differ greatly in the architecture, business dynamics, and social significance. A broadcast-oriented regime may be inappropriate for ad hoc sensor networks, both technically and in the substantive tradeoffs it makes.
Future usage patterns are heavily uncertain. Consumer demand for information technology is notoriously difficult to predict. Recent history is littered with examples of smart, successful companies wasting huge sums pursuing chimeras like video-on-demand while ignoring fax machines and the Internet. Customers in Europe or Japan have different preferences than those in the US, causing different usage patterns for technologies such as mobile messaging and broadband Internet connections. Usage is influenced by regulation, which can itself be unpredictable. Mobile phones were launched earlier in the US than Europe, but grew more quickly in Europe once the GSM standard and calling party pays roaming agreements were in place.
Usage also depends on substitutes. Eighty-five percent of American homes get their television from something other than a terrestrial broadcast signal, even though those broadcast licenses are considered the crown jewels of the spectrum. If those transmitters were turned off, only a handful of Americans would notice. Finally, spectrum usage depends on technology, which is always changing. We can make educated guesses about how much room cognitive radios linked into meshed networks will have to maneuver forty years from now, but no one can really be sure. Infinite capacity and total gridlock are both speculative outcomes.
The usage fallacy matters when we attempt to define what the ultimate objective of wireless policy should be. Different kinds of networks are measured based on different variables. For example, Benkler declares that the proper objective of wireless policy is to maximize network capacity, and describes an “ideal” wireless network as a ubiquitous mesh of interconnected nodes. He builds an economic model to show the superiority of the commons on a capacity-maximization basis. Yet Benkler never elaborates on his choice of capacity as the proper economic objective. More capacity is certainly a good thing, but it is not the only thing. Why not optimize for the network that reaches the most users, or that transmits the farthest distance, or that saves batteries by using the least power?
Furthermore, capacity has more than one meaning. Does it refer to the number of connections or the number of bits transferred? The network that transfers the most bits in a given period of time, or the network that delivers the most valuable bits? And how should value be measured – in dollar terms or based on some normative concept?
These are the issues network designers consider case-by-case. Design decisions in any network will optimize for some values and therefore make certain services more difficult or even impossible. The Internet’s “best efforts” transmission policy has a myriad of benefits, but makes it difficult to use the public Internet for latency-sensitive applications such as telephony and videoconferencing. As a general rule, meshed wireless networks do not optimize for latency, because a message may pass through a large number of intermediate nodes to reach its destination. And even if an unlicensed ad-hoc meshed network could deliver broadcast television, it would likely be more complex (and thus more costly) than the existing high-power, dumb receiver broadcast infrastructure. A property advocate would argue that markets should decide which uses should be advantaged, because any external decision to privilege one type of network will be biased.
Yet property advocates are not immune from the usage fallacy. Coase and Herzel’s original proposals focused on broadcasting, because that was the dominant form of commercial wireless communication at the time. The property regime implies certain uses in the very nature of the rights it grants, especially when those rights are based initially on existing FCC licenses. Those licenses were, in most cases, designed with specific services in mind. Finally, as Benkler explains, an exclusive rights regime for spectrum leads to a top-down market structure of few service providers controlling what their customers can do. This in turn produces services oriented toward the most widely inoffensive second-best options, rather than reflecting the broad range of individual preferences.
The best response to uncertainty is diversity. Many experiments are more likely to hit on the right approach than a few all-or-nothing guesses. Conversely, locally unpredictable risks can even out when aggregated with other risks. This is why insurance is such a large industry, and why financial advisors recommend diversification of stock portfolios.
2) Architectural choices
Uncertainty is also high regarding how wireless networks will be organized. This is one area in which frequency does matter. The frequency of radio waves affects their propagation characteristics, making frequencies more or less amenable for certain uses. Lower-frequency waves better penetrate walls, trees, water, and other obstructions, and travel farther through the air. Wireless communications systems first exploited the lowest frequencies and have gradually migrated up the spectrum as technology improved. Radio operates in the kilohertz or lower, television broadcasts in frequencies between about 300 and 800 MHz, and most cellular phone systems operate between 900 MHz and 3 GHz. Still, 90% of all use of the radio spectrum occurs in the one percent of frequencies below 3 GHz. Higher-frequency systems tend to operate over shorter distances, or on a more directional basis, to compensate for propagation loss.
Frequency is not, however, the absolute determinant of which applications can be supported. There are many ways to skin a cat (or in Einstein’s terms, to skin no cat). For example, low-power systems, such as UWB or in-home devices, can be connected with a small number of high-power “backhaul” links or chained together into meshed networks to provide wider coverage. A higher frequency that requires a more expensive network of devices to exploit may be cheaper overall if there is no fee to access that spectrum. These tradeoffs can best be made in real-time on a distributed basis. The appeal of a property rights regime for spectrum is that it seems to allow market mechanisms to execute such real-time transactions. In reality, though, as Noam argues, exclusive ownership of frequencies would be more of a barrier to real-time transactions than a facilitator. Below in Part IV, I analyze the competing spectrum proposals’ influence on real-time transactions more closely.
Wireless systems may also trade off among the three fundamental variables in any data network: transmission capacity, switching, and storage. Consider an analogy from the wired Internet. The Internet is packet-switched: all communications are broken up into small pieces called packets. Because it is a best-efforts network, delivery of an individual packet can never be guaranteed. Yet some applications, such as streaming video, cannot tolerate jitter, the degradation of picture quality due to excessive packets lost in transit. One solution is buffering. A small portion of the video file is downloaded to the user’s computer before the video stream starts running. While the user watches that section, the next chunk is being downloaded in the background, and so forth. The added buffer between transmission and viewing effectively makes the system more forgiving to packet loss in the network.
Buffering works by trading storage (the user’s hard drive) for transmission. It is a special case of a fundamental equivalence in networking among transmission, switching, and storage. Another case is packet-switching “wasting” computation by distributing switching to routers operating throughout the network. In so doing, it economizes on transmission in contrast to circuit-switched networks that hold open an end-to-end circuit for the duration of every call. The costs of computing are falling faster than those of end-to-end transmission, which is one reason the Internet is so successful. Spread-spectrum is another example of wasting computation in order to economize on transmission, in this case the wireless channel.
Traditional broadcast networks are “better” than the Internet in their ability to guarantee delivery of the same content to every user, and to do so with simple (and therefore cheap) equipment. Yet the Internet is “better” than broadcast in its ability to support two-way interactivity, a multitude of services and content offerings, as well as rapid innovation. If there must be a choice between using spectrum for broadcast or Internet applications, there is no neutral way to decide.
1) The right kind of mistakes
The fact that spectrum isn’t a thing, or that assumptions about wireless systems cannot be sustained, doesn’t necessarily mean those viewpoints aren’t useful. We merrily act as through the physical world were solid and deterministic even thought quantum mechanics tells us with great confidence that it isn’t. Neoclassical economics rests on the idea that even though people may not always behave rationally, we can act as though, in a collective sense, they do. The spectrum fallacy is different. It produces ideas that are not just wrong, they are dangerous.
The idea of spectrum as a concrete thing divided by frequencies is a paradigm, in the sense made famous in Thomas Kuhn’s The Structure of Scientific Revolutions. A paradigm provides an organizing worldview that is useful most of the time, when incremental advances take the form Kuhn called “normal science.” At some point, though, the paradigm can become a liability. It obscures truly radical ideas that may be major improvements on the current paradigm. On rare occasions the paradigm itself must shift for science to move forward. The classical spectrum paradigm has served us well, but is nearing the end of its usefulness.
Treating spectrum as a thing begs the very question spectrum policy must answer: how scarce spectrum really is. Two people cannot farm the same plot of land at the same time. Saying that spectrum is like land implies that the same exclusivity applies. But as I have explained, it does not. Such constraints on simultaneous usage as do exist are a complicated function of system design, equipment capacity, and application robustness. Treating spectrum as a resource defined by frequencies makes it difficult or even impossible to allocate it along different dimensions, even if that would increase the capacity and value of that spectrum.
A more subtle danger of the spectrum fallacy is that it implies false certainties. This in turn makes the inevitable errors of spectrum policy more difficult to correct. An analogy to the scientific method may prove helpful. A scientific theory can never be right. It can only be wrong. That is because no theory is ever a perfect description of reality, or at least, we can never prove that it is. The object of a theory is explanation, which leads to prediction and action. The beauty of the scientific method is that, purely through constant iteration of prediction and experimentation, it shows which theories are wrong and therefore produces better theories.
Newton’s mechanics told astronomers that the orbit of the planet Uranus was being deformed by some other massive body, and told them where to look for it. Lo and behold, they discovered Neptune in precisely that place. Yet as we now know, Newton’s mechanics are wrong. They suffice to describe the macro-scale world of our subjective experience, but not the cosmological world of astronomical scale nor micro-world of fundamental forces and particles. Einstein’s theories of special and general relativity, combined with quantum mechanics, form a more accurate understanding of reality.
An incorrect theory may, however, still be useful. It is Newton’s physics, not Einstein’s, taught to every high school student today. For most phenomena we are concerned with, Newton’s physics still gives the correct answer. And Newton’s physics is simpler in application and more tractable in concept than Einstein’s. By contrast, no physics or astronomy course teaches Ptolemy’s theory of planetary epicycles. Not only is it more complicated than the alternative – elliptical orbits around the Sun – its assumption that the planets revolve around the Earth offends our modern perspective.
So our goal is a theory of wireless communications that, at worst, encourages Newtonian errors. It should remain useful even if some of its assumptions later turn out to be false. Ideally, it should adapt gracefully to such shifts. The only constancy, after all, is change. The best policy approach to wireless communications is one that mimics the current regime if capacity and other limitations are in fact as Congress believed them to be in 1927. It should adapt to handle new efficient usage patterns under the technical backdrop now widely accepted, one of digital convergence and significantly lesser scarcity.
Crucially, it shouldn’t stop there. If the rules governing radio transmissions simply allow more high-power broadcasters or cellular telephony systems to exist, we will have won the last war. Yet if technology and society continue to evolve, making new systems viable that roll back scarcity beyond the perceptual horizon, rules that build in today’s assumption will produce the same welfare losses over the next seventy years as the licensing regime has over the last seventy.
For all its flaws, the government licensing model of spectrum is resilient to certain errors. Herbert Hoover, opening the first National Radio Telephony Conference in 1922, expressed the technical conviction of the day that “the use of the radio telephone for communication between single individuals as in the case or ordinary telephone is a perfectly hopeless notion.” Yet Hoover’s licensing regime was able, many years later, to accommodate cellular telephony. Because the government retains ultimate control over the spectrum, it can reallocate or reassign frequencies to allow for new possibilities. Unfortunately, that control comes at a cost in efficiency. The FCC didn’t license cellular until the 1980s, though such systems were technically feasible decades before. This delay represented a huge economic loss.
The property rights and commons models both remove the paralyzing requirement that any change in the configuration of spectrum rights require government approval. However, both risk pushing spectrum in sub-optimal directions that are difficult to reverse. Three main dangers for the property model are monopolization, hold-out and transaction costs. Spectrum owners may refuse to engage in transactions that would in a global sense be efficient, or transactions won’t take place because of the overhead of market pricing and negotiation. The perceived danger of the commons model is chaos. Either users will step on each other to the point where spectrum is useless, or the administrative costs of dispute resolution will be too great.
In both cases, the failures could be difficult to fix. Private property can only be taken back by the government with compensation, which can be a costly, political, and time-consuming process. Unlicensed devices, once sold, are under the control of individual users, making it difficult to switch from a commons to another form of spectrum rights.
Given the risks inherent in the property and the commons approach, the best spectrum policy framework should tolerate both. As I explain in greater detail in Part IV, the choice between property and commons is essentially a bet about scarcity and transaction costs. Where wireless devices can easily coexist, and transactions for proprietary transmission rights are costly, a commons is the right approach. I believe this will ultimately be the case more often then not, but such an outcome is not guaranteed. If equipment coexistence is expensive (aka “spectrum is scarce”) and transactions are cheap, exclusive property rights are a superior mechanism.
Fortunately, the property and commons models are not mutually exclusive. The FCC today controls some spectrum through traditional government licensing (for example, the broadcast bands), allows flexible use of some spectrum (the PCS bands), and provides for unlicensed commons (in the 2.4 GHz ISM band and the 5 GHz U-NII bands, as well as for Part 15 and UWB devices). Far from being a completely new idea, unlicensed operation has been recognized in FCC rules since 1938, in the form of Part 15 devices. Part 15 devices can transmit in a large section of the spectrum, so long as they operate with very low power (less than one watt).
Property and commons coexisting makes sense for other reasons. The easiest way to make a wireless commons work is to limit power output of devices. This significantly reduces the number of situations where two transmissions will be in conflict, because less power means less range. Unlicensed systems have therefore tended to be either short-range or highly directional. WiFi, for example, delivers signals less than 300 feet. Broadcast applications, which seek to blanket an area the size of a city with the same content, generally use high-power narrowband transmission. A single WiFi hotspot is not a substitute for an over-the-air television broadcast tower. On the other hand, a broadcast system cannot support millions of independent transmitters in a city the way WiFi can. It is more difficult to guarantee quality of service or equipment compatibility in an unlicensed environment than under an exclusive use model, but it is more difficult to allow for experimentation by equipment vendors under exclusive use.
Despite the intensity of the debate between property and commons advocates, the FCC, at least, sees no reason to choose. It adopted its secondary markets order and its decision to allocate 255 MHz of new spectrum for unlicensed operation on the same day in May 2003. And as noted above, the Spectrum Policy Task Force recommended expanded use of both the exclusive rights and commons paradigms. Both overcome the artificial scarcity and inefficiencies of the FCC licensing mechanism that has dominated up to now. What the FCC recognized is that spectrum scarcity is not a global value. It may exist in certain places and times, but not in others. Whether spectrum is scarce depends on propagation characteristics, other users, the nature of the desired service, and the state of technology.
The FCC was mistaken, however, in assuming that scarcity and transaction costs can be mapped mechanically to frequencies. Specifically, the FCC proposed that exclusive property rights be the primary mechanism at lower frequencies, and commons the primary mechanism above 50 GHz, because there is less scarcity and higher transaction costs at those higher frequencies. Such a blanket statement is simply not accurate. The FCC’s own Television Band Notice of Inquiry proposes allowing unlicensed devices to operate in the low-frequency broadcast bands, recognizing the significant amount of “white space” in those frequencies. The Northpoint technology that allows terrestrial and satellite systems to coexist reduces scarcity in ways that have nothing to do with the frequency it operates in. The FCC’s thinking is still constrained by the frequency-denominated spectrum fallacy, in which lower bands have many assigned licensees and most frequencies in the tens or hundreds of gigahertz are “empty.”
The way for property and commons to coexist is not to give the beachfront to one model and a few acres of empty desert to the other. That assumes too much about scarcity and transaction costs. If the assumptions prove wrong, the basic split among the two models would need to be changed, but at that point it would be too late. The “property” section of spectrum would be locked in place by the Constitutional ban on uncompensated takings, and the “commons” section would be filled with transmitters.
An appropriate division between commons and property in the spectrum would have to take into account the multitude of possible technologies and architectures for wireless communication, as well as the architectural biases of each regime. Each regime can hypothetically support any outcome, but each creates incentives that make certain business models more or less viable.
The previous Part detailed the problems with the conventional understanding of spectrum, concluding that the optimal regime would be on that accommodated both property rights and commons mechanisms in a sensible way. In this Part, I develop such a proposal. The building blocks of this new framework are bundles of use rights associated with wireless communications equipment. As explained above, nothing else is real. After reconstructing the spectrum debate in terms of equipment rights, I outline a proposal for expanding on previous property and commons ideas.
All the spectrum policy regimes involve property rights. These are not rights in spectrum or frequencies, but in equipment. The seemingly wide gap between the property and commons models is actually two small differences in the configuration of those rights. The exclusive rights model vests rights in the first instance in intermediary service providers, while the commons model generally vests them in end-users. And the exclusive use model associates the transmission rights with correlative duties upon other transmitters that the commons model does not impose. Proponents of both models have typically linked them to frequencies, though there is no fundamental reason to do so.
A wireless communications model without the limiting assumptions of the spectrum fallacy could exploit the many dimensions of freedom for adding capacity identified in the previous Part. It would be similar to the commons model, but would find virtual commons throughout the spectrum. Such a model, which I call supercommons, combines incremental experimentation from current baseline licenses with a universal access privilege wherever a transmission would not be harmful to other systems. The white space that is now off-limits may turn out to be bigger than the entire “usable” spectrum today. It is as though we have been mining quartz and tossing aside the shiny nuggets of gold that are pulled up as well.
1) Uncertain borders
The debate about whether to turn spectrum into private property, or to open up unlicensed commons, has often ignored the critical issue of how to define the rights involved. Saying there should be wireless property rights makes no sense without a clear understanding of what those rights entail. A vaguely defined right invites litigation, which increases the administrative costs of the property system. As the nations of the former Soviet Union and Eastern Europe have found, capitalist mechanisms that thrive elsewhere fail if implemented poorly. For land, there are centuries of legal precedent and webs of formal regulatory as well as informal customary boundaries that make abstract property rights useful. This backdrop does not exist for spectrum, which, to reiterate, is not a physical resource like land.
Nor does the possibility of private contracting after the rights have been assigned solve the problem. Parties engaged in such Coasian bargaining must understand what they have as a baseline for negotiation. Otherwise their agreements will be inefficient at best, impossible at worst, and costly to achieve regardless. Initial conditions matter. Inefficient initial definition of rights can be reinforced through path dependence, becoming an insurmountable obstacle to optimization. The terms of bargains that can be struck depend on the parameters of the rights granted. If spectrum property rights are all frequency-delimited, private negotiation will artificially be constrained in terms of frequencies.
The boundaries of spectrum rights seem clear today because the FCC has restricted the possibility space, and because many techniques have only recently become technically feasible. Under the prevailing government licensing model, wireless transmission systems may not be altered beyond government-mandated specifications, which are tied to frequency bands. How far the licenses extend has been a moot point because no one else could test their limits without themselves running afoul of regulatory constraints.
The FCC’s ultra-wideband (UWB) proceeding gives a glimmer of the difficulties to be expected with ill-defined spectrum rights under a property or commons model. The FCC proposed in 1998 to authorize UWB devices across a wide swath of licensed spectrum, on the grounds that such devices used extremely low power and could coexist without causing harmful interference. It finally did so in February 2002.
The UWB proceeding was intensely controversial, stretching over four years and generating many hundreds of comments and significant Congressional interest. A central issue was whether the UWB devices would be detrimental to various licensed users of the spectrum, including public safety services, Global Positioning System receivers, and mobile phone systems. Despite intense pressure, the FCC concluded they would not. However, the Commission imposed extremely conservative initial limits on UWB to protect incumbent services. The FCC stated that it planned to review those limits in the future, to determine if they could be relaxed.
The UWB proceeding illustrates how difficult it can be to analyze interference claims when non-frequency limited communications systems are involved. It also provides a good example of how fuzzy spectrum property rights would be. Sprint, which operates a licensed Personal Communications System (PCS) cellular network, opposed authorization of UWB on the grounds that its FCC licenses gave it exclusivity over the bands where it operated. In other words, Sprint claimed its license implicitly granted it a degree of control similar to an exclusive use property right, even below the noise floor.
The FCC rejected Sprint’s claims, stating that “[t]his spectrum is not, and has never been, exclusive to Sprint or to any other licensee or user.” However, this is no guarantee that a court would hold similarly if Sprint’s license were a property right. The FCC after all, was the very agency that granted Sprint its license, and that license had explicit limitations and conditions.
It is one thing for the FCC to hold that the license, which already carved out a space for low-power Part 15 devices, also was not exclusive to underlay UWB devices. It would be a different thing for a court to conclude that a Sprint with exclusive property rights was subject to the same limitation. If Sprint and the FCC disagree today about the boundaries of Sprint’s license, how will the answer be any clearer with no authoritative FCC in the picture? The judicial process of fixing such boundaries, and then transacting around them, would create costs and uncertainties for both Sprint and potential UWB entrants.
The FCC’s determination doesn’t mean Sprint has no rights against other users in the frequencies covered by its licenses, merely that its rights are not absolute. The FCC couldn’t authorize an identical PCS system in the area of Sprint’s license that would cause massive interference, or, to take an extreme case, a PCS jamming service. Sprint’s rights must have some boundaries. They are just not well-defined. And that means the rights of erstwhile UWB transmitters that wish to underlay Sprint’s system are equally ill-defined.
What about the FCC’s commitment to review those rules? The FCC’s imposition of stringent limits on UWB was based on the expectations that a cautious approach would not cause too great a loss for the UWB market, because those limitations could be relaxed in the future. Why should the UWB vendors lose the possibility of relaxing the restrictions, or be forced to pay each property rights owner for a larger underlay, simply because inalienable licenses have turned into permanent property rights? The UWB underlay commons is property as well, and UWB device owners have a right to use their equipment like the owners of devices operating in licensed spectrum. On the other hand, exclusive use property owners can make a legitimate argument that the underlay rights must end somewhere, or their property rights have no meaning.
The question, therefore, is how to draw boundaries. And those boundaries will change over time, based on evolving technology and usage of both the exclusive and underlay portions of the spectrum.
The UWB case is one example of a much larger problem. The rights encoded in existing FCC licenses are broadly under-specified or mis-specified. They seem clear only because participants in the market have been artificially constrained from pushing on them for so long. If we want to move from the brittle and inefficient government licensing model to one that offers significantly greater flexibility, as both the property and commons camps propose, the content and contours of wireless communications rights will be extremely important. They will be the baselines for negotiations, litigation, or technical standards.
Consider a conventional high-power television broadcaster today. What exactly does it possess? The traditional formulation is that it has a government-issued exclusive license to a six-megahertz range of frequencies, with certain geographic, service, and technical limitations. Though such a description has proven useful over the years, it is misleading. The frequencies do not belong to the government to license to the broadcaster; they are simply a parameter of a physical phenomenon, electromagnetic radiation. The restrictions on the license are expressed in terms such as power output, antenna location, waveforms, or protocols, which are characteristics of equipment.
In the broadcast case, the license also incorporates a mandatory protocol, NTSC, which applies not only to the broadcaster’s own equipment, but to FCC-certified television receivers. The ability to broadcast would be of no value if standard televisions could not receive those broadcasts. Those receivers, however, are neither built nor owned by the broadcaster, and they receive many frequencies other than the ones our broadcaster controls.
How far does the broadcaster’s exclusive license in the frequencies extend? “Exclusive” would seem to imply that no one else may transmit in those frequencies. However, as the Sprint UWB challenge shows, that statement is too broad. Many intentional or unintentional emitters of radiation will radiate discernable signals within the frequency range and geographic footprint of the license. Two physically adjacent licensees in the same frequencies will experience mutual interference or cross-fading at some intermediate point. A transmitter in an adjacent frequency to the license using different waveforms may disturb reception for some of the broadcaster’s customers, even though it isn’t “in” the licensed frequency. The ever-present “noise floor” will vary over time and location, and at some distance from the licensed broadcast transmitter the signal may be difficult to pick up purely due to ambient noise. In portions of the spectrum other than the broadcast bands, the FCC has authorized low-power Part 15 and ultra-wideband devices to operate in the same frequencies as licensed services. All of these are implicit constraints on the licensee’s exclusive rights.
The FCC expresses the licensee’s exclusivity as freedom from “harmful interference,” which is defined as unwanted energy that “seriously degrades, obstructs or repeatedly interrupts” a licensed radio communications service. In the static, broadcast-oriented environment that prevailed when the government licensing regime was established, these concepts have some meaning. An observer could compare the television picture with and without an offending transmission and make a judgment about whether the degradation is “serious,” for example. Even in this simple case, though, there may be difficulties. The same signal will more seriously degrade the picture on a TV far from the transmitter. The license, however, is expressed in terms such as frequency and power, not geography, because the radio waves do not stop at geographic boundaries.
Now assume that the broadcaster’s license is converted into a fee simple property right. The time duration, public interest obligations, and restrictions on alienability in the FCC-granted license go away. However, the frequencies are still just a parameter of equipment, not something the broadcaster can own. In fact, surprisingly little has changed. The broadcaster still has a transmission right bounded by frequency and technical restrictions, and ill-defined protection against interference.
Though the broadcaster now has the nominal right to alter its service, for example changing from UHF television to cellular telephony, in practice it is limited to the dimensions of the erstwhile FCC license. This is because the technical parameters of the license were designed to prevent interference between the licensee and other licensees. Changing the service by altering the power, transmitter or antenna location, or waveforms may cause serious degradation of adjacent property rights holders. Or it may cause them to claim such degradation as a means to preclude competitive entry.
Deciding whether the interference claim is legitimate will not be easy. The dimensions of the existing license were defined with regard to the authorized service, and the dimensions of adjacent licenses were also defined based on assumptions about interference from the authorized service. If other spectrum owners complain that some of the hundreds of towers that make up the new cellular data network are causing interference that the single central UHF broadcast tower did not, some dispute resolution mechanism must come into play. Common law doctrines of nuisance and trespass may be used to resolve such disputes, but the only recognizable baseline will be the pre-existing license constraints.
Property proponents have advanced two concrete mechanisms for defining spectrum rights. The first is organic evolution of common law. Tom Hazlett argues that the Federal Radio Act of 1927 stopped the natural development of a spectrum jurisprudence in state courts based on the doctrine of “first in time, first in right.” This claim is based on a single unreported state case and is therefore problematic on its own terms. Even granting that a first in time doctrine would have worked in 1927, however, does not mean it is anywhere near sufficient in the 21st century. The case of two well-defined adjacent broadcasters is light years away from the situation of many roving ad hoc transmitters using some combination of spread-spectrum, directional antennas, meshed networking, and software-defined radio. Saying that the first transmitter in a frequency has rights against any future transmissions in the frequency is simply a recapitulation of the problematic FCC licenses.
The second set of proposals involve formal definitions of initial property rights. The effort led by Arthur De Vany is the most detailed. Unfortunately, the oft-cited law review article by De Vany and his collaborators was published in 1969, virtually as close to the creation of the FCC in 1934 as it is to the present day. De Vany’s system, like Hazlett’s aspirational common law, implicitly assumes the static broadcast world of the past rather than the dynamic environment today’s technologies make possible. It is a means to resolve interference among users of frequencies, when neither interference nor frequency hold sway in the way they once did.
For example, the De Vany system delineates rights in terms of total received field strength. Consequently, it always treats multipath propagation as a harmful phenomenon. A transmitter is liable if multipath reflection causes another receiver to encounter more unwanted energy than the transmitter is permitted to radiate. However, as described above, today’s intelligent systems can use multipath effects to enhance communication. There is no way to match the capacity-enhancing value of a multipath-aware system against the increased costs the unwanted energy imposes on other systems, because the multipath effect is always treated as harmful. The De Vany proposal would have the perverse effect of restricting techniques that improve spectral efficiency.
Moving from a government licensing system to a property system does not eliminate, or even significantly reduce, the difficult conflicts that arise among overlapping wireless transmissions. The rights a spectrum property holder would possess are in many ways identical to those a licensee possesses today. The major difference is in the way the configuration of rights can be changed. Any changes under government licensing must be reviewed and approved by the FCC. Under a property regime, the rights holder would be free to make any changes that do not alter the frequency band of transmission or interfere with other owners. That includes subdivision or sale of its rights. The owner could make other changes, but it must negotiate with owners who would have their own property rights affected.
Surprisingly, a commons environment is conceptually similar in many ways. A wireless commons is often thought of as an absence of property rights, or one in which transmission parameters are defined by government agencies or standards bodies rather than the holders of the transmission rights themselves. The first view is simply wrong. The second is relatively accurate in some cases, but fails to differentiate commons rights from exclusive property rights in any substantial way.
It is true that, in a commons, the rights involved are generally vested directly in device users, rather than in an intermediary such as a carrier. The carrier may impose additional limits on the devices that can operate within the space of the rights, but ultimately its transmission rights are passed through to the end-users who operate those devices. Whether wireless systems are user-defined or operator-defined has significant implications in terms of economic incentives, innovation, and the social values of the resulting communications. These differences may well be determinative in choosing one regime or the other. The policy decision should turn on such considerations, however, not on some simplistic comparison of “property” and “regulation.”
In a commons environment, as under the property regime, rights-holders are entitled to transmit under defined parameters, and to transfer their rights to others. What is missing is protection against incursions from other transmitters. Commons rights-holders may not claim protection against other commons rights-holders. Under current FCC rules, they also may not do so against licensed transmissions.
2) A legal-realist perspective on spectrum
Whether the regime is government licensing, property, or commons, what is at issue are usage rights for wireless communications equipment. Just as the property right in a gun allows its owner to shoot it at a firing range but not, absent extraordinary threats, on a crowded street, the property right in a wireless transmitter allows its owner to emit certain kinds of radiation but not others.
Table 1: Comparison of Transmission Rights
A right to transmit can be described or circumscribed in any number of ways, some of which involve frequencies and some of which don’t. However, wireless communications rights can be distinguished based on two major parameters: what they allow the holder to do, and the correlative duties (if any) they impose on others.
The rights in FCC-granted licenses are heavily encumbered. They do not include alienability: the ability to sell or subdivide the right. Nor, in most cases, do they permit the licensee to alter the physical parameters or service offering it delivers. The property camp attacks these limitations. Their primary claims is that the spectrum resource will be more efficiently exploited if the rights to exploit it can be bought, sold, traded, torn apart, and recombined.
Yet the lack of alienability is not the only characteristic of traditional government spectrum licenses. The licenses give something – a right to transmit in certain frequencies – but also take something away. They prevent others from transmitting in that same frequency by defining such transmissions as interference. In fact, all property rights necessarily have a reciprocal character. This is because an individual’s property rights mean nothing in isolation. Individual ownership implies that there are other owners, and therefore that there is not just property but a property rights system. By commission or omission, any statement in that system about one owner’s right is a statement about other owners as well.
The idea that rights impose duties on others was explored in detail by the legal realist Wesley Hohfeld early in the last century. Hohfeld’s formal language of jural logic eliminates the sloppiness that often mars discussions of rights. A right is an entitlement to act, which in the wireless case means the ability to emit radiation with the authorized power, frequency, or other characteristics. Its opposite is “no right.” Hohfeld’s innovation was to recognize that in addition to opposites, these categories have what he called “correlatives.” Where opposites concern the same entity, correlatives describe the effect of the category on others. The correlative of a right is a duty. In other words, a broadcaster’s right to transmit on certain frequencies imposes a duty upon the rest of the world not to do so, or face penalties. An affirmative duty on others is more precise than a right to exclude, which incorrectly suggests an all-or-nothing physical boundary.
Duty also has an opposite, called privilege. Tying the relationships neatly together, the correlative of a privilege is no right. In other words, a privilege to transmit, unlike a right, does not impose a duty on others. It doesn’t by itself give them a right to transmit either; they are simply spectators. A spectrum commons is a system for wireless communications built on privileges rather than rights. Each device user has a privilege to transmit, but that doesn’t come with a club to prohibit others in the commons from doing so.
The Hohfeldian framework provides a better way to understand what is being granted to spectrum property holders or commons participants. It does not, however, indicate how to define the rights or privileges that are granted. Even if the category is clear, conflicts will arise at the boundaries of the right or privilege. Such conflicts are pervasive in all forms of property. For wireless communications they are even more pronounced than usual. As described above, overlap is a ubiquitous aspect of wireless communication. Every transmission may impinge on the rights of others. Whether or how users of wireless communications devices impose correlative duties on others is thus absolutely critical.
Reconceptualizing the spectrum debate in terms of usage rights illuminates the space of possibilities. The legal regime should take into account the full web of responsibilities and opportunities that can inure to wireless devices, in all their possible gradations. A license or property right based on frequencies privileges frequency-based techniques. Alternative mechanisms such as wideband underlay or opportunistic sharing through cognitive radio are square pegs among the round holes of frequency blocks, presumptively prohibited and difficult to accommodate. Consequently, a variety of techniques that could improve efficiency of spectrum usage are ignored or barred.
The Northpoint case is illustrative. Northpoint developed a system to underlay existing satellite broadcast systems by distinguishing terrestrial and satellite transmissions based on the angle of arrival. Devices smart enough to know that satellites are overhead and ground based transmitters are not can send and receive signals in the same frequency bands without interference. Northpoint’s system is backward-compatible; the existing satellite system need make no changes to its equipment to coexist with the new terrestrial system.
This sounds like a huge win: a way to get something for nothing. Given the alleged spectrum drought and the FCC’s promotion of technological innovation, one might think Northpoint’s system would be quickly approved. Yet it wasn’t. Like ultra-wideband, another technology that didn’t fit the frequency-oriented paradigm, Northpoint’s proposed service was the subject of bitter, protracted regulatory wrangling. Nearly a decade after Northpoint first brought its technology to the FCC, the company still has not gained the spectrum access it seeks.
The Northpoint case looks like yet another example of how government allocation of spectrum is political and inefficient. It shows the public choice problems with a system that vests allocation and assignment decisions in regulators. Property rights, however, would not necessarily do better. If the incumbent satellite system had a property right rather than a government-issued license, it could make the decision through private bargaining whether to allow Northpoint’s system to operate. Or could it?
First, that assumes the satellite system’s property right is exclusive to all transmissions in its frequencies, even though that do not interfere because they operate on another dimension (angle of arrival). That control is part of the FCC’s residual authority today, but that doesn’t mean it would be part of the satellite operator’s property right tomorrow. Saying that property rights will be “exhaustive” assumes potential transmission mechanisms can be enumerated ahead of time. As I have explained, however, that assumes too much.
Second, it would be difficult to determine the price and terms for such a subdivision of the satellite system’s property. With no precedent or comparable transactions, the results are unlikely to be efficient. If in fact Northpoint’s system causes no interference with the existing satellite transmission, whatever the property owner charges is an inefficient rent that reduces the likelihood Northpoint will find it economically worthwhile to deploy. Perhaps the property system will get to the capacity-maximizing result more quickly than the government did. Then again, perhaps not.
The narrowness of the traditional frequency-based view of spectrum also corrupts the debate among property and commons models. The alternative to property rights in frequency blocks is more than frequency blocks open to any device following technical protocols. That is too limited a formulation. It ignores many possibilities, including underlay uses such as ultra-wideband, opportunistic sharing such as agile radio in the broadcast television guard bands, short-range communications within the home, and truly unregulated frequency blocks. As I have stressed, the real choice is among configurations of usage rights in wireless communications devices.
Focusing on the devices is also important because many of the new techniques for enhancing spectral efficiency are by nature cooperative. They use other devices as part of a communications system that is more efficient than any simple transmitter-receiver pair could be alone. The result is diversity or cooperation gain, which I outlined in Part II. The “same” spectrum may support more devices, more capacity, or different kinds of devices based on the overall architecture of the communications systems those devices support. What matters, therefore, is how and whether the devices can be configured to participate in such cooperative efforts. Though the end result applies to systems of many devices, that is a question about the properties of each individual device. It is not a question about the spectrum in which they operate.
The best regime for allocating spectrum and resolving disputes would be decentralized. The basic problem with the FCC as the arbiter of spectrum use rights is not that it is a public actor instead of private market participants. The FCC cannot possibly be efficient because it is a single actor seeking to allocate resources ahead of time. Fortunately, as complexity science has demonstrated, complex adaptive systems can self-organize without any central control mechanism. In other words, a collection of independent actors, all seeking to maximize their own welfare, may produce a more efficient global result than any central traffic cop. This is in some ways an elaboration and expansions of Adam Smith’s insights about self-interested economic activity, with a more rigorous explanation of how and why the “invisible hand” operates.
Consider by analogy the two very different legal regimes called copyright. Originally, the term was used in connection with royal printing monopolies. Those who wished to print needed a copyright from the Crown. This system operated very much like the government licensing system for spectrum. It centralized control in an actor impervious from market forces, predictably depressing production, stifling speech, and allowing inappropriate factors to influence decision-making.
One solution would be to take the right to control printing away from the king and give it to private printers in the form of a charter or property right. The printers could then determine who could publish, and could buy or sell their charters to do so. This approach decentralizes rights, but only somewhat. The goal is to maximize the social welfare from printing, and print has enormous positive externalities. A better solution would be to give individual authors the right to publish, constrained only by their economic ability to pay the relevant costs, and establish exclusive rights in the publications themselves. The last part is of course what we today call copyright.
1) Beyond unlicensed bands
Commons has from its beginning been associated with the FCC’s existing unlicensed bands and with the WiFi and U-NII systems that operate in those bands. There are many good reasons for designating frequency bands for unlicensed operation rather than granting licenses or exclusive rights in those frequencies. That doesn’t mean unlicensed bands are the full realization of the commons model. In fact, they are but a small part of it. The commons comprises all those virtual locations where a wireless system can operate on the basis of privileges rather than rights. As discussed in Part II, there are a large and growing number of mechanisms that can create non-interfering “white space” throughout the spectrum. Like the invisible dark matter that cosmologists infer must make up the bulk of the universe, the wireless commons may be all around us.
The primary reason to broaden the commons is that it could allow many additional wireless systems that otherwise would not be permitted to operate, expanding the capacity of the spectrum. There are, however, other benefits. Opponent of the commons approach such as Stuart Benjamin equate the wireless commons with regulated unlicensed bands, and then critique it for being subject to inefficiencies of government management that private market arrangements avoid. This line of attack rests on the false assumption that the commons relies on a specific legal or technical approach.
Recent scholarship acknowledges that property and commons models should coexist. Property advocates such as Faulhaber and Farber argue that something like a commons can be established within the property regime (in the form of private or public “parks”) or alongside property rights (thought easements). Hazlett proposes overlay and underlay rights in broadcast spectrum. Noam suggests that access could be open but still subject to a price mechanism. Benkler, though rejecting the assumption that property rights should be the baseline, nonetheless proposes experiments with both property and commons systems for a period of time, building in reversibility once the experiment is over.
Once some frequency bands are turned into property and others are opened up as commons, what remains are all the opportunities to communicate without dominating a frequency. This superstructure may have more capacity than all established wireless systems exploit today. It should be open to innovation and free from the forced exclusivity of formalized property rights.
In other words, the commons would be the baseline, with property encompassed within it, rather than the reverse. The initial legal rule for wireless communication should be universal access. Anyone would be permitted to transmit anywhere, at any time, in any manner, so long as they did not impose an excessive burden on others. Conflicts among users of wireless devices should be addressed through a “negative” regime of tort and an “affirmative” regime of safe harbors. To a first approximation, this regime collapses to the system we have today, but it allows for expansion, potentially in radical directions.
I call this regime supercommons.
As the name suggests, the supercommons is a superset of the approach that Benkler and Lessig advocate. Benkler’s primary policy recommendation is to expand the frequency bands dedicated to unlicensed use, either by loosening rules or by adding additional bands. Entry would be based on technical standards developed in part by private standards bodies and endorsed by central authorities.
The supercommons picks up where these steps, which I endorse, leave off. The two differences are that the supercommons is not limited to designated frequencies, and it uses tort and other backstops rather than equipment standards as the legal threshold for entry. Benkler hints at something similar in his proposal for underlay and interweaving privileges for ultra-wideband and software-defined radio. However, he devotes a single paragraph to the idea, and still suggests ex ante technical requirements rather than common law backstops.
2) The space around exclusivity
Universal access is not as radical as it may seem. Portions of the U-NII band, for example, are limited in terms of power and little else. The FCC’s UWB decision, rejecting Sprint’s claim that its license rights granted it exclusivity within its bands, suggests there is some white space that is not part of the rights granted to licensees. The FCC in 1981 floated the idea of allowing spread-spectrum devices to underlay licensed transmitters throughout the spectrum, though it never followed through. Noam proposes an “open access” regime that would only require that prospective wireless users pay a dynamically calibrated price for the privilege. In the real world, Haiti appears to have no limits on use of devices in its unlicensed bands, though admittedly it is a less-crowded spectral environment than the US.
The supercommons also parallels the FCC’s longstanding policy approach to the Internet. The FCC has never subjected Internet services to most of its requirements for telecommunications services. However, Internet services are clearly within the FCC’s broad jurisdiction. What one FCC staff working paper calls “unregulation” was a conscious decision not to impose certain rules, and a further decision to fence off certain kinds of services from incumbents.
The important parallel with spectrum is that the FCC’s policy, when formulated in the 1980s, was made amid deep uncertainties about how technology and markets would develop. The FCC didn’t predict the emergence of the commercial Internet. Its officials were as surprised as the private sector about the growth of the World Wide Web. What the FCC appreciated was that computers attached to communications networks could create fundamentally new types of services and applications. Without both regulatory restraint and protective regulatory intervention, those new services could be stillborn. So the FCC created a protected space for innovation, which it called enhanced services.
The uncertainties about spectrum are of a similar character. We know that powerful computers controlling radios have the potential to revolutionize wireless communications, just as the FCC knew that powerful computers attached to communications networks had great potential. We don’t know how and when those revolutionary possibilities will manifest themselves. For example, the degree to which interference will be a significant practical problem can only be guessed at. A legal regime that assumes interference will always be a serious problem requiring costly price or property rules is as unjustified as one that eliminates all legal protections against interference today.
The supercommons also bears some similarities to fair use in copyright law. Fair use helps to reconcile the contradictory public policy imperatives to protect property rights and free expression. It also mitigates market failure when transaction costs of licenses are excessive, when societal benefits of non-commercial sharing are external to the copyright holder, or when new technologies cannot easily be accommodated in the copyright framework.
In short, fair use is outside but not opposed to the exclusive rights copyright grants. It is a realm of unconstrained sharing that balances a complex array of competing claims on published work. All of these rationales can be applied to supercommons transmissions around the exclusive transmission rights that administrative licensing or private ownership guarantee. The primary difference is that fair use is limited to functions such as education and parody that do not directly compete with the primary commercial exploitation of the work. The supercommons is a full-fledged communications space that may be utilized for any purpose.
The universal access privilege, in effect, says that any transmission that is not otherwise prohibited is allowed, though whether it is subject to a Hohfeldian privilege depends on whether it exceeds a flexible set of boundaries developed through decentralized legal mechanisms. This proposal reverses the current approach, under which actions must be expressly authorized by the government, or in a property regime by the property owner. It resembles the unambiguous language of the First Amendment, which is nonetheless limited and balanced in application.
The supercommons approach can theoretically apply to the entire radio frequency spectrum. However, two sets of frequencies deserve special treatment. Radio astronomy bands such as broadcast channel 37 must be as free as possible of any other signals to maximize reception of distant astronomical phenomena. In the interests of science, intentional transmissions there should continue to be prohibited. There is no means to weigh potential benefits against the societal losses if basic research into the nature of the universe were disrupted.
Other non-commercial bands for uses such as public safety, military, and aerospace communications do not necessarily raise such concerns. The same boundary enforcement mechanisms and incentives that protect commercial wireless systems could apply to these uses, perhaps with more-protective default rules or liability standards. Technical mechanisms could be used to “reclaim” spectrum during times of emergency. However, in an abundance of caution, some or all of the non-commercial frequencies could be declared off-limits from supercommons transmission for an initial period.
3) Universal entry in a world of uncertainty
If wireless communications will be limited in practice, why adopt a default rule of universal access? Because there is such profound uncertainty about where to draw boundaries between permissible and impermissible uses. The danger of any preconceived limitations on wireless communications techniques is that they will be self-fulfilling prophesies. They will make spectrum scarce by ruling out approaches that could mitigate scarcity. Or they will push usage toward sub-optimal equilibria.
Going forward, the only certainty is that wireless technology will evolve through more powerful computing devices and more sophisticated mechanisms to share spectrum. Legal rules that make innovative forms of communication too difficult will produce sub-optimal results, because there will be less communication, smaller markets for services or equipment, and less innovation. As I’ve stressed repeatedly, wireless systems must tolerate some interference. At some level, though, the cost of upgrading systems to be robust to interference, or the administrative costs of detecting and punishing those responsible, will be greater than the benefits from the additional communications.
The property vs. commons debate has largely hinged on which of these scenarios is more likely. One side worries about innovative technologies being blocked by spectrum owners; the other worries about legitimate systems being rendered useless by interference. Yet there is another way rules can fail. They may be too brittle. They may strike an appropriate balance between entry and protection, but be unable to adapt when conditions change. What is impractical or unlikely today may be routine in the future, thanks to changing technology and usage patterns. The optimal level of interference is steadily increasing. A legal regime that cannot adapt quickly is doomed to failure.
For this reason, the solution is not simply to define existing rights more concretely. As noted above, the boundaries of existing wireless licenses are vague. Yet eliminating that vagueness without providing an efficient mechanism for change could make things worse. The defined boundary might be too high or too low to accommodate efficient entry. It might be right for urban areas but not rural ones, or for certain system architectures but not others. It would require the FCC to engage in a top-down definitional exercise, subject to the same inefficiencies as the current licensing process. And it would likely consider only some sharing mechanisms, such as low-power underlay, excluding other possibilities.
As a threshold matter, if the legal regime is going to be wrong, better for it to be too lenient towards new services. If incumbents know that they must tolerate other systems and that the boundaries of their rights are subject to review, they will have incentives to make their receivers more robust. This is itself a beneficial outcome. Better receivers mean less interference, in effect increasing the available spectrum.
The optimal situation is one in which each user takes steps that increase the marginal capacity of the spectrum as a whole which exceed the marginal cost of the step. Where users enjoy legally enforceable protection against interference, they have no incentive to exercise a higher standard of care, even if such steps were efficient from the view of the system as a whole. The only exception is if prospective entrants could pay the incumbents to improve their equipment. Such transactions are unlikely for reasons I discuss below. Even if they did occur, the least cost avoider of the harm is likely to be the incumbent, who can design robustness into its original equipment specifications.
In proposing a universal access privilege, I am not arguing for anarchy. Even when rights are expressed in absolute terms, common law doctrines, administrative rules, custom, and economic interests all function to constrain what real-world actors do. Undoubtedly there will be conflicts, especially at first. That will be true under any new spectrum regime. The mere possibility of disputes is not a reason to reject the supercommons, if those disputes can be avoided or resolved in an efficient manner.
In this section, I first discuss why a regime built on a principle of universal access will not necessarily lead to chaos. I then outline the basics of a legal regime that could address those disputes that do arise among competing wireless users.
1) Of commons and tragedies
The idea that unconstrained entry causes a tragedy of the commons is so patently obvious to most scholars as to hardly need defending. Hazlett calls this point “undisputed.” Consider it disputed. Try walking a few blocks down Fifth Avenue in midtown Manhattan during the Christmas shopping season. The sea of humanity surging along the sidewalks will slow your trek, but it won’t stop it. People may jostle one another, but fistfights are quite rare. Why doesn’t this street scene, with open entry, potential for “interference,” and incentives for free riding degenerate into a melee? The answer is that individual actors in a complex adaptive systems can sometimes self-organize and find globally efficient arrangements. They don’t need property rights or price mechanisms to do so.
Garrett Hardin may have coined the phrase “tragedy of the commons”, but that every commons doesn’t lead to a tragedy. For example, people don’t chop down trees and build houses in public parks, because there are rules and enforcement mechanisms to preserve the public character of the space. CB radio may have failed for peculiar reasons not universal in shared forms of wireless communication. Scholars such as Elinor Ostrom have examined conditions under which commons are self-regulating. Stuart Buck has applied this work to wireless communications, arguing that spectrum can be thought of as a common pool resource.
Countering the tendency toward over-exploitation of any commons is the fact that wireless communications systems involve strong network effects. One transmitter is as valuable as one fax machine, which is not very much. Systems gain value the more users they can support. This is why WiFi was such a significant boost to usage of the 2.4 GHz band. The band was available for unlicensed devices for years before WiFi was standardized. Then and now, non-WiFi devices operate in the same band, but they have a far smaller market. Vendors build devices to the WiFi standard because they benefit from interoperability with millions of existing users of other vendors’ products. In the future, manufacturers who want to sell large numbers of devices will have incentives to build devices that don’t cause a paralyzing tragedy of the commons or constant litigation over interference with other services. Both situations would limit or even destroy the market for devices.
Incentives on equipment manufacturers may prevent theoretical conflicts from appearing. Assume A and B are communicating on an unlicensed basis. C arrives on the scene and wants to send a message to D, but doing so on the same channel would result in neither message being received. C could use higher power and “shout” above A and B, assuming there was enough headroom in the relevant power limits. This might work at first, but would ultimately provoke an arms race that prevented anyone from communicating effectively. We must take a step back. C is constrained by two conditions – it wants to communicate, and it is using a particular piece of equipment. C’s equipment can only do what its manufacturer allows. The manufacturer wants to sell lots of equipment, which means it wants to see lots of communication. It therefore has an incentive not to sell C a device that makes it easy to produce the shouting match described above. There is no law preventing C from buying such a device, but perhaps there need not be.
The evolution of unlicensed wireless local area networking standards shows how this dynamic works. Two other standards were developed around the same time as the WiFi: Bluetooth and HomeRF. Bluetooth is a short-range low-power specification initially developed as a wire replacement, and HomeRF was designed for networking computers within the home. All three used the same 2.4 GHz band. Because they employed different technical forms of spread-spectrum, there were concerns the three technologies would interfere with one another if used in close proximity. The feared conflicts never arose. WiFi obliterated HomeRF in the marketplace, causing vendors to switch to the winning technology. Meanwhile, the relevant standards bodies implemented technical modifications, some of which had to be incorporated by the FCC into the 2.4 GHz rules, to ease the potential collisions between the systems. It is simply in no vendor’s interest to sell devices that don’t work.
As wireless devices become smarter, participants in the supercommons will have a wider range of choices for avoiding conflicts with other users. Their decision will involve whether to expend some additional cost – either in a more sophisticated device or in using a different method to communicate – or try a brute-force approach in the hope that someone else won’t have the same idea. It is wrong to assume the answer is necessarily the same as it would be for shepherds in a common meadow.
2) The common law of spectrum
When the incentives described in the previous section are insufficient to ensure harmonious coexistence of unlicensed supercommons devices, a common law tort-like regime can take up the slack. Those aggrieved by the actions of a supercommons transmitter could sue for injury or violation of a duty of case, and obtain either damages or injunctive relief.
Many property advocates accept that a workable legal regime for spectrum disputes was in the process of developing prior to the Federal Radio Act of 1927. There were no rules prohibiting any kind of transmission in those days, save for the requirement of a license from the Commerce Department (which the courts held could not be denied). The claim is that, had the spectrum not been taken over by the federal government, courts would have developed common law doctrine turning broadcasters’ de facto rights into de jure private property, and would have fashioned rules on a case-by-case basis to resolve disputes as they arose.
It would be hypocritical for property advocates to argue that courts are a good means of resolving boundary disputes over spectrum in an environment of exclusive property rights, but not a good mechanism when there is a presumptive universal privilege to transmit. The supercommons is still a world of property rights. Those rights are vested in equipment, not the spectrum, but as I argue above, this is no different than under the exclusive rights model.
Coase offered an analogy, albeit for a different purpose, that precisely illustrates this point. He argued that spectrum was mistakenly treated as a physical thing because of a faulty analogy to the law of airspace. The problem was not so much that the association could not be made, but that it needlessly complicated the issue. A case involving a man who scared away a flock of ducks on a neighboring property by shooting his gun could, Coase explained, be decided based on the shooter’s rights to violate his neighbor’s airspace. However, a more straightforward analysis would focus on the shooter’s rights to use his gun.
Similarly, if there is a dispute about whether A’s transmitter improperly degrades B’s reception, we could look at whether A “trespassed” on B’s exclusive territory. Or we could consider whether A exceeded the bounds of his legal privilege to transmit. The only difference is the perspective, and the complexity of the analysis. The second formulation is the one I propose to apply to supercommons devices under the universal entry privilege.
The usual expectation is that trespass and nuisance, two doctrines that apply to land, would form the basis for a common law of spectrum. As Ellen Goodman explains in a forthcoming article, these concepts are often invoked in the spectrum concept, but seldom explained in detail. Goodman articulates how the application of nuisance or trespass law to disputes over spectrum property rights would not be as simple as property advocates assume. Goodman applies to spectrum the famous Cathedral framework originated by Guido Calabresi and Douglas Melamed for resolving property disputes.
Nuisance and trespass are doctrines deeply rooted in land. Because spectrum is not a physical resource, there is no precise analogy to the physical entry onto another person’s land. Trespass is especially inapt for spectrum. Trespass law is built on bright-line distinctions, with the central notion being intrusion on land owned by another. With spectrum, however, there is no difference between a signal that “intrudes” on spectrum controlled by another entity. All wireless signals potentially intrude on all others, because radio waves do not stop at any defined physical boundary. The same signal may be invisible in one case or totally prevent communication in another.
Nuisance law makes more sense in the wireless context. It does away with the requirement of physical intrusion and focuses attention property on the effects of different uses of property. Modern nuisance law requires a finding that a nuisance is substantial and unreasonable in order to impose damages or injunctive relief. As Goodman explains, these tests require a public interest determination very similar to what the FCC does today.
If we go down the path of using tort concepts to assign liability for impermissible uses of wireless devices, why only apply this regime to property holders? Nuisance and trespass assume the injured party is also a land owner. Torts, however, include all sorts of injuries or accidents that do not involve property ownership. For example, products liability law imposes liability on defective products that cause harm. This body of doctrine has evolved to balance incentives on product manufacturers between protecting consumer safety and investing in new products. The modern rule is generally strict liability, imposing damages on the manufacturer without requiring a showing of negligence. However, liability generally requires proof of a harm and that the product was defective. A wireless transmitter that allows its user to emit signals that disrupt other communications systems can be thought of as a defective product that causes harm. The harm requirement bears a striking similarity to the FCC’s notion of “harmful interference.”
Tort is far from perfect. It could produce several kinds of errors in the wireless context. Courts could adopt too lenient a standard, forcing incumbents to tolerate more interference than they should. The delay and cost of litigation could produce a similar result. Incumbents might be forced to deal with excessive interference because they are forced to upgrade their equipment to protect their customers while litigation is pending. Or they may decline to pursue an action that would ultimately succeed, because the time, expense, and uncertainty involved were too great. In such cases the legal standard would be correct but the system would still be inefficient.
These risks, however, exist in most of tort law. Unauthorized transmission is not a pure harm. It produces a social benefit – more communication – that must be weighed against the social costs it also may impose. Courts have developed a range of mechanisms and doctrines to deal with these problems. I am not suggesting that importing product liability law whole hog into the wireless context would solve all the complexities of interference disputes. A device that causes harm only when used in certain places at certain times with certain settings is not obviously defective. Products liability law directly polices manufacturers, whereas in wireless the manufacturers are stepping stones to the real actors, who are users. My point is that tort has wrestled with the kinds of questions that a common law of wireless communications will need to address. The basic building blocks for a workable legal regime to deter, police, and punish inefficient and irresponsible signals are already there.
Interference is a negative externality of wireless communication. It is a cost that the transmitter imposes on others, that may not be figured into the transmitter’s economic decision about whether to transmit. The point of liability rules backing exclusive property rights or technical standards for unlicensed devices is to internalize those externalities. The reason for doing so is that the one doing the transmitting is likely to be the Coasian least-cost avoider of the undesirable result. Imposing a legal duty on transmitters therefore creates incentives to avoid excessive interference or to negotiate with those affected when the transmitter values its interfering transmission highly enough.
Tort does this by imposing duties of care. We are not always free to do as we please, regardless of whether we are doing so with our own property. In the modern administrative state, tort-based limits on action coexist with regulatory requirements. Bodies such as the Occupational Safety and Health Administration or the Consumer Product Safety Commission impose mandatory baselines upon actors they regulate. Regulation can function as a screen to overcome the inefficiencies of the court system. As I describe in the next section, a government entity could facilitate spectrum dispute resolution as well by fashioning default rules.
There are three classes of interference that could take place. Supercommons devices might unintentionally interfere with each other, they might unintentionally interfere with licensed systems, or they might deliberately degrade other transmissions. There are also two kinds of systems that can be harmed: private communications networks and public systems such as public safety networks, military radars, global positioning system satellite transmissions, and communication with airplanes.
One simple way to apply tort law to spectrum disputes would be to ratchet the duty of care up or down depending on the situation. Interference with public systems, which can cost lives, could be subject to strict liability as well as strong remedies including injunctive relief and even criminal punishments. As noted above, the supercommons might initially have a flat prohibition on transmissions in public bands. At the other extreme, suits for interference among unlicensed devices could require a showing of recklessness or even intentionality. In the middle, claims of interference between unlicensed devices and private networks subject to government licenses or exclusive property rights could use the negligence standard that is the most common threshold for ordinary torts.
Enforcement under the regime I’ve described would not be significantly more difficult than it is today. Under the current government licensing system, unauthorized “pirate radio” transmitters or authorized transmitters that cause harmful interference must be identified and held responsible for their actions. The FCC conducts regular monitoring, but the licensees affected by the interference have the best incentives to report and track down violators. An unlicensed device operator that blasts high-power signals in a way that causes significant degradation of other systems is no more difficult to find and prosecute than a pirate radio broadcaster today. Lower-power, more adaptive, and more mobile devices may be harder to track down, but for the same reasons they are less likely to cause significant, ongoing interference to other systems.
Finding the proper boundaries among growing networks of wireless devices is not an easy matter. As Goodman describes, it will necessarily involve a messy process informed by public interest calculations, and in all likelihood looking to a government regulator for help. If, however, a common law liability regime can work to resolve spectrum disputes, there is no reason to require pre-assigned exclusive property rights as the price of entry to that mechanism.
The uncertainties of a common-law liability regime suggest that experimentation would be valuable. The supercommons could be tried initially in a portion of the spectrum where the risks of failure are lower, for example, excluding public safety and military bands. Legal doctrines could be developed and tried out for a period of time. This could reduce the transaction costs of the liability system when applied to the whole spectrum.
E) Safe Harbors and Backstops
The common law liability regime for wireless disputes that I describe will function more efficiently if default rules can be established. Litigation is costly. “Rough justice” rules can reduce the number of situations in which market participants will avail themselves of the court, and will provide guidelines to equipment vendors and users considering engaging in a novel form of transmission. Below I list some mechanisms that could be employed. Further work would need to be done to determine which would make sense in what situations, and the specifics of how these safe harbors and backstops would be defined.
Commons opponents would argue that any safe harbor standards mandated or approved by the government are subject to significant inefficiencies and public choice concerns. The restrictions I’m describing, however, are not government-defined protocols. They are procedural mechanisms to facilitate private negotiations. Moreover, an exclusive property rights regime does not eliminate the need for such mechanisms.
1) Technical Standards
One option for supercommons devices is a mandatory certification regime, similar in some ways to the FCC’s current rules for unlicensed bands. Benkler proposes such a system, under which device standards approved by recognized standards bodies would receive fast-track approval by the FCC, and proprietary designs could be certified through a more detailed review.  A certification mechanism addresses the concern that a larger spectrum commons would collapse into a tragedy of mutual interference. Participants on technical standards bodies have an incentive to produce standards that allow for widespread usage of the spectrum in question. They are generally representatives of hardware vendors who want to sell more hardware, and they can only do so if there is a market for more devices.
With a common law liability system standing behind the standards, they become default rules or safe harbors, rather than mandates. One way to merge private standards with a liability regime is to shift the burden of proof if a device is certified to comply with technical standards adopted by a property constituted open industry standards body. Or industry best practices and standards could be incorporated into the “reasonable person” standard that is used as the liability test in assessing tort liability.
Another way to look at technical standards for supercommons devices is as external boundaries around hybrid limited common property. Buck examines this approach, drawing on the work of Carol Rose. Limited common property arrangements are exclusive to the outside world, but shared among participants. To use in the existing unlicensed bands, for example, equipment manufacturers must receive a certification that their devices comply with FCC-mandated technical requirements. WiFi devices are subject to a further requirement that they comply with industry-defined standards, and subject to interoperability testing by the Wi-Fi Alliance if they wish to use the WiFi trademark. Once devices pass those hurdles, they are free to operate wherever and whenever their owners desire.
An abundance of caution may justify imposing such restrictions on the supercommons initially. Any fixed technical requirements will prevent some transmissions that would be efficient to permit. However, reasonable technical standards would allow much more than either that status quo or a system of exclusive property rights.
2) Trademark concepts
An alternate safe harbor regime would be a publication mechanism similar to that used in trademark law. Those seeking federal protection for trademarks must submit an application to the Patent and Trademark Office (PTO), which engages in an initial review to determine if the trademark is confusingly similar to existing marks or is non-trademarkable, as with descriptive terms. The PTO then publishes the proposed trademark. This gives those who believe the application would infringe on their existing rights the opportunity to oppose the trademark prior to its issuance.
A similar mechanism could be used for wireless devices. Those who wished to deploy devices that used techniques or protocols not covered under existing authorizations would be required to publish a technical description of their system through an open publication system managed by the FCC or some other agency. The government would perform only a perfunctory review of the proposal to ensure it met basic technical and administrative requirements, and did not duplicate already-approved technologies. Those who felt the design would interfere with their own systems would have an opportunity to file an opposition. If none were filed after a reasonable period of time, the design would be approved. There would have to be limitations on frivolous filings to prevent companies from strategically opposing any new technology.
Trademark law provides other concepts that could profitably be incorporated into spectrum policy, as Tom Bell has noted. The key connection is that trademark rights are based on use, not on abstract ownership. A trademark owner gains access to remedies based either on actual use in commerce (through common law or registration), or on filing its bona fide intent to use the mark. Similarly, a wireless rights-holder should not be able to preclude transmissions that do not affect its own use of devices, even if they cross the wavelengths where it is entitled to transmit.
Finally, trademark law has wrestled with many of the same difficulties that arise in the wireless context. It has a formal taxonomy of industries – the SIC code system – to distinguish parallel uses that are, in the language of spectrum, per se non-interfering. It has a hybrid system of common-law and statutory rights, facilitated by a non-regulatory government agency. It even has a concept, dilution, similar to the “pollution” that can occur when wireless devices collectively raise the noise floor even though they are individually distinguishable.
The analogy between wireless communications and trademarks is just that. Beyond the publication concept outlined above, further analysis would be needed to determine how specifically trademark concepts could inform a supercommons wireless regime. It may turn out that the specific solutions used in trademark are inapposite here. Regardless, trademark shows how a legal regime can grow up to police conflicting uses in a complex space of economic activity, without assigning exclusive ownership rights in an abstract concept.
3) Interference temperature
The FCC’s concept of noise or interference temperature is another possible mechanism for per se boundaries. The FCC Spectrum Task Force introduced the idea in its November 2002 report, and FCC staff have indicated the Commission intends to initiate a formal proceeding on the concept. The starting point for the FCC’s proposal is that the boundaries of licensees’ rights are, as I’ve explained, not well-defined. In particular, the strength of a wireless signal decays as distance from the transmitter grows, meaning that the same signal will be more difficult to receive in some places than others. At some point, the signal will be so faint that it will drop below the ambient noise floor. That point is not always well-identified today, leaving licensees unsure whether the inability of the distant receiver to understand the transmission results from background noise (which it must tolerate) or an interfering signal (which it can ask the FCC to shut down).
The noise floor is not constant. It varies over time and location, based on the radiators (intentional or unintentional) and physical geography in any given area. Since the signal decays over distance, there may be areas within the geographic scope of the license where transmissions already cannot be received. As a practical matter, the licensed transmission can only guarantee reception where its signal strength exceeds the highest value of the noise floor.
The FCC’s interference temperature proposal seeks to formalize this noise floor boundary. It would define peak noise floor levels, presumably on a band-by-band basis, and permit unlicensed devices to operate presumptively below this floor. The FCC’s argument is that, since the licensed operators cannot reliably transmit below this level anyway, they are giving up nothing by accepting such underlay uses. In fact, the licensees would benefit from knowing exactly how far down their rights extend. Conversely, unlicensed devices would gain greater headroom above the one watt Part 15 limits, which are very low and do not vary across the spectrum.
Interference temperature is a novel concept. It has already provoked opposition from licensed wireless operators. And at best, it would provide a safe harbor only for low-power underlay, not for other supercommons mechanisms such as opportunistic sharing. Nonetheless, as a rough cut to minimize unnecessary litigation, interference temperature may be a useful metric. Its virtue is that it makes quantitative what is today undefined or uncertain. As long as it is not the only boundary between the supercommons and other devices, interference temperature may prove to be a useful measure. More work should be done to determine how to implement interference temperature, and to ensure that it provides enough room for additional unlicensed devices to operate.
Owners of real property could have a per se right to transmit however they wish within their own property. There is no reason to limit this right to certain frequency bands, as Chartier proposes, except possibly to exclude public safety bands. Transmissions that stay within a building are non-interfering with other users, except for those the building owner invites in. This rule would allow some equipment to be operated only within the home or other private property. The same device would no longer be subject to the safe harbor if used on a street corner or a neighbor’s lawn.
The local safe harbor works only for short-range devices. However, there may be significant markets and applications for such systems if restrictions on transmission are eased. Various mechanisms could be employed to ensure the devices are not mis-used, including geo-location technologies and beacons to check that the device is in range of some fixed point in the home.
5) Cognitive radio
Software-defined radio opens up a new set of possibilities for backstops. Cognitive radios can sense the spectrum around them, not transmit if they see a licensed signal, and move out of the way when a licensed signal appears after they’ve begun transmitting. A requirement that some unlicensed devices include such mechanisms would impose a significant cost tax on the equipment, but that might be worth it, especially if spectrum agility is already necessary for the kinds of activity the device engages in.
Devices could even be required to have remote cutoff or override capability. An authorized entity such as a court or a public safety authority could send out a signal that would temporarily or permanently disable the device. Such a capability would have to be designed carefully, and the devices might be subject to pranks or other attacks to disable them unjustifiably. However, this is no more a show-stopper than the fact that any wireless device is subject to eavesdropping or identity theft.
Cognitive radios are not yet on the market. However, the basic technology required is just a function of computational price/performance and miniaturization, which have followed steady improvement curves for decades. Once such devices are commercially feasible, they will have a disruptive effect. A cognitive radio is computer that happens to talk through radio waves. Computers are radically reconfigurable, since the cost of software upgrades is far less than hardware changes. When new communications techniques, standards, or protective measures involve software downloads rather than new equipment, adaptations that seemed infeasible become inevitable. Greater utility and lifespan ultimately offset higher initial cost.
Finally, if all else fails, various risk-sharing mechanisms are commonly used to deal with situations of potential liability. Manufacturers or users of unlicensed devices could purchase insurance against damages for interference. Requiring manufacturers to have insurance could guarantee that those harmed by over-reaching unlicensed devices would be made whole. It would also create incentives for the manufacturers to build “polite” devices, and ensure that their customers do not exceed the bounds of politeness. The insurer would become another private, decentralized actor pushing the unlicensed devices toward optimum coexistence with each other and with licensed devices. Bonding and indemnification could also be employed.
Proponents of exclusive property rights may argue that a supercommons, even if workable, fails to provide adequate certainty to incumbent users of the spectrum. If some new system might interfere, they will have less reason to spend money building their own systems. Yet this possibility already exists under the status quo. The supercommons regime still prohibits and imposes liability on systems that cause excessive harm to other transmitters. Where the supercommons might be different is the degree to which it protects exclusivity for its own sake, or enshrines poor quality receivers as the standard for interference. These are the cases where absolute protection of incumbents, whether through government licenses or exclusive spectrum rights, makes it harder to achieve the welfare-maximizing configuration of wireless devices.
The property rights system always vests rights in incumbents, which is inefficient on three levels. The incumbent may hold out or refuse to negotiate with an entrant that would enhance spectrum efficiency. The incumbent may be unaware of whether the new technology would work. Or the entrant itself may be unable to determine whether access is worth negotiating for, because even experimentation with possible shared use mechanisms can only occur at the sufferance of the incumbent rights-holder. The company best able to take the risks in developing and testing the new technology is the entrant, but it needs the freedom to do so. The supercommons regime gives the entrant the right to transmit in the first instance, which is likely to be the least-cost rights allocation.
The other problem with the investment incentives argument is that it is frequently used to justify inefficient monopolies. Companies invest in capital-intensive projects all the time without a guarantee they will earn a rate of return. It is reasonable for companies to want to understand the risks so they can make appropriate decisions about how to act, but the existence of some risk does not paralyze investment. Legitimate concerns about risk that investments will be undermined by unexpected factors sometimes mask assumptions that the right to deploy a communications system is a right to be free from competition. The FCC and the courts have consistently rejected the argument that an entitlement from the government grants protection against future government entitlement of competitors.
The argument about investment incentives also ignores the fact that there are different types of investments. Wireless systems can be built through capital-intensive deployment of network infrastructure, which is then recouped through service revenues. Or they can be built through sales of equipment directly to end-users, who themselves pay the costs of building out the network. The former model is the traditional model for wireline and wireless operators; the latter is how WiFi has grown. As Benkler explains, both are legitimate capital formation mechanisms. They simply create different kinds of incentives.
With the universal entry privilege as a baseline, the question then becomes whether to employ exclusive property rights or expressly defined spectrum commons for everything else. Both should be used, and doubtless both will. Bearing in mind that this is not an absolute question, this Part examines the relative benefits of the two models through the lens of use rights I have developed.
Both commons and property advocates reject the assumption in the government licensing model that changes in the contours of use right for wireless equipment should be forbidden unless specifically authorized through a government regulatory process. Both allow new users or technologies to gain access to the spectrum, thereby improving efficiency and enhancing welfare.
The difference between commons and property regimes is how they manage these boundaries. Exclusive property rights give the rights holder a veto power over any new entry. Non-rights holders must either acquire the owner’s spectrum rights completely, or must negotiate a subdivision of rights such as a low-power underlay. Commons approaches allow the new entry by default. The traditional commons position bounds this entry right with technical standards; my supercommons proposal would expand the entry right using a liability system as the backstop.
I am joining this debate in progress. Proponents of exclusive property rights advance five primary arguments against the commons position:
• Markets are the best mechanism for allocating scarce resource
• Commons = regulation
• “Parks” will naturally emerge
• Scarcity and transaction costs favor property rights
• Commons may be accommodated through easements
The earlier analysis in this paper provides a framework to evaluate these claims. In every case, the attacks fail to destroy the commons position. At best, there are strong reasons for continuing to experiment with both exclusive rights and unlicensed systems, because we cannot be certain the commons approach will be superior. There is no basis, however, for concluding that exclusive rights are necessarily the best answer.
The property camp’s primary argument is that however we define the situation, market mechanisms are the best way to allocate resources. This viewpoint is perhaps best summarized by Peter Huber, who rebuts George Gilder’s claim that spectrum property rights are unnecessary by asserting that, “the one certain thing is that true wisdom in matters this complex does not emerge from centralized commission, nor even from visionary pundits. Wisdom emerges from markets.”
The problem with this argument is that a wireless commons is not an absence of markets. It is a different form of market from those prevalent in exclusively controlled spectrum. Commons are markets for equipment, not access. By making access costless, they actually facilitate more active and efficient markets for devices and ancillary services, much as the government’s construction of highways facilitates better markets for cars. There may be reasons to prefer a market for access over a market for equipment in some situations, such as when access is a scarce resource subject to a low-cost transaction regime. Or there may be situations in which a market for access generates an access price of zero. However, these are not arguments for a blanket preference toward exclusive rights. When pressed, the markets-über-alles argument devolves to either one about scarcity and transaction costs, or to a claim that private parks will naturally develop. I address both of these arguments below.
Furthermore, markets are not the right answer for every situation. There is no market for the right to breathe the air, for example. Air is so abundant that a market mechanism would be overkill, even if it would ultimately produce a zero price in most cases. And in situations where air wasn’t abundant, few would support letting people suffocate based purely on who can pay the most to breathe. The more spectrum looks like air, both in its abundance and its social value as a fundamental speech entitlement, the less markets for spectrum access make sense.
The problem is that we don’t know with certainty how abundant spectrum will be in the future. Huber is right that just because George Gilder thinks spectrum has infinite capacity doesn’t mean we should bet the future of spectrum on that prediction. The reason that, as Huber puts it, “wisdom emerges from markets,” is that markets can be good distributed signaling mechanisms under conditions of uncertainty. However, under some conditions, markets may not be the best means to perform that signaling function. Spectrum is such a case.
The critical resource for any form of social organization is information. The value of a good is a piece of information that is usually costly to obtain, so costly as to be practically impossible in most social situations. Markets obtain this information by delegating decision-making to the distributed mechanism of price signaling rather than using a central decision-maker. The market clearing price of a transaction is a distributed regulator of individual actions, giving participants better information than any central allocation of resources.
There can, however, be other distributed signaling mechanisms that play the same role as markets in allocating resources. Scientists studying the phenomenon of complexity have found numerous examples of physical and biological systems that self-organize without central control mechanisms or the formality of a price mechanism. Ant colonies, for example, find short paths to food sources through a purely random, uncontrolled process of exploration. They do so with an ingenious signaling mechanism: pheromone trails that are reinforced as more ants follow the efficient path. Scientists see no evidence of ant banking systems or currencies.
Intelligent wireless communications devices can fulfill the same function as the ants’ pheromone trails. By communicating with one another and adjusting their behavior dynamically, they can perform the distributed, bottom-up signaling function that markets achieve through the transaction mechanism. In fact they may do better. Markets involve time lags for price signals to propagate, they can be distorted through market failures or hold-outs, they can be misdirected toward maximizing transaction revenue rather than the value of communications, or they may signal efficiently but for the wrong variable. In these cases, non-market signaling mechanisms may prove superior. Software-defined radios and meshed networks in particular may be able to adapt more effectively to optimize spectral efficiency than transaction-based market mechanisms.
The argument comes back to Coase. Only this time, not the Coase of The Federal Communications Commission and The Problem of Social Cost, but the Coase of The Nature of the Firm. Coase’s insight there was that firms use non-market mechanisms internally for management because the transaction costs of market mechanisms would be too great. Similarly, some wireless communications systems may operate better in a non-market configuration.
As evidence that property-based markets are superior to commons, Hazlett points to the large sums companies were willing to pay at auction for exclusive rights to spectrum. The spectacular bids he references have since been revealed as bubble-era irrational enthusiasm. More fundamentally, focusing on the monetary value paid for spectrum under the current regime is a mistake. Both property and commons approaches would sharply reduce the value, on a per-unit basis, of the hypothetical spectrum resource. In so doing, however, they would increase the collective value of economic activity associated with spectrum use. Spectrum is expensive today because it is artificially scarce. A major goal of both property and commons advocates it to reduce that scarcity, either through transfers of un-used or under-used spectrum to those who will use it more, or by encouraging innovations that improve spectral efficiency. Increase supply without an equivalent increase in demand, and price will drop.
The cost of spectrum access is not a proper yardstick for comparing spectrum uses. PCS cellular operators (who paid billions for their spectrum rights) compete against successors of the local exchange carriers that received original cellular licenses for free, and against Nextel, which cobbled together taxi dispatch licenses through regulatory entrepreneurship. What spectrum costs is far less important than how spectrum is used. And legal structures that require an ex ante transaction to acquire transmission rights from a private owner make certain forms of spectrum use more difficult.
Next, property advocates attack spectrum commons as regulation in disguise. Hazlett and Benjamin, for example, point out that bands such as 2.4 and 5 GHz where WiFi operates were set aside and protected by the FCC for unlicensed use. They therefore equate commons with explicit government management of the spectrum, and list in great detail the ways that government is a poor decision-maker for this function compared to private actors.
There are several responses to this argument. Benkler offers two. First, exclusive property rights are hardly free from government involvement. As I note in Part II, the shape of these entitlements are not predetermined by nature. Whether or not they track existing FCC license terms, such rights are artificial constructs of the same government actors that define the boundaries of an unlicensed band. Second, unlicensed bands may require initial government actions to facilitate a commons, but after that they devolve power to individual users, who can employ equipment however they wish within the bounds of the commons. This contrasts with the centralized approaches in which spectrum owners determine the services available to their users. If freedom from regulation is defined as government non-intervention in individual decision-making, therefore, government action to create a commons is the less regulatory approach.
Another difficulty with the commons as regulation argument is that it presumes too stark a difference between the two regimes. As I have explained, commons and property are two configurations of use rights in wireless equipment. They are not radically different concepts. Under each arrangement a variety of rules are possible. Some involve more up-front government decision-making, and some require less. Even within a “public park” commons, government can impose many levels of requirements. The Unlicensed PCS bands have more stringent protocol mandates than the 2.4 GHz band, which in turn has more detailed requirements than some portions of the 5 GHz band.
Similarly, property rights can involve more or less interference protection, based on government decisions in defining those rights. In simple cases the rights holder and prospective entrants or adjacent rights holders may be able to negotiate around the initial rules. If the owner holds out, though, or if the proposed transaction requires re-aggregation of rights from too many sources (the “anticommons” problem), the inefficient government allocation will have decided the boundary between competing uses.
Finally, the equation of commons with regulation assumes expressly established commons bands. The underlay rights the FCC established through Part 15 and its ultra-wideband decision, for example, do not involve frequencies that the FCC has allocated for unlicensed operation instead of licensed uses. They involve an area below the exclusivity bounds of pre-existing licenses. Software-defined radios in the broadcast bands would similarly not represent public uses crowding out private uses, because no private uses exist in the guard bands.
The universal entry privilege I propose in Part III is fundamentally a deregulation of spectrum. It puts decisions about who can transmit in the hands of those who wish to transmit, and makes use of the private mechanism of common law courts to sort out disputes. The regime that prevents transmissions because government has issued an exclusive property right is arguably the regulatory one.
Property advocates, most notably Faulhaber and Farber, assert that if unlicensed commons are beneficial, they can and will appear within an over-arching property framework. Government can buy spectrum in a property regime and set it aside as a commons (the equivalent of public parks). Or private owners can open up their spectrum to any users who employ certain forms of equipment (private parks). The notion is that private parks would likely be created by equipment vendors, or on service providers who could charge a royalty on devices to cover the costs of acquiring the spectrum. These examples show that commons can exist within a property world. Exclusive property rights cannot, however, exist within a commons, because a commons means the absence of exclusivity. Therefore, property should be the baseline rule.
As with the transactions around default property rights in previous argument, parks sound nice in theory but are unlikely to happen in practice. As Benkler explains, there is a collective action problem for either the government or any private actor to purchase the necessary spectrum to establish a park. Many existing licensed spectrum bands, such as the PCS and Wireless Communications Service (WCS) bands offer significant flexibility to licensees. Yet no company has even proposed to purchase some of that spectrum at auction and make it available as a commons through an equipment royalty model, let alone actually doing so.
Unlicensed parks will not emerge within a property regime because of an information capture problem. A company bidding for spectrum property rights in a government-run auction or a private negotiation must determine whether the value it would receive from the spectrum would exceed the cost it will pay. For a centrally controlled, licensed service, benefits are straightforward to calculate. Service revenues are a function of the number of users. The discounted cash flow from those revenues can be matched against the net present value of the expenses involved, including the initial spectrum cost, the infrastructure build-out to provide service, and ongoing maintenance.
In an unlicensed environment, however, the discounted cash flow analysis is not so simple. Commons depend on competition among manufacturers to develop better equipment and sell it to users. As more users take advantage of the spectrum, the value of communications within it grows thanks to network effects. The spectrum owner cannot measure these values ahead of time, because it must calculate the price it will pay for exclusive control of the entire band, not whether there is enough of a market to recoup the expenses of selling a device. If the spectrum owner mandates that only certain equipment may be used, it is no different than a licensed service provider.
In other words, auctions require an ex ante decision about value of the spectrum, whereas value determinations in a commons are made real-time by many actors. The 2.4 GHz band is a perfect example. Before WiFi, this was known informally as the “junk” band, because it was so heavily congested with uses such as cordless phones, baby monitors, and microwave ovens. These “industrial, scientific, and medical” devices were thought to cause so much interference that the band would have little or no value if given exclusively to a licensee. Opening it up as a commons therefore had little opportunity cost for the FCC and significant potential benefits.
At first, nothing happened. Gradually, pioneering manufacturers such as Apple began selling wireless local area networking gear that took advantage of the 2.4 GHz band. The WiFi equipment market took off, reaching $2 billion in 2003 and still growing rapidly. Today, the 2.4 GHz band is the foundation of an extremely valuable and dynamic industry. Yet how could a prospective owner of an exclusive property right over the 2.4 GHz band predict that economic value ahead of time?
Public parks in spectrum are subject to the same failings. Government is no better able to capture ahead of time the value that could be generated from buying up spectrum and making it into a commons. The 2.4 GHz unlicensed band happened because there was effectively no cost for making it unlicensed, both because there was no charge for government to acquire the spectrum and because the pre-existing uses ruled out exclusive licensed operation. Public spectrum parks have the added problem that they require government to muster the resources to buy back spectrum rights. Given the vagaries of the budget process and the fact that spectrum in the age of auctions is thought of as a revenue generator for government, it is hard to see this happening.
Ultimately, the analogy with parks is flawed, and not just because it is an extension of the land metaphor I attacked in Part II. Parks compete with ownership of land, not with use of land. Central Park has different functions from the rest of midtown Manhattan – that’s the point of having a park. In the spectrum case, by contrast, the erstwhile parks would compete to provide the same kinds of services as privately controlled spectrum. The decision about whether to apply one model or the other is therefore not a fair one.
I have already touched on the next argument against wireless commons: scarcity is high and transaction costs are low. The FCC’s Spectrum Task Force correctly identified these as important variables to assess the proper legal regime. It failed, however, in claiming these measures called for exclusive property rights to be the dominant paradigm in all but the highest frequencies.
Contrary to the Spectrum Task Force’s blithe assertions, scarcity and transaction costs are not simple to determine. Both are recursive. How spectrum is used, in terms of architectures, services, and technologies, influences both scarcity and transaction costs. And these influence how spectrum is used. The branch of mathematics known as chaos theory demonstrates that seemingly simple equations can become hopelessly complex when they include recursive elements. Even when inputs follow deterministic laws to outputs, reconstructing the inputs may be impossible. Once again, the difficulty is informational. There is no way to keep track of the tangle of interacting variables, even knowing their starting points precisely.
Imagine, for example, a frequency band licensed to a mobile telephony operator that uses code division multiple access (CDMA), a common technology for second and third-generation cellular networks. An equipment manufacturer wishes to sell unlicensed ultra-wideband gear to operate across a frequency range that includes this band. Is spectrum in the band scarce? In general, the level of activity in the spectrum will depend on how densely the operator has built out its network of transmission towers and how many customers it has. For the UWB system, scarcity is even more complicated. UWB operates below the noise floor of background radiation because it employs such low-power signals. Spectrum below the noise floor is therefore not scarce, at least from the perspective of high-power systems above it, because these systems ignore radiation at that level. CDMA, however, dynamically alters power output of its transmitters to maximize utilization. In effect, it manages the noise floor. UWB systems, though they individually use infinitesimal power, can gradually increase the noise floor in aggregate by increasing the overall radiation levels in a band.
The point at which the CDMA and UWB transmissions meet is the scarcity boundary. Where this boundary appears is highly contingent on time, location, the usage of the CDMA system, and the usage of the UWB devices, not to mention other background radiators and natural thermal fluctuations. Given any set of initial conditions, there is an optimal solution for the two systems to share the spectrum, and an optimal transaction to reach this solution. There is, however, no simple way to find that optimal point, especially since the initial conditions are constantly changing. The value of a universal entry privilege for spectrum is that it would not absolutely preclude the UWB system from operating in the band absent government authorization or a transaction with the spectrum owner. It would give the UWB manufacturer leeway to experiment, with liability or technical standards as evolving backstops. It is a distributed solution to the intractably distributed problem of scarcity and transaction costs in this type of situation.
In general, the scarcity of spectrum is entirely dependent on the observer’s perspective. Today, the “ultra high frequency” (UHF) television bands are near the bottom of the spectrum chart, because so much higher-frequency spectrum that was once un-usable has been populated. FCC orders through the years are replete with categorical statements about the channel capacity available for broadcasting. None of these supposed limits have withstood the progress of technology. Furthermore, capacity is a function of system architecture. A broadcast network may deliver more bits to each user, but a cellular network carries more total bits and more independent conversations. A system employing inefficient receivers has less capacity than a system that swaps those out for better devices, even if the system architecture remains constant. Hence, each digital television channel can support up to six broadcast-quality streams in the same bandwidth that today carries just one.
The choice between exclusive property rights and commons affects the scarcity value of spectrum as well. At first blush, a commons would seem to increase scarcity. Unconstrained entry can lead to a tragedy of the commons, in which every user rationally consumes more than its efficient share of the resource. Yet as WiFi and the Internet itself demonstrate, not all commons end tragically. Well-designed technical or legal constraints can keep the system from collapsing.
In fact, commons can reduce scarcity by changing the incentives on participants. Operators of licensed systems have incentives to make receivers as cheap as possible within the constraints of their license authority and system design goals. Less robust devices cut down on costs, which means more profit to the bottom line for the service. Such devices, however, make the spectrum more scarce, because they are less able to tolerate other users outside the licensed system. In other words, the fragility of the receivers is a negative externality of the licensed system.
This externality is internalized in an unlicensed environment. Because there is no guarantee to be free from interference from other devices, equipment vendors have incentives to make their gear more robust. An unlicensed receiver less able to tolerate interference is likely to be less successful in the market, forcing the equipment manufacturer to factor this variable into its decision-making. Because a commons is open, equipment manufacturers are free to innovate and experiment with new approaches. They need not gain the permission of a regulator or owner to launch a better device into the marketplace.
Transaction costs similarly are higher than they appear in a property regime, and lower in a commons. Exclusive rights mean that every change requires a formal transaction. The potential entrant must find the rights owner, determine how much it is willing to pay, negotiate a transaction, and have that transaction recognized through the ancillary legal regime that accompanies every functioning property system. The rights-holder effectively holds a veto over any transaction, no matter how efficient.
Noam’s proposed regime, in which transactions would occur in a real-time market based on dynamic pricing, removes some of these barriers, but still requires payment and adds overhead to enforce those payments. The overhead of a transactional system can exceed the value of the transactions themselves. In the telephony world, a substantial portion of the price of a local call goes to the convoluted web of billing and tracking infrastructure that carriers operate to meter calls. In local area data networking, the winning protocol was Ethernet, which has no mechanism to prevent collisions among packets passing through the network. Competing protocols such as token ring prevented such “interference,” but at too great a cost.
The aggregate costs of spectrum transactions increase as the transactional environment becomes more dynamic. If frequency is the only variable, and frequency blocks are well-defined, transactions may be relatively cheap. This is the animating vision of the propertization literature going back to Coase. It was a fair approximation of reality in the past. Today, however, frequency blocks are no longer the only mechanism for exploiting spectrum. A property regime limited to frequency would preclude many uses that increase capacity and value of spectrum, such as wideband underlay. The trouble is that the more degrees of freedom there are, the more complex the transactional environment. Coase himself acknowledged that, “When the transfer of rights has to come about as a result of market transactions carried out between large numbers of people or organizations acting jointly, the process of negotiation may be so difficult and time-consuming as to make such transfers a practical impossibility.”
Property advocates themselves inadvertently confirm this point. Kwerel and Williams, in advocating a “big bang” auction mechanism for propertizing the spectrum, describe several aspects that would be efficient to incorporate into the auctions, such as combinatorial bidding, which are too complex for current software. And this is just for initial reassignment of existing frequency blocks! The better the property system at taking into account the alternative uses of spectrum, the more potential transactions, and hence the greater the transaction costs. This is also a reason property advocates propose that initial property rights track existing FCC licenses, even though those licenses may have inefficient or outdated boundaries.
The De Vany and Minasian proposals for output-delineated spectrum rights were efforts to develop better rights configurations, but as a result they create new problems and complexities. Property advocates today cite these works religiously, yet despite the wealth of recent spectrum scholarship, there have been no serious attempts to update their decades-old proposals. The property camp seems to have concluded that moving to a property systems for spectrum as quickly as possible is more important than working through the thorny matter of getting that system right. Once the market is in charge, they assume, the magic process of Coasian bargaining will right all wrongs. There is, however, no guarantee this will happen, for reasons Coase himself identified.
Transaction costs are a mirror of interference. The more complicated and uncertain a concept interference is, the more difficult it is to determine ahead of time the efficient configuration of duties to prevent interference. Prior to Coase, interference was seen as an absolute barrier to flexibility in assignment of wireless transmission rights. Government had to manage the spectrum and limit private actors, because the alternate was chaos.
One of Coase’s major contributions was his recognition that interference boundaries could be set through property rights rather than government management, because each rights-holder would have incentives to both exploit and police its spectrum. Coase’s proposed solution, however, was based on the contemporary view of interference when he proposed it. Interference was still thought to necessitate exclusivity in spectrum. The less we think of interference as a rigid and high barrier, and the more we see it as a phenomenon technology is gradually conquering, the more the transaction cost ledger favors commons.
Consider what might have happened had the FCC adopted Coase and Herzel’s proposed spectrum property regime in the 1950s. In those days, wireless services offered to the public meant broadcasting. Exclusive property rights in spectrum would therefore have been broadcast rights. The interference boundaries would have assumed central transmission towers with passive receivers, and the systems deployed would have in fact looked like that. Owners could happily buy and sell spectrum in this broadcastopia.
Ten or twenty years later, let us assume, AT&T decides the cellular telephony technology it has developed is ready for commercial deployment. To introduce cellular service, AT&T must purchase spectrum from an existing owner, who is by definition a broadcaster. What AT&T needs, however, is not a right to build a broadcast tower in one geographic location, but the ability to erect may two-way cellular towers across the country. It is not clear that AT&T could acquire those rights at any price. Mobile telephony, which today generates more revenue than broadcast television, would have no place in broadcastopia. The property system would be efficient on its own terms, but transaction costs would mean it was heavily sub-optimal in the aggregate social welfare terms that matter.
What about low-power unlicensed devices, such as those permitted under the FCC’s Part 15? Broadcasters’ property rights would give them exclusive control over “their” spectrum, so the Part 15 devices would be prohibited. Makers of low-power devices could negotiate to pay a spectrum owner for an “easement,” but those direct and transaction costs would raise the price of their equipment. It would make many such systems impractical, because they couldn’t transmit on the same frequencies everywhere in the country. The problem gets worse with technologies such as ultra-wideband. Aggregating easements from the large number of rights-holders involved would be a practical impossibility. Scholars have labeled this scenario the anti-commons.
Perhaps the courts would read a Part 15 underlay into the pre-existing property rights. The analogy would be the decision that, even though land rights theoretically extend forever above and below the ground, airplanes need not negotiate overflight rights with individual land-owners. This is an easy case because the benefits of allowing aviation were so clear, and the interference with the land-owners’ notional property rights so theoretical. We cannot simply assume that courts will reach the right result in the spectrum context. When the default rule is private ownership and the new underlay services isn’t yet commercially available (which it couldn’t be until it is authorized to use the spectrum), courts are likely to err on the side of the incumbent property owners.
The transaction costs of dispute resolution and monitoring are similarly complicated to calculate. Faulhaber and Farber assert that such costs are higher in a commons because there will be more disputes. I as have explained, however, this would not necessarily be the case. Whatever the legal regime, administrative transaction costs will depend on how effective and settled the dispute resolution regime is. Costs will be higher at first under either system, as parties explore the boundaries and adjudicators develop precedent. Also, as I detail above, a commons regime could use the same administrative mechanisms, the courts, as a property regime.
The final counter to the commons position, Faulhaber and Farber’s proposed non-interfering easement, is of a different character from the others. The easement is an insurance policy. It concedes that commons may be a viable approach, and therefore mandates as commons alongside the property rights allocation. Faulhaber and Farber present the easement as a simple adjustment to the property regime that achieves the best of both worlds. On the contrary, a non-interfering easement is a radical limitation of the core exclusive rights in a property system. It means that rights are always exclusive... except when they are not. And they are not wherever a parallel transmission is considered non-interfering.
The easement is, in effect, the universal entry privilege I propose above. Far from being a reason to choose property as the over-arching legal paradigm, it is a reason to scale back the bundle of rights granted in those situations where the property regime applies.
Faulhaber and Farber’s choice of wording is unfortunate. An easement is a legal concept plucked from the law of real property. Importing it into spectrum policy reinforces the inaccurate analogy between spectrum and land. Calling the easement “non-interfering” is also problematic, because no transmission can be proven to avoid interfering with all other transmissions. “Non-interfering easement” connotes something akin to a land owner’s deed specifying that the public can drive on a road passing though his or her property. Even on their home turf of property law, easements are not so simple, as any first-year law student can attest.
The easement concept implies that the spectrum owner is the primary rights-holder, with the easement as an intrusion that it must tolerate if it suffers no harm. If the burden of proof is solely on the potential user of the easement to be non-interfering, the easement may not be viable. The property owner has every incentive to put up artificial limits on the easement or claim interference. In the CDMA vs. UWB example I outlined above, the cellular operator could crowd out the UWB systems at any time by making its equipment more sensitive. The non-interference restriction is effectively a choice to protect the cellular network, even if that is the economically inefficient choice. Unless, of course, the easement boundary is subject to a legal regime such as I propose in Part III, at which point the easement looks much larger than Faulhaber and Farber envision.
The easement as Faulhaber and Farber present it effectively chooses the surgeon over the confectioner in Coase’s illustration. That may be the right outcome in some or most cases. To assume that it is, though, is to make a leap not supported by the evidence. The problem here is similar to the one with dedicated unlicensed parks amid owned spectrum, something Faulhaber and Farber also bring up. Neither government nor the spectrum owner knows ahead of time how the easement will be used. So there is no way to calculate the benefit or optimal boundaries of the easement.
One alternative is to make the interference protection implied by the easement mutual. Easement users could have some rights against the spectrum owner, just as it has against them. The effect would be similar to the “Part 16” rules that Apple proposed for the U-NII band, under which unlicensed devices would collectively enjoy the same protection against out-of-band interferers as licensed systems. The FCC rejected Apple’s idea. Benkler argues such a rule should be adopted broadly for unlicensed devices. Merely doing so, however does not indicate where the boundary should be drawn when rights overlap. The equivalent statement that I have the right two swing my arm until it connects with the bridge of your nose is the beginning of the legal analysis, not the end.
Faulhaber and Farber properly define the easement not just as a low-power underlay, but as a general privilege to transmit where that transmission would not excessively burden other systems. An adaptive or cognitive radio that exploited temporary holes in owned spectrum would be subject to such an easement, for example. If in fact a majority of spectrum frequencies are unused any given time, as the Shared Spectrum survey found, this portion of easement would in fact be larger than the entire scope of exclusive property rights.
If the easement is a serious proposal, not a hypothetical bone tossed toward the commons, there must be a viable mechanism to determine its boundaries. Benkler, for example, proposes an expanded version of the Faulhaber and Farber easement, which he calls underlay and interweaving rights. Benkler’s easement would have two constraints. Systems would have to use power low enough to not “appreciably affect the information flow” of incumbent devices deployed as of that date. They would also be required to incorporate automatic sensing and of licensed systems and mechanisms to vacate the band when such transmissions appear.
As I detail above, we can go further. The unlicensed system is not a necessarily subordinate use; it is a competing projection of property rights. The incumbent user of the band has legitimate reliance and expectation interests that must be weighed against any claim that a competing unlicensed use is more efficient. These interests, however, are not absolute, especially at margins. When technological limitations made sharing of spectrum frequencies difficult, a blanket rule privileging incumbents made some sense. As that constraint is more and more fully lifted, however, exclusive rights become a greater and greater drag on efficiency.
Ultimately, commons and property come together as equals. They are different shades in the rainbow of possible use rights that can be accorded to wireless communications equipment. Neither should exist at the sufferance of the other, though participants in the market should have some ability to choose among them. The option of buying exclusive rights or participating in a commons should be a choice spectrum users can make, just as they must choose what service to deliver and what technology to employ. The presence of more substitutes is what will make spectrum effectively less scarce, even as usage increases. The more ways there are to avoid the messy process of dispute resolution by simply routing around conflicts, the more likely participants will choose that route.
The policy challenge today is how to achieve such a result, given the many deep uncertainties I have outlined. If we knew all the situations in which property rights or commons were more efficient, based on which set of boundary conditions, we could simply mandate that world. But we do not know that, nor can we. And any solution that is correct today will be wrong tomorrow as usage and technology change. Therefore, caution and experimentation must be part of the agenda for moving forward. There is nothing wrong with trying out property rights in spectrum, but there is everything wrong with acting as though property rights are the only solution, everywhere and for the end of time.
The legal states for wireless communications can be defined in terms of two variables: whether entry is open to any conforming device or controlled by a rights-holder, and whether the boundaries of the state are fixed ahead of time by law or evolve through the interactions of participants.
Table 2: Possible legal arrangements
Before the emergence of the commons critique, the spectrum debate focused almost entirely on the top left quadrant of the table. Both government licensing and property rights involve exclusive control of a defined allocation. That is why moving from licensing to property seems so temptingly simple. Government simply must transmute heavily encumbered licenses into permanent fee simple ownership of the same assets. The idea that there might be other ways to slice the spectrum pie, using techniques such as wideband underlay, occasionally cropped up. De Vany, for example, argued in 1996 that property rights would allow owners to share their spectrum with what he called “broadband broadcasters.” The boundary between the high-power and low-power uses would not be fixed in the allocation of property rights; it would evolve through negotiation. This was at most a side discussion, though, not the primary argument for the property regime.
The commons critique revealed that there was another dimension that neither government licensing nor property rights took into account. Smart digital devices backed by technical standards can coexist without exclusivity, as WiFi and UWB demonstrate. Open entry encourages different business models, with equipment manufacturers taking the place of centralized infrastructure builders. It also allows for uses that may better achieve the normative goals of democracy, diversity, and autonomy.
My goal in this paper has been to fill in the final quadrant. The supercommons has both open entry and open boundaries. It is the white space around the other forms that have been the subject of the debate. Transmissions that are neither exclusive nor confined to bounded spaces have always been defined as impermissible interference. In many cases, they are. As technology increases the scope of possibilities for wireless communication, however, the spectrum white space grows to the point at which it can be a viable platform in its own right. Even if the supercommons is not exploited, its existence helps sharpen our understanding of the scarcity and transaction cost tradeoffs among the property and commons forms in the various possible manifestations.
Going forward, we should proceed along two tracks. Fixed-boundary frequency bands should gradually be converted to both exclusive property rights and commons, with mechanisms for reversibility at some future point. With over ninety percent of the spectrum below 3 GHz still subject to service-specific licenses, even a relatively small shift could have significant economic consequences.
A commons may be more difficult to reverse than a time-delimited property right, but it is not impossible to move from a commons to something else. The FCC has in the past reassigned private microwave users, who share licensed spectrum under a regime that has elements of a commons. The more devices there are operating in the commons, the harder it will be to declare those devices illegal and turn the band over to an exclusive owner. However, more devices mean the commons is more valuable. If a commons has failed, either because it has few users or because it has descended into a tragedy of over-use, reassigning it to another use would not cause much of a loss for users. Though there would be a cost and a time lag, there is no reason the FCC could not order certain unlicensed devices shut down the way it plans to order virtually every household in America to upgrade their television sets when analog broadcasting ceases.
Simultaneously with its expansion of bounded property and commons uses, the government should begin to open up the supercommons. At first, stringent limits including sensing and avoidance capability, strict liability for interference, insurance or bonding requirements, and remote shut-off capability could be required. As Benkler proposes, “anything goes” spectrum could also be set aside as an experimental space. The limits on the supercommons could be removed over time, either on a set schedule pursuant to defined milestones or based on petitions from potential equipment or service providers. The other option is to start with a supercommons that has less severe backstops, but to do so with more constraints on how devices can operate. For example, the newly minted property rights or commons could be tied to a strong supercommons underlay or interweaving privilege. This would give the bounded users notice that they may have to share with others at the margins, and would net this restriction on their rights against the benefits they receive at the same time.
The goal of spectrum policy is to maximize national welfare derived from wireless communications. Welfare involves some measure of efficiency, which means optimizing the amount and economic value of communications that occur. The capacity of wireless systems is a function of their architecture and equipment, which are themselves shaped by legal rules. Welfare also involves values such as autonomy, diversity, and democracy, which are not always reducible to dollar amounts.
The actions that are properly the subject of spectrum policy are actions of individuals. The issue is whether users can operate certain kinds of devices. Those devices may be built to specifications defined by service providers, technical standards bodies, or hobbyist groups. The device in question may be a five-hundred-foot broadcast tower or a mobile phone that fits in a shirt pocket. Individual use of those devices generates collective value in the form of communications capacity and revenue from services or equipment. As a general matter, the more individuals who can participate in wireless communications, and the more they can do with their devices, the more value will be produced.
Coase himself hinted at such a user-centric perspective. After pointing out that spectrum is no more a resource susceptible to allocation than sound or light waves, he observed that, “[t]o handle the problems arising because one person’s use of a sound or light wave may have effects upon others, we establish the right which people have to make sounds which others may hear or do things which others may see.” These are not exclusive or possessory property rights. They are individual privileges to act and rights to be free from certain invasions by others. Their boundaries are determined dynamically by law and custom.
At any point in time, there exists a theoretically optimal arrangement of wireless devices. This would be a situation in which the marginal value of any change would be less than the marginal cost. It is impossible to determine ahead of time what this arrangement is. It will vary over time depending on technical capabilities, existing systems, and demand. It may not be the arrangement that passes the most bits, since some bits are more valuable than others. Nonetheless, the optimal arrangement would generally be one that utilizes the spectrum as fully and efficiently as possible.
The status quo is far from the optimal arrangement, because it leaves large portions of the spectrum fallow. Property rights in spectrum would be an improvement, but only to a point. Their inherent exclusivity would systematically prevent many possible avenues from being explored. Unlicensed bands open up further opportunities, but remain tied to frequencies. The supercommons around traditional communications techniques opens up the final frontier of wireless. By shifting from regimes that require explicit permission for new transmission techniques to one that uses dynamic boundary-setting mechanisms, the supercommons would over time allow the wireless communications ecosystem to move toward the theoretical ideal.
The outcome of the spectrum debate is extremely important. In the years to come, the uses and importance of wireless communication will only grow. As radios become cheaper and more flexible, they will be incorporated into many products and processes that today are unconnected. More than eight billion microprocessors are shipped ever year, a number exceeding the population of the earth, yet less than two percent of them are networked. And though virtually every household in America has at least one television, only tens of broadcast stations are available through the air and a couple hundred through cable or satellite.
A vast opportunity lies between those hundreds of channels and hundreds of millions of users, as well as between those millions of networked devices and billions of microprocessors. A new world of communication awaits. Its coastline is only now being mapped; the extent of the hidden continent and its territorial riches remain to be discovered. Now is not the time to fear mythical dragons on the electromagnetic high seas.
* Founder, Supernova Group LLC. Former Counsel for New Technology Policy, FCC and former Editor, Release 1.0. Thanks to Yochai Benkler, Ellen Goodman, and Phil Weiser for comments on earlier drafts. Feedback welcome to firstname.lastname@example.org.
 Apologies to Karl Marx and Friedrich Engels. Actually, unlike communism, the commons position is neither anti-property nor anti-markets. See infra Part IV(A).
 Eli Noam, Spectrum Auctions: Yesterday’s Heresy, Today’s Orthodoxy, Tomorrow’s Anachronism. Taking the Next Step to Open Spectrum Access, 41 J. Law & Econ. 765 (1998) [hereinafter Noam, Yesterday’s Heresy].
 Gerald R. Faulhaber & David Farber, Spectrum Management: Property Rights, Markets, and the Commons (working paper), at http://rider.wharton.upenn.edu/~faulhabe/SPECTRUM_MANAGEMENTv51.pdf; ; Eli Noam, The Fourth Way for Spectrum, FT.com, May 29, 2003, at http://news.ft.com/comment/columnists/neweconomy; Ellen P. Goodman, Spectrum Rights in the Telecosm to Come, __ San Diego L.Rev. ___ (forthcoming February 2004).
 See, e.g., Yochai Benkler, Some Economics of Wireless Communications, 16 Harv. J.L. & Tech. 25, 63 (2002) [hereinafter Benkler, Some Economics].
 Recent property and commons scholarship acknowledges the possibility of non-frequency-based modalities, but fails to grapple with their implications. See infra notes 220, 224, 229.
 Richard Feynman, There’s Plenty of Room at the Bottom, Engineering and Science, Feb. 1960, available at http://www.zyvex.com/nanotech/feynman.html. Feynman was referring to the potential for what is now called nanotechnology: machines that operated at the molecular scale.
 See Michael Chartier, Enclosing the Commons: A Real Estate Approach to Spectrum Rights, unpublished paper presented at the AEI-Brookings Joint Center Conference, “Practical Steps to Spectrum Markets,” Nov. 9, 2001 (on file with author) at 5 (noting that at the International Radio Telegraph Convention of 1927, all services resided below 1715 kHz and the extreme range of “experiment” possibility ended at 60,000 kHz, whereas today the FCC’s table of allocations ends at 300,000,000 kHz.).
 This is not the only enduring aspect of wireless communication. An 80-year-old AM radio can still be a useful device today. No other consumer electronics device has anywhere near that degree of longevity.
 Cf. Ithiel de Sola Pool, Technologies of Freedom (1984).
 See Erik Barnouw, A History of Broadcasting in the United States, Volume 1: A Tower in Babel 12 (1966)
 “Free-space optics” communications systems can now be built using laser beams that operate at visible-light frequencies above the radio spectrum. Vendors such as Terabeam and AirFiber sell free-space optics equipment for high-speed data links comparable to systems using radio frequencies. See Doug Allen, The Second Coming Of Free Space Optics: New Technology Called Free Space Optics May Be the Answer to Bandwidth Bottlenecks, Network Magazine, March2001, at 55. However, since the FCC’s authority extends only to “communication by wire or radio,” the free-space optics systems are outside of FCC jurisdiction. 47 U.S.C. § 152(a).
 Officially, FCC rules defines the radio spectrum or “Hertzian waves” as the frequencies between 9 kilohertz and 300 gigahertz. 47 C.F.R. §2.1 (2002). A hertz is one cycle per second. A kilohertz (khz) is a thousand hertz; a Megahertz (Mhz) is a million; and a Gigahertz (GHz) is a trillion, or a thousand Megahertz.
 Ironically, this archaic method of “carrierless” wideband wireless transmission has now reappeared in the form of ultra-wideband (UWB), with precisely the opposite result. Where spark gap transmitters prevent any other radios from operating, UWB systems operate at such low power that they can “underlay” virtually any other transmission without noticeable interference. See infra TAN 118.
 Marconi received British patent no. 7,777 for the use of “resonant tuning” to divide radio communications by frequency.
 See Paul Baran, Is the UHF Frequency Shortage a Self Made Problem? Paper delivered to the Marconi Centennial Symposium (June 23, 1995) (describing the history of frequency division); Andrew J. Viterbi, The History of Multiple Access and the Future of Multiple Services Through Wireless Communications, available at http://www.gte.com/Showcase/Cdma/Feature/editorial.html.
 See Barnouw, supra note 10.
 See Ronald H. Coase, The Federal Communications Commission, 2 J. L. & Econ. 1, 1 (1959) [hereinafter Coase, The FCC].
 See Hoover v. Intercity Radio Co., 286 F. 1003 (D.C. Cir. 1923); United States v. Zenith Radio Corp., 12 F.2d 614 (1926).
 The 1927 Act was expanded in the 1934 Communications Act, which folded the Federal Radio Commission into the Federal Communications Commission (FCC) that endures to this day. Communications Act of 1934, 48 Stat. 1064 (1934) (codified as amended at 47 U.S.C. §§ 151 et seq. (2002)).
 See National Broadcasting Co. v. United States, 319 U.S. 190 (1943).
 For an overview of the current spectrum management process, see Charles L. Jackson, Use and Management of the Spectrum Resource, in Paula L. Newberg, ed., New Directions in Telecommunications Policy (1989) at 247-71.
 47 U.S.C. § 303 (2002). The FCC has responsibility for privately used spectrum. The National Telecommunications and Information Administration (NTIA) of the Department of Commerce oversees government spectrum usage, in conjunction with the agencies (including the Department of Defense, Federal Aviation Administration, and NASA) that actually use the spectrum. See 47 U.S.C. § 305(a)(2002). The government either shares or controls two-thirds of the most easily used spectrum (in the range between 30 khz and 3 Ghz), and outright controls a quarter of it.
 47 C.F.R. § 2.1(c) (2002).
 See Leo Herzel, Comment, Public Interest and the Market in Color Television Regulation, 18 U. Chi. L. Rev. 802 (1951).
 See Coase, The FCC, supra note 17.
 See Arthur S. De Vany et al., A Property System for Market Allocation of the Electromagnetic Spectrum: A Legal-Economic-Engineering Study, 21 Stan. L. Rev. 1499 (1969); Jora Minasian, Property Rights in Radiation: An Alternative Approach to Radio Frequency Allocation, 18 J.L. & ECON. 221 (1975); Harvey J. Levin, The Invisible Resource: Use And Regulation Of The Radio Spectrum (1971); Milton Mueller, Property Rights in Radio Communication: The Key to the Reform of Telecommunications Regulation, Cato Policy Analysis No. 11 (June 3, 1982).
 It is also part of a more recent deregulatory movement in regulated industries. See Joseph D. Kearney & Thomas W. Merrill, The Great Transformation of Regulated Industries, 98 Columbia L. Rev. 1324 (1998).
 See Thomas W. Hazlett, The Wireless Craze, the Unlimited Bandwidth Myth, the Spectrum Auction Faux Pas, and the Punchline to Ronald Coase’s “Big Joke”: An Essay on Airwave Allocation Policy, 14 Harv. J. Law & Tech. 335 (2001) [hereinafter Hazlett, Wireless Craze]; Lawrence J. White, “Propertyzing” the Electromagnetic Spectrum: Why It’s Important, and How to Begin, 9 Media L. & Pol’y 19 (2000). See also Pablo T. Spiller & Carlo Cardilli, Towards a Property Rights Approach to Communications Spectrum, 16 Yale J. Reg. 53 (1999).
 See Promoting Efficient Use of Spectrum Through Elimination of Barriers to the Development of Secondary Markets, Policy Statement, 15 F.C.C.R. 24178, 24181 (2000) (“[T]he best way to realize the maximum benefits from the spectrum is to permit and promote the operation of market forces in determining how spectrum is used”).
 See Yochai Benkler, Overcoming Agoraphobia: Building the Commons of the Digitally Networked Environment, 11 Harv. J.L. & Tech. 287 (1998) [hereinafter Benkler, Overcoming Agoraphobia]; Lawrence Lessig, The Future of Ideas: The Fate of the Commons in a Connected World (2001); Nobuo Ikeda, The Spectrum as Commons: Digital Wireless Technologies and the Radio Policy, RIETI Discussion Paper Series 02-E-002, March 2002; Kevin Werbach, Open Spectrum: The Paradise of the Commons, Release 1.0, November 2001. The commons position is also sometimes referred to as “open spectrum” or “open wireless.” Cf. Scott Woolley, Dead Air, Forbes, Nov., 25, 2002 (“It is the bitter irony of America’s skies: Open airwaves are everywhere, yet the people desperate to use them cannot.”) I use the term commons here because it is widely used to describe the argument, and because it emphasizes the relationship of this viewpoint to a broader critique of current legal orthodoxies related to the digital world. See infra note 46
 Frequency is a physical property; frequency division is a technical design choice.
 See Claude E. Shannon, A Mathematical Theory of Communications, 27 Bell System Tech. J. 379-423, 623-56 (1948) , available at http://cm.bell-labs.com/cm/ms/what/shannonday/shannon1948.pdf.
 Lessig and Benkler draw a parallel between the possibility of an open-entry commons at the “physical layer” of networks with the commons that the public domain represents vis a vis copyright at the “content layer.” See Lessig, The Future of Ideas, supra note 30; Yochai Benkler, From Consumers to Users: Shifting the Deeper Structures of Regulation Toward Sustainable Commons and User Access, 52 Fed. Comm. L.J. 561, 568-570 (2000) [hereinafter Benkler, Consumers to Users]. For a further explication of the layered model of communications, see Kevin Werbach, A Layered Model for Internet Policy, 1 J. Telecom. & High Tech. L. 37 (2002),
 See Revision of Part 15 of the Rules Regarding the Operation of Radio Frequency Devices Without an Individual License, 4 F.C.C.R. 3493 (1989) (establishing the spread-spectrum bands at 900 MHz, 2.4 GHz, and 5 GHz); Amendment of the Commission’s Rules to Provide for Operation of Unlicensed NII Devices in the 5 GHz Frequency Range, 12 F.C.C.R. 1576 (1997) (establishing the Unlicensed National Information Infrastructure Bands) [hereinafter U-NII order].
 The FCC originally designated as unlicensed bands that were so full of other uses, including microwave ovens, medical equipment, and garage door openers, as to be unsuitable for licensed operation. Unlicensed bands are sometimes described as “licensed by rule” because they are in fact subject to FCC licenses like any other band. See, e.g., See, e.g., “Commission Seeks Public Comment on Spectrum Policy Task Force Report,” ET Docket No. 02-135, Comments of Microsoft at n.1 (Jan. 27, 2003). The difference is that the license is extended by rule to any device meeting the FCC’s technical criteria, rather than those approved by an individual licensee.
 See George Gilder, The New Rule of the Wireless, Forbes ASAP, March 29, 1993; Paul Baran, Visions of the [sic] 21st Century Communications: Is the Shortage of Radio Spectrum for Broadband Networks of the Future a Self Made Problem? Keynote Talk Transcript, 8th Annual Conference on Next Generation Networks Washington, DC (Nov. 9, 1994), at http://www.eff.org/GII_NII/Wireless_cellular_radio/false_scarcity_baran_cngn94.transcript. Baran is best known for developing the technique of packet-switching, on which the Internet is built. Gilder’s support for a wireless commons is notable given his anti-government bent in other areas of communications policy. Gilder, in fact, is on most topics a leading advocate of the deregulatory communications policies the commons camp opposes. See George Gilder, Telecosm: How Infinite Bandwidth Will Revolutionize Our World (2000). Gilder’s advocacy on technical grounds of unlicensed wireless systems rather than exclusive rights hints at the deep connections between the supposedly opposed property and commons positions in the spectrum debate. See infra TAN 179.
 See Benkler, Overcoming Agoraphobia, supra note 30; Benkler, Some Economics, supra note 4.
 See Noam, Yesterday’s Heresy, supra note 2. Noam’s 1998 paper expanded on his earlier work endorsing what he calls “open access” for wireless. See Eli Noam, Taking the Next Step Beyond Spectrum Auctions: Open Spectrum Access, 33 IEEE Comm. Mag. 66 (1995). Noam argues not for free access, but for open entry subject to a variable fee. He acknowledges that his proposal may not be practical today, but that when it is, it will achieve the best of both the commons and property worlds.
 See Benkler, Overcoming Agoraphobia, supra note 30.
 See id.
 See Werbach, Paradise, supra note 30; Kevin Werbach, Open Spectrum: The New Wireless Paradigm, New America Foundation Spectrum Series Working Paper No. 6 (October 2002); Spectrum Policy Task Force Seeks Public Comment on Issues Related to Commission’s Spectrum Policies, DA 02-1311, ET Docket No. 02-135, Comments of Kevin Werbach (July 8, 2002).
 See Lessig, The Future of Ideas, supra note 30.
 See FCC Spectrum Task Force Public Notice, ET Docket No. 01-135, Comments of David P. Reed (July 8, 2002); David Weinberger, The Myth of Interference, Salon, Mar. 12, 2003, at http://www.salon.com/tech/feature/2003/03/12/spectrum/print.html (explaining Reed’s ideas).
 See Stuart Buck, Replacing Spectrum Auctions with a Spectrum Commons, 2002 Stan. Tech. L. Rev. 2 (2002).
 Though economists developed the property position and proponents of the commons emphasize technology, it is too simplistic to cast this as a debate between the virtues of economics and engineering. Commons advocates are perfectly capable of framing their arguments in economic terms. See Benkler, Some Economics, supra note 4, at 49-56; Noam, Yesterday’s Heresy, supra note 2. The normative prong of the commons attack offers an independent justification even if the economic debate is stalemated.
 See Benkler, Overcoming Agoraphobia, supra note 30; Yochai Benkler, Siren Songs and Amish Children: Autonomy, Information, and Law, 76 N.Y.U.L. Rev 23 (2001); Lessig, The Future of Ideas, supra note 42. Like the economists who developed the property rights proposal for spectrum, the leading academic supporters of a wireless commons have a larger program in mind. They envision communications, media, and technology industries that respect the value of commons to promote innovation, and allow greater freedom and control to individuals. See Benkler, Consumers to Users, supra note 33.
 In this case, though, the market is for end-user equipment rather than tradable spectrum rights.
 See Noam, Yesterday’s Heresy, supra note 2.
 Thomas W. Hazlett, Spectrum Flash Dance: Eli Noam’s Proposal for “Open Access” to Radio Waves, 41 J.L. & Econ. 805 (1998); Hazlett, Wireless Craze, supra note 28.
 When Coase testified before the FCC in 1959, one FCC Commissioner began by asking whether his proposal was just a big joke. See Hazlett, Wireless Craze, supra note 28, at 5.
 See, e.g., Reed E. Hundt, Spectrum Policy and Auctions: What’s Right, What’s Left, Remarks to Citizens for a Sound Economy (June 18, 1997), at http://www.fcc.gov/Speeches/Hundt/spreh734.html.
 See Omnibus Budget Reconciliation Act of 1993. Pub. L. 103-66, 107 Stat. 312 (1993) (codified at 47 U.S.C. § 309).
 See FCC Opens First Ever Airwave Auction, News Release (July 25, 1995).
 See Hundt, supra note 51; Gregory L. Rosston & Jeffrey S. Steinberg, Using Market- Based Spectrum Policy to Promote the Public Interest, 50 Fed. Com. L.J. 87 (1997); Evan Kwerel & John Williams, A Proposal for a Rapid Transition to Market Allocation of Spectrum, FCC Office of Plans and Policy Working Paper No. 38, November 2002.
 See FCC Issues Guiding Principles for Spectrum Management, News Release (Nov. 18, 1999), at http://www.fcc.gov/Bureaus/Engineering_Technology/News_Releases/1999/nret9007.html.
 See FCC Takes Steps to Make More Spectrum Available Through the Development of Secondary Markets, News Release, at http://www.conformity-update.com/fcc-spectrum-001201.htm.
 See Federal Communications Commission, Spectrum Policy Task Force Report, ET Docket No. 02-135 (rel. Nov. 15, 2002) [hereinafter Spectrum Task Force Report].
 See id. at 35-37.
 See id. at 38-41.
 See Revision of Parts 2 and 15 of the Commission’s Rules to Permit Unlicensed National Information Infrastructure (U-NII) Devices in the 5 GHz Band, ET Docket No. 03-122 (rel. June 4, 2003). The FCC also released a staff working paper reviewing the benefits of unlicensed spectrum. See Kenneth R. Carter, Ahmed Lahjouji & Neil McNeal, Unlicensed & Unshackled: A Joint OSP-OET White Paper on Unlicensed Devices and their Regulatory Issues, FCC Office of Strategic Policy Working Paper Series No. 39 (May 2003), available at http://hraunfoss.fcc.gov/edocs_public/attachmatch/DOC-234741A1.pdf.
 See Additional Spectrum for Unlicensed devices Below 900 MHz and in the 3 GHz Band, Notice of Inquiry, 17 F.C.C.R. 25632 (2002).
 WiFi (wireless fidelity) is a family of protocols for wireless local area networks issued by the Institute of Electrical and Electronic Engineers (IEEE). See Wi-Fi: It’s Fast, it’s Here – and it Works, Bus. Week, Apr. 3, 2002; infra note 303.
 See Peter W. Huber, Telecom Undone - A Cautionary Tale, Commentary, January, 2003, at 34-40.
 The FCC’s rules governing “unintentional” and “incidental” radiators technically constrain property rights in other kinds of devices, such as microwave ovens and the microprocessors powering personal computers. 47 C.F.R. §§15.13, 15.20. Since these devices are not intended for communication, they do not fall within this analysis.
 The exception is a commons with a “Part 16” duty imposed on devices that are not part of the commons. Apple proposed such a mechanism for the U-NII band, but the FCC rejected it as unnecessary. See U-NII Order, supra note 34, at ¶ 91. Benkler advocates such a rule. See Benkler, Overcoming Agoraphobia, supra note 30; Benkler, Some Economics, supra note 4. Part 16 is a further refinement of the commons model in which the absence of corollary duties extends only to devices of the same class as the transmitting device.
 Saying that the right to exclude is fundamental to the grant of rights begs the question. See Loretto v. Teleprompter Manhattan CATV Corp., 458 U.S. 419, 435 (1982) (“The power to exclude has traditionally been considered one of the most treasured strands in an owner’s bundle of property rights.”) Why should a right to transmit, or a possessory right in real property, include a right to exclude as part of the bundle? The answer is that the overall right becomes meaningless if others interfere with the signal or trample the land. This is just another way of saying that rights are inherently social, and dependent on assumptions about how third parties can and will behave.
 Cf. White, supra note 28, at 7 (“From the early 1900s to the present day the basic problem of using the spectrum has been seen as that of interference.”)
 See Watch This Airspace, The Economist, June 20, 2002 (describing four disruptive technologies that “could shake up the wireless world.”)
 Einstein no cat cite.
 The desire to treat spectrum as a physical resource parallels the tendency to see cyberspace as a place. Dan Hunter has explored the significance of the “cyberspace as place” metaphor in detail. See Dan Hunter, Cyberspace as Place and the Tragedy of the Digital Anticommons, 91 Calif. L. Rev. 439 (2003). Drawing of the work of cognitive scientists, especially George Lakoff, Hunter argues that the physical metaphor for cyberspace is so embedded as to be nearly impossible to replace, despite scholarly rejection of the association between the online world and a distinct physical space. The best we can hope for is to contest the implications of the metaphor. See id. Spectrum poses a similar challenge. Like cyberspace, it is an unfamiliar, difficult concept. A physical space of frequency bands is much more comfortable to imagine. By attacking the “spectrum as land” metaphor, I do not imagine it can be eradicated. If policy-makers could understand not to treat spectrum as property simply because land is property, that would be sufficient. For the same reason, I will continue to use the term “spectrum” in the remainder of this article.
 The word “wireless” is not without its own difficulties. It is, like the horseless carriage, defined by what it is not. In the 1880s, when essentially all long-distance communication passed through wires, it was useful to speak of Marconi’s invention as removing those wires. The lack of wires, however, no more describes the essential elements of wireless communication than the lack of a horse describes an automobile. An electromagnetic wave propagating through the air is no different than a wave propagating through a wire. The wire confines the signal to a defined physical space, which from a practical standpoint makes problems of interference and reception go away. It also reduces the legal and economic difficulties of determining the ownership or location of a wireless transmission. These are distinctions, but whether they make a difference is an empirical matter.
 The ether has occasionally made explicit appearances. Coase quotes the Congressional testimony of Josephus Daniels, the secretary of the Navy, in 1918: “There is a certain amount of ether, and you cannot divide it up among the people as they choose to use it; one hand must control it.” See Coase, The FCC, supra note 17, at 3.
 Coase used this analogy. See id. at 32-33. The musical scale may have some instructional or emotive value, but the fact the notes may be arranged this way provides no basis for rationing them through a price mechanism.
 Ironically, Herzel’s original article was properly focused on equipment rather than spectrum. Herzel suggested a market mechanism to address competing transmission standards for color television, which had been the subject of controversy at the FCC. He did not recommend treating the spectrum as freely alienable private property. What he actually proposed were long-term leases with complete technical flexibility. See Herzel, supra note 24, at 811. It was Coase who, while stating that spectrum was not a thing, redirected the property approach toward interference optimization in a hypothetical spectrum resource.
 See Carol M. Rose, The Several Futures of Property: Of Cyberspace and Folk Tales, Emission trades and Ecosystems, 83 Minn. L. Rev. 129, 164 (1998) (pointing out the prevalence of “hybrid” property rights).
 Existing spectrum licenses and proposed spectrum property rights involve other restrictions, principally power, geographic location, and time. (De Vany and his collaborators combine power and location into a single variable representing the output area of the signal. See De Vany et al, supra note 26.). However, frequency is always one of the parameters. It is usually the distinguishing one. No one talks about owning a right to emit 10,000 watts or a right to transmit all day; the discussion always centers on control a frequency band such as 800-806 MHz.
 See, e.g., eBay, Inc. v. Bidder’s Edge, Inc., 100 F. Supp. 2d 1058 (N.D. Cal 2000) (finding trespass to chattels for automatically extracting data from an auction Website); Intel Corp. v. Hamidi, 94 Cal. App. 4th 325, 114 Cal. Rptr. 2d 244 (Ct. App. 2001) (finding trespass to chattels for sending unwanted email to company employees); Compuserve, Inc. v. Cyber Promotions, Inc., 962 F. Supp. 1015 (S.D. Ohio 1997) (finding trespass to chattels for sending unsolicited bulk email).
 See Goodman, supra note 3 at 13; White, supra note 28, at 3 (“Throughout this essay I will employ the analogy of real estate and the property rights that attach to real estate....”); Thomas W. Hazlett, The Rationality of U.S. Regulation of the Broadcast Spectrum, 33 J.L. & Econ. 133 (1990) (applying legal doctrines derived from land to spectrum); Peter Huber, Law and Disorder in Cyberspace: Abolish the FCC and Let Common Law Rule the Telecosm 29 (1997); Hazlett, Spectrum Flash Dance, supra note 49, at 814 (“The analogy to land – an excellent analogy...”). Even Coase, who later in his article stated quite clearly that spectrum was not a physical resource, fell into the trap of drawing parallels between spectrum and land in connection with interference. See infra note 90.
 The land analogy is problematic in cyberspace as well. See Dan L. Burk, The Trouble with Trespass, 3 J. Small & Emerging Bus. L. 27, 33 (2000) (questioning the application of trespass to cases such as Hamidi); Hunter, supra note 70.
 This analogy comes from Joseph Singer. See Joseph William Singer, Entitlement: The Paradoxes of Property (Yale Univ. Press 2000) at 3. Coase also used the example of a gun, but for a different point. See infra TAN 254.
 See Singer, supra note 80.
 See Levin, supra note 26; Harvey Levin, The Radio Spectrum Resource, 11 J. Law & Econ. 433, 433 (1968) (“For practical purposes, the radio spectrum is a three-dimensional natural resource...”). See also Christian A. Herter, Jr., The Electromagnetic Spectrum: A Critical Natural Resource, 25 Natural Resources J. 651 (July 1985).
 Id. at 437 (emphasis added).
 See id. at 454. Levin argues that spectrum has elements of both a fixed “stock” resource (like minerals) and a fluid “flow” resource (like water). Noam argues that spectrum is all flow, see Noam, supra note 2, leading Hazlett to counter that there is no difference between the two categories, see Hazlett, Spectrum Flash Dance, supra note 49. The reason for this disagreement is that the basic analogy to natural resources is flawed.
 Subsequent scholarship has challenged Levin’s assumption that common property resources always experience the tragedy of the commons. See Elinor Ostrom, Governing the Commons: The Evolution of Institutions for Collective Action (1990). I discuss this line of argument below. See infra TAN 247.
 See White, supra note 28, at 5 (“For spectrum users, however, transmission interference has been considered fundamental and has provided the tried-and-true justification for the rejection of explicit property rights...”.)
 See Benkler, Some Economics, supra note 4.
 And in fact, Benkler develops an economic analysis of why the equipment-oriented market structure is likely to maximize communications capacity. See id.
 See Coase, The FCC, supra note 25, at 32-33.
 With regard to interference, Coase fell into the very trap of equating spectrum with a physical resource such as land that he warned against later in his paper: “It is clear that, if signals are transmitted simultaneously on a given frequency by several people, the signals would interfere with each other and would make reception of the messages transmitted by one person difficult, if not impossible. The use of a piece of land simultaneously for growing wheat and as a parking lot would produce similar results.” See Coase, The FCC, supra note 17, at 25.
 See Hazlett, Wireless Craze, supra note 26, at 20 (recognizing that the proper goal is not to minimize but to optimize the level of interference). The ubiquity of interference is a consequence not only of the fact that signals radiate indefinitely, but also of sub-harmonic and intermodulation effects. Geographically adjacent systems can experience interference when their transmissions are on exact fractions of each other’s frequency, or when the sum of two frequencies equals a third. See Minasian, supra note 26, at 226 (describing sub-harmonic radiation); De Vany et al, supra note 26, at 1520-21 (describing intermodulation).
 As a matter of fundamental physics, the two waves are in a state of quantum superposition. See David P. Reed, How Wireless Networks Scale: The Illusion of Spectrum Scarcity, Presentation to FCC Technological Advisory Council, Apr. 26, 2002, at http://www.reed.com/OpenSpectrum/Spectrum%20capacity%20myth%20FCC%20TAC.ppt.
 The common engineering measure of interference is signal to noise ratio, where noise represents any detectable emissions other than the desired transmission for that particular receiver. Of course, if the receiver always knew what was signal and what was noise, it would have no trouble distinguishing between the two. Smarter systems can tolerate lower signal to noise ratios.
 See Shannon, supra note 32.
 Even in that simplest case, there are more possibilities than it seems. Most computer dial-up modems sold today deliver 56 kilobits per second, which exceeds that Shannon’s Law limit for the relevant channel characteristics. . The secret is that the modems are asymmetric: they send data down to the user faster than the user can send data back to the network. Since most online applications today, such as email, browsing the Web, and downloading files, involve more downstream than upstream transmission, this is a tradeoff worth making.
 See infra Part II(A)(3)
 See Glenn Fleishman, Take the Mesh-Networking Route: Mesh Networks Offer an Agile, Cost-Effective Alternative, Infoworld, March 7, 2003; Steven M. Cherry, Broadband a Go-Go , IEEE Spectrum, May 30, 2003. Tim Shepard’s 1995 dissertation outlined how a meshed network could overcome traditional capacity constraints. See Timothy Shepard, Decentralized Channel Management in Scalable Multihop Spread-Spectrum Packet Radio Networks (1995) (unpublished Ph.D. dissertation, Massachusetts Institute of Technology), available at http://www.lcs.mit.edu/publications/pubs/pdf/MIT-LCS-TR-670.pdf. Meshed networks may be particularly valuable for providing “last mile” broadband connection. See Tim Fowler, Mesh Networks for Broadband Access, IEE Review, January 2001, at 17.
 See Benkler, Some Economics, supra note 4, at 44; Comments of David Reed, supra note 43. The two concepts are similar. Cooperation gain refers to mechanisms in which devices act together to enhance capacity; diversity gain means examining more than one signal path for the same purpose.
Though these cooperative approaches are important technical foundations for the commons critique, the argument does not rest on a specific architectural proposal such as a densely meshed network. Stuart Benjamin, for example, takes Benkler’s “ideal network” as a blueprint rather than a theoretical construct, thereby reducing the many variations of wireless commons to a single form, which he calls “abundant networks.” See Stuart Minor Benjamin, Spectrum Abundance and the Choice Between Private and Public Control, __ N.Y.U. L. Rev. __ (2003). Benjamin then proceeds to attack the commons position based on problems with his straw-man conception. See id. Dense meshed networks have exciting potential, but they are not the only mechanism for a commons.
 See Reed, supra note 92.
 See News Release, FCC Chairman Kennard Urges Three-Pronged Strategy To Promote Wireless Web, May 31, 2000, available at http://www.fcc.gov/Bureaus/Miscellaneous/News_Releases/2000/nrmc0032.html
(“All of the new technologies -- mobile phones, faxes, wireless computers -- are consuming spectrum faster than we can make it available, and we are in danger of a spectrum drought.”)
 See New America Foundation & Shared Spectrum Company, Dupont Circle Spectrum Utilization During Peak Hours, available at http://www.newamerica.net/Download_Docs/pdfs/Doc_File_183_1.pdf.
 An equivalent concept appears in tort law in the form of the “eggshell plaintiff” problem. People and companies are not always required to act in a manner that protects even the most abnormally sensitive person. The “reasonable person” standard is an attempt to codify this distinction. I consider tort law below in Part III(D)(2).
 See Ronald H. Coase, The Problem of Social Cost, 3 L.J. & Econ. 1 (1960).
 The case is Sturges v. Bridgman, 11 Ch. D. 852 (1879). See Coase, The FCC, supra note 25, at 26; Coase, supra note 103, at 2. Ironically, the court granted the doctor an injunction even though he built his examining room substantially after the confectioner began operating his machinery. In other words, the court applied exactly the opposite of the “first in time, first in right” doctrine Hazlett suggests would govern a spectrum common law regime. See infra note 197.
 Benkler uses Coase’s example to attack Faulhaber and Farber’s contention that a non-interfering easement within a property system eliminates the need for commons. See Benkler, Some Economics, supra note 4, at 39. I address the easements proposal in more detail below in Part IV(E).
 This is not a theoretical question. In the FCC’s ultra-wideband (UWB) proceeding, manufacturers of Global Positioning System (GPS) equipment argued that UWB should not be permitted because it would prevent reception of GPS signals. See Revision of Part 15 of the Commission’s Rules Regarding Ultra-Wideband Transmission Systems, First Report and Order, 17 F.C.C.R. 7435 (2002) [hereinafter UWB Order]. Engineering studies revealed that most GPS equipment was robust enough to ignore the UWB signals, but some poor-quality receivers might experience interference. The FCC had to choose between protecting UWB transmitters and protecting GPS receivers.
 47 C.F.R. § 2.1(c) (2002).
 See Coase, The FCC, supra note 17, at 27 (“It is sometimes implied that the aim of regulation in the radio industry should be to minimize interference. But this would be wrong. The aim should be to maximize output.”)
 Hazlett recognizes that this is the correct formulation. See Hazlett, Wireless Craze, supra note 28. The idea that interference must be eradicated is a legacy of the old government licensing regime, which he hopes to replace with a more fluid property system. However, when it comes to the commons proposal, Hazlett suddenly frets about “over-exploitation and airwave chaos.” Id. at 130. He takes the large sums spent on FCC license auctions as evidence investors value freedom from interference. Indeed they do, but the auctions provide no mechanism to determine whether those investors are making the optimal tradeoff between interference they tolerate and additional interference-generating communications. The licenses are exclusive, and moving to any arrangement that allows some interference if its value exceeds its costs requires transactions that are unlikely to occur. See infra Part IV(D).
 Transmitters even generate internal noises that interferes with themselves.
 See, e.g., Thomas W. Hazlett, The Rationality of U.S. Regulation of the Broadcast Spectrum, 33 J.L. & Econ. 133, 138 (1990) (“The interference problem is widely recognized as one of defining separate frequency ‘properties’...”).
 Frequency means what it sounds like: how often the wave oscillates. The higher the frequency, the shorter the wave. (Think of an undulating piece of string bunched up or pulled taut.)
 See supra note 76. Even Part 15 and ultra-wideband devices are limited in which frequencies they can exploit.
 See U-NII Order, supra note 34; Benkler, Overcoming Agoraphobia, supra note 30, at 42-43 (calling the U-NII Order “inspiring”).
 See Benjamin, supra note 98.
 See Faulhaber & Farber, supra note 3.
 Levin makes the most serious attempt to incorporate enhancements in equipment capability into a model of exclusive spectrum rights. See Levin, supra note 82. He describes possible improvements in the intensive margin (denser multiplexing of frequencies) and extensive margin (transmission at higher frequencies than was previously possible). Because his analysis assumes the primacy of frequency, improvements in the intensive and extensive margins as he describes them are simply further subdivisions of the pre-existing frequency pie represented by the traditional spectrum assignment chart. Improvements and sharing along planes orthogonal to the boundaries of established rights do not enter into his consideration. Hazlett similarly dismisses novel wireless techniques as “not new, not unique” because he reduces them incorrectly to frequency re-use. Hazlett, Wireless Craze, supra note 28 at 132.
 See UWB Order, supra note 106. The FCC has for some time allowed very low-power Part 15 devices to operate on an unlicensed basis throughout most of the spectrum, except for certain restricted bands. See infra TAN 173. These devices, however, tend to be frequency limited. The power limits for Part 15 devices are so strict that they generally cannot be used for high-speed or long-range applications.
 See David G. Leeper, Wireless Data Blaster, Sci. Am. May 2002.
 The FCC allowed UWB to operate between 3 GHz and 10 GHz. See UWB Order, supra note 106. However, it defined UWB as any system with bandwidth exceeding 500 MHz. See id. at . Some vendors are building “multiband” UWB equipment that splits the available spectrum into slices, and can simultaneously transmit on any of them.
 Coase presciently anticipated this problem in 1958. Though UWB had not been developed, he pointed out that “some types of medical equipment can apparently be operated in such a way as to cause interference on may frequencies and over long distances.” Coase, The FCC, supra note 17, at 30. Coase believed that such a situation called for regulation, even within the general context of property rights he advocated.
 See Additional Spectrum for Unlicensed devices Below 900 MHz and in the 3 GHz Band, Notice of Inquiry, ET Docket No. 02-380 Comments of Intel, at 4 (Apr. 7, 2003) (“Preliminary technical analysis conducted by Intel and testing performed by the Communications Research Centre Canada, on Intel’s behalf, demonstrate that technically viable broadband services can be operated on a non-interfering basis with both analog and digital TV broadcast services in a major metropolitan area in which many overlapping TV service contours exist.”) See also FCC Spectrum Policy Task Force, Report of the Interference Protection Working Group at 4 (Nov. 15, 2002), available at http://www.fcc.gov/sptf/files/IPWGFinalReport.pdf, at 5 (“Due to advances in digital signal processing and antenna technology, communications systems and devices are becoming more tolerant of interference through their ability to sense and adapt to the RF environment.”)
 See William Lehr, et al., Software Radio: Implications for Wireless Services, Industry Structure, and Public Policy, Massachusetts Institute of Technology Program on Internet and Telecoms Convergence (August 2002) at 11 (“It is not unreasonable to consider software radio as a disruptive technology, with the potential for radically altering the structure of the industry within which radios are designed, manufactured, deployed and operated.”), at http://itc.mit.edu; Authorization and Use of Software Defined Radios First Report and Order, 16F.C.C.R. 17373 (2001); Spectrum Policy Task Force Report, ET Docket No. 02-135, Comments of Vanu, Inc. at 1-2 (“Software-Defined Radio (SDR) technology brings unprecedented flexibility to wireless systems, and will be able to take advantage of new, more flexible regulatory policy.”).
 Christian Bourge, New Tech Feeds Spectrum Debate, Wireless Newsfactor, June 30, 2003, at http://www.wirelessnewsfactor.com/perl/story/21828.html. The military is a major contributor of funding and research for advanced wireless systems. Soldiers, especially when operating in a foreign country, may not have dedicated spectrum assigned for their communications needs. The hope of XG, as well as the military’s Joint Tactical Radio System, a software radio program now in procurement, is to use spectrum opportunistically, without needing prior allocations.
 See Joseph Mitola, Cognitive Radio for Flexible Mobile Multimedia Communications, IEEE Mobile Multimedia Conference, 1999; Tomorrow’s 5G Cell Phone, Infoworld, Feb. 28, 2003 (interview with Joseph Mitola on cognitive radio).
 See Nobuo Ikeda, Spectrum Buyouts: A Proposal for the Transition to Open Spectrum, ISART 2003. In theory, time division could be applied to many wireless communications systems. With the exception of broadcast services that transmit continuously, most forms of wireless communication are intermittent. Even when the transmission appears continuous to the receiver, it may in fact be a series of packets interspersed with white space so quickly that the ear cannot detect the discontinuity. For example, such time-division multiplexing is used in GSM, the dominant standard for today’s digital mobile phone networks.
 See Robert Matheson, The Electrospace Model as a Tool for Spectrum Management, ISART 2003.
 See P. Gupta & P.R. Kumar, Towards an Information Theory of Large Networks: An Achievable Rate Region, IEEE International Symposium on Information Theory, Washington, DC, June 2001; M. Grossglauser & David Tse, Mobility Increases the Capacity of Adhoc Wireless Networks, Proceedings of the IEEE Infocom Conference, April 2001.
 See Bell Labs Scientists Shatter Limit on Fixed Wireless Transmission, Lucent Technologies News Release (Sept. 9, 1998), at http://www.bell-labs.com/news/1998/september/9/1.html; Andrew Backover, Wireless Links Could Blast Off, USA Today, November 17, 2002; G.J. Foschini & M.J. Gans, On Limits of Wireless Communications in a Fading Environment When Using Multiple Antennas, 6 Wireless Personal Communications 311 (1998).
 See John Markoff, Start-Up Plans to Introduce Alternate Wi-Fi Technology, New York Times, Aug. 18, 2003.
 See Amendment of Parts 2 and 25 of the Commission’s Rules to Permit Operation of NGSO FSS Systems Co-Frequency with GSO and Terrestrial Systems in the Ku-Band Frequency Range, Memorandum Opinion and Order and Second Report and Order, ET Docket No. 98-206, 17 FCC Rcd. 9614 (2002). Northpoint first brought its technology to the FCC in 1994 and filed formal applications for spectrum allocation in 1999. The FCC, though finding in principle that the Northpoint system can share spectrum with direct broadcast satellite systems in the same band, has not yet assigned the spectrum Northpoint seeks. See Catherine Yang, The Scuffle over Sharing Spectrum, Bus. Week, April 19, 2002; Ted Hearn, Northpoint Can’t Pry Loose Spectrum, Multichannel News, April 29, 2002. Northpoint has sued the FCC to gain access to the spectrum. See Ted Hearn, Northpoint Takes FCC to Court, Multichannel News, June 26, 2002.
 Martin Cooper, Antennas Get Smart, Sci. Am., July 2003.
 Space-time coding distinguishes based on the relative location of transmitters and receivers. Multiplexing by physical location, which could be called space division, means distinguishing by the absolute location in space of the overall wireless system.
 The signal is not confined to the house. As noted, it can penetrate walls or other obstacles. However, many low-power indoor signals attenuate so greatly in that process as to be largely undetectable outside.
 See Chartier, supra note 7.
 See Reed, supra note 43 (observing that the technological changes affecting communications systems are “far from reaching any fundamental limits.”) To be clear, I am not saying that wireless capacity is infinite. At any given time there will be an optimal amount of communication that can be supported, based on demand and the state of technology. How congested spectrum seems reflects the ratio between actual communication and what is theoretically possible, modified by the cost of routing around other signals and the degree of degradation those signals generate. At some point spectrum can feel uncongested for important applications even if there are real limits on how much can be transmitted.
Hazlett invokes Say’s law, popularly formulated as “supply creates its own demand,” to attack the “physical abundance” arguments of commons advocates. See Hazlett, Wireless Craze, supra note 28, at 135. Say’s Law states that total demand will never fall below or exceed total supply in an economy as a whole. It does not suggest that supply and demand for any one input are always in equilibrium. Even in the absence of legal or regulatory constraints, there are costs for any potential spectrum user to make use of available communications capacity. For example, better or different equipment may be required. There are also substitutes for wireless communication, including wireline transmission and storage, that factor into the decision whether to exploit any “empty space” for wireless transmission. Says’ Law is not violated in a world where total communications capacity is in equilibrium but some users can engage in wireless transmission as though capacity were not scarce.
 See Kevin Werbach, The Architecture of Internet 2.0, Release 1.0, February 1999 (arguing that the closed architecture of some broadband networks would constrain growth and innovation); Lawrence Lessig: Code and Other Laws of Cyberspace (1999). The first prominent thinker to use “architecture” in this context was Mitch Kapor, founder of Lotus Development Corp. and co-founder of the Electronic Frontier Foundation. cite
 The Internet and the public switched telephone network both route messages among distant nodes, but they have very different architectures. The Internet’s distributed packet-switching and end-to-end principle make it a more open platform for innovation than the centralized, circuit-switched phone network.
 See Reed, supra note 43 (“[T]he economic value in a communications system architecture does not inhere in some abstract ‘ether’ that can be allocated by dividing it into disjoint frequency bands and coverage areas. Instead it is created largely by the system design choices...”).
 These three variables make up the TAS system that De Vany and his collaborators propose.
 Each tower is only responsible for communication within a small radius. Towers therefore can use relatively low power, enabling users connected to one tower and users connected to another on the same frequency to talk at the same time. This phenomenon is known as spectrum or frequency re-use. Cellular technologies are often compared on the basis of their frequency re-use efficiency. Re-use is an example of how technology can create “more” spectrum out of nothing. It is not, however, a complete description of current capacity-enhancing techniques, as Hazlett suggests. See Hazlett, Wireless Craze, supra note 28, at 132 (“The key scientific break-through claimed to unleash unlimited bandwidth is frequency re-use.”) Frequency re-use is a form of subdivision that leaves the basic pattern of wireless transmission intact. In effect, it is a simple trade of power for range. Systems that exploit cooperation and diversity gain, as well as software-defined or cognitive radios, do not rely on this mechanism alone.
 See Benkler, Overcoming Agoraphobia, supra note 30. De Vany et al acknowledge their system of spectrum property rights systematically benefit transmitter owners rather than receiver owners. See De Vany et al, supra note 26, at 1518. This choice may have been justified in a broadcast context, where receivers are merely passive devices. As receivers become more intelligent and contribute to optimization of spectrum use, however, the bias in the legal regime away from them becomes a negative instead of a positive. I consider further the different incentives of property and commons regimes in Part IV.
 See supra note 78; White, supra note 28, at 20 (“In sum, this ideal system would look much like the current system of property rights that apply to real estate.”). Mueller does focus on equipment rights, but these are rights to deploy particular kinds of transmitters and antennas, rather than rights to emit particular kinds of radiation. See Mueller, supra note 26.
 See Goodman, supra note 3.
 See Hunter, supra note 70 (explaining the importance of metaphor).
 See infra Part IV(D).
 Hazlett and Huber argue that policing radio transmissions through a generalized legal framework not specific to communications is itself beneficial, because it reduces the possibility of regulatory capture or inefficiency. See Hazlett, supra note 49, at 807; Huber, supra note 78. There are three responses to this argument. First, one size doesn’t always fit all. If wireless is sufficiently distinct, shoehorning it into generalized models does more harm than good. Second, I proposing to use generalized common law frameworks to resolve spectrum disputes, just not on the basis of exclusive property rights. Third, the supercommons model can be seen as part of a different generalized framework. The framework involves networks, which have powerful characteristics that have not fully been analyzed for their legal significance. However, this paper is not the place for a full explication of network law.
 Herzel’s groundbreaking article was, according to its own title, focused on color television standards.
 Annual Assessment of the Status of Competition in the Market for the Delivery of Video Programming, 17 FCC Rcd 26901, 26975 (2002). The number is higher for UHF stations. Nationwide, only seven percent of assigned UHF channels are even in use. See J.H. Snider & Max Vilimpoc, Reclaiming the “Vast Wasteland:” Unlicensed Sharing of Broadcast Spectrum, New America Foundation Spectrum Series Issue Brief #12 (July 2003). Most of the value of the broadcasters’ licenses today is actually tied to their government-issued “must-carry” rights to have their content broadcast on cable and satellite distribution platforms.
 The bulk of television broadcast revenue now comes from legally mandated “must carry” retransmission on cable and satellite platforms. The biggest stick in the broadcaster’s bundle of “spectrum rights” is the entitlement to transmission over a wire!
 See Benkler, Some Economics, supra note 4, at 29 (“The correct object of optimization is wireless network communications capacity.”)
 Benkler does identify other benefits of the commons approach outside the realm of comparative economic efficiency. Following Lessig’s discussion in The Future of Ideas, he argues that a commons policy is better for innovation. He also notes that, because a commons-based market is based on end-user equipment purchases rather than carrier infrastructure build-outs, the commons network will respond more quickly to user preferences. See id. at 50.
 Benkler’s position is that policy-makers optimize for information-theoretic capacity in terms of bits, because the costs of determining whether subjective value has been maximized is likely to be too high in lost capacity. He may be right. However, my argument is that a real-time mechanism of common law backstops and safe harbors can optimize for value better than any a priori allocation such as a price mechanism.
 See Dale N. Hatfield, The Current Status of Spectrum Management, in Balancing Policy Options in a Turbulent Telecommunications Market: A Report of the Seventeenth Annual Aspen Institute Conference on Telecommunications Policy (Aspen Institute 2003), available at http://www.aspeninst.org/AspenInstitute/files/CCLIBRARYFILES/FILENAME/0000000137/balanceturbtelecom.pdf; Werbach, Paradise of the Commons, supra note 41, at 8-9 (quoting Hatfield).
 See Faulhaber & Farber, supra note 3, at 7-8; Minasian, supra note 26, at 263.
 See Goodman, supra note 3.
 See Benkler, Overcoming Agoraphobia, supra note 30; Benkler, Siren Songs.
 Testimony of Peter F. Guerrero Before the U.S. Senate Committee on Commerce, Science, and Transportation (June 11, 2002), at 4.
 Cf. FCC Interference Working Group Report, supra note 122, at 12 (“[T]he Working Group recommends that the Commission consider addressing its long-term interference measurement and spectrum policy challenges by supplementing is transmitter-centric approach with a new paradigm based on (1) real-time measurements of spectrum use and the RF environment and (2) adaptive responses of transmitters and receivers to these measurements.”)
 See Noam, Yesterday’s Heresy, supra note 2.
 In this case, the tradeoff imposes a penalty in the form of latency, or delay prior to the start of the video transmission. Caching, or storing frequently-accessed content close to the user and intelligently delivering it from those nearby stores rather than across the network, has the opposite effect.
 Hazlett mistakenly sees this feature as a bug. See Hazlett, Wireless Craze, supra note 28. Noting that spectrum commons advocates often draw analogies to the Internet, he argues that the Internet is inherently wasteful and will fail to deliver promised service without imposition of usage-based pricing mechanisms. It is true that some services such as real-time streaming video are difficult to deliver on a best-efforts network. One response has been content delivery networks such as Akamai that use local caches to improve performance. Caching is another example of “wasting” storage and processing to economize on transmission. Another solution has been the deployment of private Internet connections, in some cases employing usage-based pricing. Contrary to many predictions, though, the existence of such “premium” services has neither displaced the best-efforts Internet nor relegated it to awful performance. The Net has been exceptionally resilient to its own growth and to attacks and failures which might have shut it down. Hazlett points to fiber backbone operators that raised large amount of capital and had huge market capitalization based on the premise they would overcome the Internet’s messy capacity limitations. Most of these companies subsequently saw their stock prices collapse and several are in bankruptcy. Few of the promised bypass routes have been built. Yet the “wasteful” Internet continues to grow.
 Under the famous Moore’s Law, computational power doubles approximately every eighteen months, all other things held equal. Transmission capacity on a per-link basis is doubling roughly every 12 months, though end-to-end capacity improves more slowly because it is no better than the weakest link in the chain. Storage is improving the fastest of all – doubling in capacity every nine months. See Dan Orzech, Rapidly Falling Storage Costs Mean Bigger Databases, New Applications, CIO Update, June 4, 2003, at http://www.cioupdate.com/trends/article.php/2217351.
 Thomas Kuhn, The Structure of Scientific Revolutions (1962).
 For this reason, the philosopher Karl Popper has described the scientific method in terms of falsibiability. See Karl Popper, The Logic of Scientific Discovery (1934). A non-scientific contention, such as “God exists,” may in fact be true. The trouble is that, if it is not, there is no means to prove it false.
 Joao Magueijo, Faster Than the Speed of Light: The Story of a Scientific Speculation (2003).
 Astronomers used the same method that had predicted Neptune to postulate a planet inside Mercury. However, this time they found nothing where the planet should have been. The discrepancy turned out to be evidence that at these great scales, Newtonian mechanics broke down. See id.
 Herbert Hoover, remarks at the First National Radio Telephony Conference, Feb. 27, 1992, cited in Chartier, supra note 7, at 3.
 One study estimates the delay reduced US economic welfare by at least $86 billion in 1990 dollars. See Jeffrey H. Rohlfs, Charles L. Jackson, and Tracey E. Kelly, Estimate of the Loss to the United States Caused by the FCC’s Delay in Licensing Cellular Communications (National Economic Research Associates, Nov. 8, 1991).
 The commons may not be as difficult to reverse as it appears. See TAN 336.
 Coase, though rightly cited as the father of the property rights approach, made clear that only with experience could it be determined how widely such rights should be employed. See Coase, The FCC, supra note 17, at 34.
 Proponents of both models now advocate a hybrid approach to move forward. See supra note 3. The convergence of property and commons models for spectrum policy tracks efforts in other areas to bring together these two approaches. See, e.g., Rose, supra note 75 (describing “limited common property” situations that are exclusive on the outside and commons on the inside).
 47 C.F.R. §15.103 (2002).
 Technically speaking, power is more complicated concept, because it can be measured in different ways for wireless systems. A spread-spectrum system may have relatively high total power output, but low power on any one of the many frequencies it uses (what is known as power spectral density). Or the power may be focused in a narrow beam which goes a long distance but does not spread significantly in other directions.
 It may, however, be possible to deliver a wide-area broadcast service at low power using some of cooperation gain techniques such as mesh networking, directional antennas, and intelligent coding. These techniques do impose a tax in the form of latency or higher equipment costs, though these are technological variables that will decrease over time. It is fair to say that existing commercial unlicensed systems are no substitute for broadcast, and that for the foreseeable future low-power unlicensed systems are likely to be a more difficult or expensive way to do broadcast than high-power narrowband.
 See Spectrum Task Force Report, supra note 57, at 39.
 See TV Band NOI, supra note 61.
 See supra note 131.
 Howard Shelanski and Peter Huber, reach a similar conclusion from a different direction. See Howard A. Shelanski and Peter W. Huber, Administrative Creation of Property Rights to Radio Spectrum, 41 J. Law & Econ. 581 (1998). Their objective is to demonstrate that the FCC has actually created significant property rights in spectrum through administrative decisions regarding licenses. To consider these attributes “property rights,” they must replace the idea of a holistic “ownership” with a legal realist bundle of rights. And to do that, they must acknowledge that spectrum is not a thing to be owned:
The mere label of “ownership” is unhelpful. With spectrum especially, that label often obscures more than it illuminates. There is no such thing as “spectrum” out there, any more so than there was “ether,” to be bottled by the Commission or anyone else. “Spectrum” is composed entirely of the engineering characteristics of transmitters and receivers.
Id. at 584. Huber and Shelanski go on to build an impressive case that the FCC has, in fact, strengthened the bundle of transmission and reception rights it allocates to spectrum licensees. Doing so does not, however, imply that the FCC should be moving toward greater property rights, merely that it can. Overcoming the spectrum fallacy as a descriptive matter destroys much of the basis for the program of spectrum propertization.
 “If the right to use a frequency is to be sold, the nature of that right would have to be precisely defined.” Coase, The FCC, supra note 17, at 25. Yet Coase himself failed to offer such a definition in his original paper. He merely stated that rights should be alienable, and subject to subdivision, aggregation, or modification upon agreement of the parties involved. See id. De Vany et al., writing a decade later, acknowledged that no one had yet provided a concrete proposal for spectrum property rights: “It is our belief that the parties to this debate have never truly joined issue. The market mechanism as a theory cannot be offered as an alternative to FCC regulation; to make the debate useful it is first necessary to articulate a detailed system of property rights in spectrum usage.” De Vany et al, supra note 26, at 1501 (emphasis in original). Coase and two co-authors did develop a more thorough proposal for spectrum property rights in 1963, but it was never published. See Ronald Coase, William H Meckling, and Jora Minasian, Problems in Radio Frequency Allocation (unpublished manuscript, RAND Corp., May 1963).
 See Goodman, supra note 3.
 “Before two parties can enter into a contract, we must define what they own. Otherwise, we cannot determine who is buying and who is selling. In situations involving neighboring owners, for example, we cannot just leave it to the market – that is, rely on private contracting alone – to determine which party should prevail.” Singer, Entitlement, supra note 80, at 50.
 Property advocates argue that any property rights are better than the current licensing regime. And they may be correct. However, a short-term efficiency gain at the cost of long-term structural inefficiency is not a worthwhile trade.
 See Revision of Part 15 of the Commission’s Rules Regarding Ultra-Wideband Transmission Systems, Notice of Proposed Rule Making, ET Docket No. 98-153, 65 Fed. Reg.37332, June 14, 2000. UWB’s basic technology is described above. See supra TAN 118.
 Hazlett criticizes the FCC for taking so long to authorize UWB. The alternative he proposes is an administrative allocation of several exclusive underlay property rights: “Several licenses could be allocated per band, up to the ceiling set by the ‘noise floor’ limits extended licensees transmitting over the underlay rights.” Hazlett, Wireless Craze, supra note 26, at 151. This simply wishes away the thorny problem of potential interference that was the cause of the drawn-out FCC proceeding. There are no clear “noise floor limits” in existing FCC licenses, nor does Hazlett propose any for the spectrum property rights he favors. Furthermore, the argument of the licensed service providers is that even if one UWB device is infinitesimally faint, many will in aggregate exceed the noise floor. Hazlett offers no mechanism by which his underlay rights-holders would be able to show they were not exceeding noise floor boundaries in aggregate.
 See UWB Order., Statement of Commissioner Michael J. Copps, available at http://www.fcc.gov/Speeches/Copps/Statements/2002/stmjc205.html (calling the limits “ultra-conservative”).
 See Petition for Reconsideration in ET Docket No. 98-153 of Sprint PCS (filed June 17, 2002) at 4-8 (arguing PCS licensees hold exclusive licenses and that the FCC cannot require them to be shared); Letter from Charles W. McKee, Sprint PCS, to Bruce A. Franca, FCC, Feb. 21, 2001, at 6-7 (“Having received valuable compensation for issuing exclusive licenses, the Commission does not now have the legal authority to convert these licenses into non-exclusie licenses and to require Sprint PCS to share its spectrum with others, much less share its spectrum for free.”).
 See Ultra-Wideband Order, supra note 118, at. ¶271.
 The distinction is a semantic one. Though the FCC did not use these terms, it found that Sprint’s “exclusive” license gave it rights to exclude other high-power transmitters, but not to be the sole exclusive transmitter operating in those frequencies.
 See Stuart Minor Benjamin, The Logic of Scarcity: Idle Spectrum as a First Amendment Violation, 52 Duke L.J. 1 (2003), at n.259 (recognizing that “new uses of allocated spectrum raises the question of the breadth of the existing licenses”).
 Even this is not always the case. In some frequencies, there are primary and secondary licensees. The secondary licensee is not protected against interference from the primary licensee, but is protected against harmful interference from others.
 47 C.F.R. §97.3(a)(23).
 In reality, the interference rules have been interpreted to bar transmissions that would interfere with a hypothetical receiver within the licensee’s footprint, even if no such receiver exists. In other words, the mere fact that someone could put a television in a particular spot and have their service degraded is enough to shut down another transmitter.
 See supra note 89.
 See Goodman, supra note 3. Property advocates acknowledge the need for dispute resolution mechanisms. See Hazlett, Wireless Craze, supra note 28 (proposing a spectrum court to replace the FCC).
 See supra note 155. The FCC, in fact, already uses doctrines similar to those in nuisance law. See Goodman, supra note 3, at 49.
 See Hazlett, supra note 111, citing Tribune Co. v. Oak Leaves Broad. Station, Inc., et al (Cir. Ct., Cook County, Ill. 1926), reprinted in 68 Cong. Rec. 215, 215-219 (1926).
 See supra note 26.
 The De Vany system is based on three parameters: time, area, and spectrum (frequency) of transmission. Such “TAS packages” are a reasonable description of broadcast systems, but break down amid the dynamic, cooperative, non-frequency-based systems that now exist. See id. at 1501-02.
 See De Vany et al, supra note 26, at 1520.
 See supra TAN 136.
 The property right also differs in being permanent and irrevocable. However, this is more a legal difference than a practical one. See Shelanski & Huber, supra note 179. Conversely, property rights may be taken away by the government through eminent domain, so long as compensation is paid.
 See Benkler, Overcoming Agoraphobia, supra note 30.
 Benkler has proposed that commons rights include “Part 16” protection against other transmitters. See id.; Benkler, Some Economics, supra note 4.
 Matheson’s electrospace model is one framework that goes beyond frequencies to take into account other properties of wireless communication. See Matheson, supra note 127. However, Matheson’s model is still incomplete. more.
 My description of the property rights in wireless devices builds on the legal realist notion of property as a bundle of rights. See Singer, supra note 80 at 82-84.
 Cf. Charles A. Reich, The New Property, 73 Yale L.J. 773 (1964) (“Property draws a circle around the activities of each private individual or organization. Within that circle, the owner has a greater degree of freedom than without. Outside, he must justify or explain his actions, and show his authority. Within, he is master, and the state must explain and justify any interference.”).
 See Singer, supra note 80.
 See Wesley Newcomb Hohfeld, Fundamental Legal Conceptions as Applied in Judicial Reasoning, 23 Yale L. J. 16 (1913). See also Joseph W. Singer, The Legal Rights Debate in Analytical Jurisprudence from Bentham to Hohfeld, 1982 Wis. L. Rev. 975; Arthur Corbin, Jural Relations and their Classification, 30 Yale L.J. 226 (1921).
 Technically, the relationship means that there is at least one entity subject to the correlative category.
 See supra note 66.
 The full Hohfeldian system includes a second set of four categories: power, immunity, disability, and liability, in the same configuration. These describe the ability that parties have to alter the system or determine where burdens are imposes.
 If licensed devices have duties not to transmit in unlicensed bands, those duties arise from the licenses themselves. This was the FCC’s rationale for rejecting “Part 16” protection for U-NII devices, which could be seen as a collective right against systems outside the commons. It concluded such a rule was unnecessary because licensed devices were already limited by the terms of their licenses. See U-NII Order, supra note 34.
 See Singer, supra note 80.
 See supra note 131; Stephen Labaton, An Earthly Idea for Doubling the Airwaves, New York Times, April 8, 2001 (describing the history of Northpoint’s efforts); Paul Davidson, Northpoint Proposes Satellites to Get License, USA Today, March 24, 2002 (same).
 See supra note 131.
 Hazlett asserts that property rights in spectrum would solve Northpoint’s problem: “In a more efficient world, innovative wireless companies such as Northpoint would simply buy the spectrum they need, much as any company buys labor, raw materials, and capital inputs.” Testimony of Thomas W. Hazlett, Senate Commerce Committee Subcommittee on Communications, May 22, 2003, available at http://www.manhattan-institute.org/html/testimony_hazlett_5-22-03.htm. Certainly, if spectrum were private property, Northpoint could buy exclusive use of a frequency block. However, the price for controlling the entire band would presumably be higher than for what Northpoint needs: only the ability to share the band with direct broadcast satellite systems. Northpoint would also be inefficiently hoarding spectrum if it controlled an entire band. Whether Northpoint could buy the right to share the DBS spectrum in a property regime is an open question. The DBS operator and Northpoint would have to work out interference boundaries. Also, the DBS operator would have strong incentives to refuse to sell to Northpoint if it feared competition from the new service. FCC economist Douglas Webbink, a long-time property advocate, acknowledges in an article considering spectrum sharing arrangements such as Northpoint’s proposal that, “[i]n reality, defining such rights may be extraordinarily complex.” Douglas W. Webbink, Communications Convergence, Spectrum Use and Regulatory Constraints (Sept. 13, 2001), unpublished? at 5.
 See Arthur De Vany, Implementing a Market-Based Spectrum Policy, Paper presented at “FCC License Auctions: From Concept to Policy” conference, Tomales Bay, CA, July 27-29, 1996, at 25 (explaining why the decentralized approach has a strong attractor to “integrated and compatible” networks). Cf. David G. Post & David R. Johnson, “Chaos Prevailing On Every Continent”: Towards a New Theory of Distributed Decision-Making in Complex Systems, 73 Chi.-Kent L. rev 1055 (1998) (using complexity theory to argue for decentralized legal regimes in cyberspace).
 This analogy resembles Benkler’s “trade with India” analogy. See Benkler, Some Economics, supra note 4. The important difference is that the decentralized solution retains some legal regime to incent activity and resolve disputes, rather than simply allowing “free trade.” Of course, “free trade” in the real world often involves tariffs, duties, legal requirements, and other limitations.
 Benkler does advocate “underlay and interweaving” in addition to designated unlicensed spectrum. See Benkler, Some Economics, supra note 4, at 79-80.
 To take just one possibility, a study commissioned by Intel found the equivalent of 19 vacant television channels in the San Francisco Bay Area, even while licensed stations were still offering both analog and digital broadcasts. See Additional Spectrum for Unlicensed devices Below 900 MHz and in the 3 GHz Band, Notice of Inquiry, ET Docket No. 02-380 Comments of Intel, at 8 (Apr. 7, 2003).
 See Benjamin, supra note 98. Commons advocates bear some blame for this confusion. Early work advocating the spectrum commons model often described the concept in too-narrow technical terms, focusing on particular advances such as spread-spectrum. See, e.g., Lessig, Code, supra note 137, at 184 (“There is a second architecture for broadcasting (which I will call “Spread Spectrum” – it has a few different names)...”).
 See Faulhaber & Farber, supra note 3.
 See Hazlett, Wireless Craze, supra note 28, at 188-89. Hazlett proposes administrative definition of exclusive underlay property rights, along with a blanket underlay for low- power devices “creating material signal degradation only in the user’s immediate jurisdiction...”. As with other property rights, the initial boundaries man not be workable, and transactions to redistribute them may be too complicated.
 See Noam, supra note 3.
 See Benkler, Some Economics, supra note 4.
 See id at 76-80.
 Paul Baran advocated something like this in his original speech contesting the need to ration scarce frequencies. The only restrictions he though necessary were power limits: “In such an environment anyone would be allowed to use the spectrum, without the high front-end costs that keep out the true innovators. Of course, the allowable power and power densities would have to be realistically restricted.” Baran, supra note 36.
 For underlay and interweaving, Benkler postulates strict power limits and automated sensing to identify and avoid licensed transmissions. See id. at 79-80.
 See supra TAN 187.
 See Authorization of Spread Spectrum and Other Wideband Emissions Not Presently Provided for in the FCC Rules and Regulations, Notice of Inquiry, GEN Docket No. 81-413 (1981).
 See Noam, Yesterday’s Heresy, supra note 2.
 See Jon M. Peha, Lessons from Haiti’s Internet Development, Communications of the ACM, June 1999; Jon M. Peha, Wireless Communications and Coexistence for Smart Environments, IEEE Personal Communications, October 2000, at 6, 7.
 Jason Oxman, The FCC and the Unregulation of the Internet, FCC Office of Plans and Policy Working Paper No. 31 (July 1999), at http://www.fcc.gov/Bureaus/OPP/working_papers/oppwp31.pdf.
 See Regulatory and Policy Problems Presented by the Interdependence of Computer and Communications Services and Facilities, Notice of Inquiry, 7 F.C.C.2d 11 (1966) [hereinafter Computer I]; Amendment of Section 64.702 of the Commission's Rules and Regulations (Second Computer Inquiry), Final Decision, 77 F.C.C.2d 384 (1980) [hereinafter Computer II]; Amendment of Section 64.702 of the Commission's Rules and Regulations (Third Computer Inquiry), Report and Order, 104 F.C.C.2d 958 (1986) [hereinafter Computer III]. See also Robert Cannon, The Legacy of the Computer Inquiries, ___ Fed. Comm. L.J. ___ (2003).
 See Dan L. Burk & Julie E. Cohen, Fair Use Infrastructure for Rights Management Systems, 15 Harv. J.L. & Tech 42 (2001); Harper & Row, Publishers, Inc. v. Nation Enters, 471 U.S. 539, 560 (1985).
 See Wendy J. Gordon, Fair Use as Market Failure: A Structural and Economic Analysis of the Betamax Case and its Predecessors, 82 Colum. L. Rev. 1600 (1982).
 See Julie E. Cohen, Lochner in Cyberspace: The New Economic Orthodoxy of “Rights Management”, 97 Mich. L. Rev. 462, 551-59 (1998); Yochai Benkler, Free as the Air to Common Use: First Amendment Constraints on the Enclosure of the Public Domain, 74 N.Y.U. L. Rev. 354 (1999).
 See Pamela Samuelson, Fair Use for Computer Programs and Other Copyrightable Works in Digital Form: The Implications of Sony, Galoob, and Sega, 1 J. Intell. Prop. L. 49 (1993).
 The analogy is no accident; wireless communication is speech. In proposing universal entry, I am not here asserting that such an outcome is constitutionally required, though that is one possible interpretation. Stuart Benjamin has examined in detail the First Amendment implications of government decisions barring low-power radio stations that cause only minimal interference, concluding that they are constitutionally suspect. See Benjamin, supra note 190; see also Mark Nadel, A Technology Transparent Theory of the First Amendment and Access to Communications Media 43 Fed. Comm. L. J. 158 (1992) (arguing that if changing regulation can make broad access to a medium possible, the First Amendment compels the government to implement some form of it). Ironically, Benjamin opposes the commons approach, which he equates commons with government control. See Benjamin, supra note 98. Others have questioned whether spread-spectrum, by undermining the scarcity rationale for spectrum regulation, makes the current regime unconstitutional. See Yochai Benkler & Lawrence Lessig, Net Gains: Is CBS Unconstitutional?, The New Republic, Dec 14, 1998., at 12 A full analysis of the constitutional implications of the supercommons is beyond the scope of this paper. My claim here is simply that the supercommons approach is the best policy framework.
 See Mark M. Bykowsky and Michael J. Marcus, Facilitating Spectrum Management Reform via Callable/Interruptible Spectrum, available at http://intel.si.umich.edu/tprc/papers/2002/147/SpectrumMgmtReform.pdf.
 See Coase The FCC, supra note 17, at 28 (noting that it may be efficient for market participants to accept some level of interference).
 Thus, for example, the FCC’s interference temperature metric, while a promising “safe harbor” mechanism within a supercommons tort regime, cannot be the sole boundary between exclusive and supercommons transmissions. See infra TAN 275.
 The fear of overuse of the spectrum commons calls to mind the Navy’s claims as early as 1910, quoted by Coase, that “there exists in many places a state of chaos” with early unregulated radio transmission. S. Rep. No. 659, 61st Cong., 2d Sess. 4 (1910), cited in Coase, The FCC, supra note 17, at 2.
 See Hazlett, Wireless Craze, supra note 28, at 130.
 For an introduction to complexity theory, see M. Mitchell Waldrop, Complexity: The Emerging Science at the Edge of Order and Chaos (1992); Roger Lewin, Complexity: Life at the Edge of Chaos (1992); James Gleick, Chaos: Making a New Science (1987).
 Garret Hardin, The Tragedy of the Commons, 162 Sci. 1243 (1968).
 See Ostrom, supra note 85; Rose, supra note 75.
 See Buck, supra note 44.
 C could build one herself, but still would have incentives to create something that could communicate with others. Moreover, one-off hobbyist devices would be built in significantly smaller numbers (and therefore cause less of an interference problem) than mass-produced equipment.
 See Amendment of Part 15 of the Commission’s Rules Regarding Spread Spectrum Devices, Second Report and Order, FCC 02-151 (May 16, 2002).
 See Hazlett, supra, note 78. See also Huber, supra note 78, at 74 (arguing that common law can solve spectrum disputes).
 See De Vany, supra note 218, at 32 (acknowledging that the system he co-developed in 1969 could just as easily be applied without property rights).
 See Coase, The FCC¸ supra note 17, at 34.
 See Goodman, supra note 3.
 See id. (explaining the difficulty of applying trespass and nuisance law to spectrum); Webbink, supra note 217.
 See Goodman, supra note 3; Guido Calabresi & Douglas Melamed, Property Rules, Liability Rules, and Inalienability: One View of the Cathedral, 85 Harv. L. Rev. 1089 (1972).
 See Goodman, supra note 3, at 45.
 See id.
 product liability literature. The De Vany group suggested strict liability for spectrum conflicts, because it proposed a “bright line” set of boundaries around exclusive rights. See De Vany et al, supra note 26.
 A regime involving private litigation does have significant inefficiencies, because litigation can be very costly and time-consuming. However, those very costs prevent most actors from entering into litigation unless the matter is very significant and they are confidence of their chances of success. Spectrum property rights advocates recognize this point. See De Vany et al, supra note 26, at 1519.
 Goodman also argues for the value of a regulator to smooth the nuisance-based dispute resolution process. See Goodman, supra note 3.
 See supra TAN 241.
 See id.
 See id.
 Benkler, Some Economics, supra note 4 (suggesting “no rules” in very high frequencies). Benkler’s proposal is initially limited to extremely high frequencies, and to non-commercial activity. This may be a valuable first step, though I believe we can go farther.
 See Benkler, Some Economics, supra note 4, at 43.
 See Buck, supra note 44; Rose, supra note 75.
 An additional technical requirement that could facilitate such a limited supercommons is transmit power control (TPC). See FCC Interference Working Group Report, supra note 122, at 16 (finding that TPC is now widely used, though generally not legally required). TPC is an engineering mechanism for politeness. In non-technical terms, it says, “don’t shout if you can whisper.” Devices can to emit no more power than necessary for the transmission they wish to engage in. A transmitter communicating with a nearby receiver will modulate its power down compared to one communicating with a distant receiver. TPC is a common element of modern wireless communications systems such as cellular networks and wireless local area networks. There is reason to believe equipment designers would employ it even absent a requirement. Mandating TPC is not the same a specifying a protocol or particular interference avoidance mechanism. It is simply a technical implementation of what a rational wireless user would do if it recognized that other users are equally free to transmit.
 See Tom W. Bell, The Common Law in Cyberspace, 97 Mich. L. Rev. 1746 (1999) (proposing a trademark-like regime for spectrum). In fact, Bell tantalizingly raises, but does not answer, some of the very questions I consider here. See id. at n. 100 (“How well would the trademark analogy protect the right to transmit via a CB radio or low power device?”).
 See id. Bell points out that the Oak Leaves case that Coase, Hazlett, and Huber cite as defining embryonic property rights in spectrum actually spoke in terms of goodwill, unfair competition, and trade names, which are trademark rather than property concepts
 See Spectrum Task Force Report, supra note 57 (describing interference temperature).
 See Spectrum Policy Task Force Report, ET Docket No. 02-135, Comments of Cingular Wireless LLC (Jan. 27, 2003) at 17-37.
 Interference temperature differs from the permeable boundary between licensed systems and supercommons devices in that it is fixed ahead of time. A fixed boundary enhances certainty, but guarantees that the line will not always be drawn in the optimal way.
 A property owner should be able to choose whether to interfere with its own broadcast television reception, but choosing to interfere with police or rescue personnel in the event they must enter the property is another matter.
 See Joseph Mitola III, Cognitive Radio: Making Software Radios More Personal, IEEE Personal Communications, Vol. 6, No. 4, August 1999.
 On the other hand, many wireless devices already on the market can be upgraded through software. The radio and chipsets in an ordinary WiFi access point are physically capable of transmitting on adjacent licensed frequencies. The firmware in these devices is software-upgradeable, so users can install security patches and other enhancements. Experimental supercommons transmissions, albeit unauthorized, may be closer than we suspect.
 See supra not 188 (rejecting Sprint’s claim that UWB underlay is prohibited because its licenses are exclusive); FCC rejecting cellular arguments against competition. Cf. Proprietors of Charles River Bridge v. Proprietors of Warren Bridge, 36 U.S. (11 Pet.) 420 (1837) (holding that a government entitlement to operate a ferry does not protect against government later authorizing a competing bridge).
 See Benkler, Some Economics, supra note 4.
 Huber, Law and Disorder, supra note 78, at 75.
 Benkler explores this point, and the comparative economics of the two models, in detail. See Benkler, Some Economics, supra note 4.
 See Werbach, New Wireless Paradigm, supra note 41, at 9. The same is true for commons at other layers of the network. For example, the public domain in intellectual property is a resource that can contribute to private works, rather than a threat to them. Most of Disney’s animated movies use fairy tales and characters in the public domain. See Lessig, supra note 30.
 Coase understood this point well. For example, in his seminal spectrum article, he noted the coordination costs involved when there were many participants in transactions might overcome the benefits of a market: “When large numbers of people are involved, the argument for the institution of property rights is weakened and that for general regulations becomes stronger.” Coase, The FCC, supra note 17, at 29. See also De Vany et al, supra note 26, at 1509 (stating that exchange an enforcement costs will increase as the number of parties increases).
 Goodman in fact sees the commons position as essentially a claim that spectrum is like air. See Goodman, supra note 3. I would not go so far. Air, though intangible, is a concrete physical resource. Spectrum is not. I use the analogy in the text only to show how markets sometimes are not the proper answer to allocation problems.
 Coase used the difficulty of information gathering as an argument that a central administrative agency could not hope to allocate spectrum efficiently: “[I]t cannot, by the nature of things, be in possession of all the relevant information possessed by the managers of every business which uses or might use radio frequencies, to say nothing of the preferences of consumers for the various goods and services in the production of which radio frequencies could be used.” Coase, The FCC, supra note 17, at 18. Only the market as a whole possesses such information. A distributed price mechanism can therefore produce more optimal results than the best central regulator, under most circumstances. As Coase acknowledged, though, the market is not always superior. Transaction costs and in particular the coordination costs of many actors can overcome the benefits of a price mechanism. See id. At 18, 29.
 See F. A. Hayek, The Use of Knowledge in Society, Amer. Econ. Rev. 35 (Sept. 1945).
 See Stuart Kauffman, At Home in the Universe: The Search for the Laws of Self-Organization and Complexity (1995); John Holland, Emergence: From Chaos to Order (1998); Kevin Kelly, Out of Control: The Rise of Neo-Biological Civilization (1994). See also Arthur De Vany, “The Emergence and Evolution of Self-Organized Coalitions,” in Manfred Gilli, Editor Computational Economic Systems: Models, Methods and Econometrics (1996).
 Eric Bonabeau, Marco Dorigo & Guy Theraulaz, Swarm Intelligence: From Natural to Artificial Systems (1999).
 See Noam, supra note 2 (discussing similar kinds of failures in the auction process).
 Ronald Coase, The Nature of the Firm, 4 Economica 386 (1937); Yochai Benkler, Coase’s Penguin, or Linux and the Nature of the Firm, 112 Yale L.J. (2002).
 The difference with Coase’s analysis in The Nature of the Firm is that the non-market mechanisms in firms are top-down, hierarchical management, whereas the non-market mechanisms for wireless devices are decentralized, bottom-up signaling processes that do not rely on prices.
 See Hazlett, Wireless Craze, supra note 28.
 See Hazlett, Wireless Craze, supra note 28; Benjamin, supra note 98.
 See Benkler, Some Economics, supra note 4, at 42.
 For example, De Vany at al acknowledge that their property rights regime would push certain kinds of systems into particular points on the spectrum, which is exactly the inefficient “block allocation” structure today’s property advocates decry. Because of the way the De Vany group defines output rights, it produces greater zones of confusion among adjacent rights-holders at lower frequencies. De Vany et al propose to overcome this by having “transmitters serving large areas” in lower frequencies, and “transmitters serving small areas” in higher frequencies. See De Vany et al, supra note 26, at 1523. They do not specify how such an arrangement would arise. Presumably it would have to be built into the initial rights assignment, or add significant transaction costs for parties to trade their initial rights to achieve the efficient organization.
 Benkler, Consumers to Users, supra note 33; Yochai Benkler, The Battle Over the Institutional Ecology of the Digitally Networked Environment, 44(2) Communications of the ACM 84 (2000).
 It is worth noting that, for all their talk of privatization, property advocates generally acknowledge the continuing need for a regulator. See, e.g., White, supra note 28, at 21 (“There would still be a role for a national spectrum agency...”). White specifically suggests that this agency “could serve as a vehicle for encouraging the coordination on technical standards that is often desirable in network industries,” id. at 21-22. In other words, precisely the sort of “regulatory” standards coordination that Benkler proposes.
 See Peha, Wireless Communications, supra note 233.
 See supra note 187.
 See Faulhaber & Farber, supra note 3; Hazlett, Wireless Craze, supra note 28; Huber, supra note 78, at 75.
 A commons can, however, coexist in the same band with licensed services. The unlicensed devices can be required to operate without interfering with the primary licensed service. Such an approach could be used for example, as a transition mechanism in the 700 MHz band which is to be cleared as part of the digital TV transition. See Werbach, New Wireless Paradigm, supra note 41.
 Benkler, Some Economics, supra note 4, at n. 47; Benkler, Overcoming Agoraphobia, supra note 30, at 362-63.
 See Wireless LAN Equipment: Worldwide, 2001-2007 (Gartner Group, January 2003). Sales of wireless LAN chipsets increased from 7.9 million in 2001 to over 23 million in 2002. See Allied Business Intelligence, http://www.alliedworld.com/prhtml/wlic03pr.pdf.html. Instat predicts the WiFi hardware market will grow to $4 billion in 2004. See It’s Cheap and It Works: Wi-Fi Brings Wireless Networking to the Masses (Instat, December 2002).
 See supra TAN 176.
 CDMA is actually a spread-spectrum technology, developed by Qualcomm, that is predominantly employed in licensed bands.
 The FCC’s current ultra-wideband authorization restricts the technology to frequencies above 3 GHz, meaning that it cannot operate in PCS cellular bands.
 CDMA is not the only technology that controls power in this way. I use the example because the boundary between CDMA PCS systems and UWB devices has already been a topic of discussion in the FCC’s UWB proceeding. At some point the FCC is likely to consider expanding the available spectrum for UWB or other underlay devices, which will bring this conflict to the fore.
 See e.g., Revision of Part 15 of the Commission’s Rules Regarding Ultra-Wideband Transmission Systems, ET Docket 98-153, Comments of Boeing ( date) at 13-14.
 FCC statements of maximum # of channels. See e.g., Margaret Kriz, Supervising Scarcity, Nat’l J., July 7, 1990 at 1660 (“On April 13, the FCC handed out the last remaining substantial portion of prime radio waves....”) Similar are the periodic declarations that all the “prime” spectrum has now been assigned, and a choking “spectrum drought” looms. See Kennard, supra note 100.
 If the more robust receiver is substantially more expensive, users may opt for the cheaper alternative. However, this only reinforces the value of the commons. It is a case of market forces determining how good a receiver to build, taking all factors into account. A functioning commons optimizes the design of devices globally to balance efficiency against cost, rather than optimizing solely to meet the private goals of a licensee or spectrum owner.
 The FCC’s struggle with low-power FM radio is an example of how spectrum incumbents seek to block entry. Congress substantially rolled back the FCC’s proposal for non-interfering low-power FM stations, on the grounds that more low-power stations would cause major interference with high-power broadcasters. See Benjamin, supra note 190. Those fears proved to be overblown. See MITRE Technical Report, Experimental Measurements of the Third-Adjacent-Channel Impacts of Low-Power FM Stations, Volume 1: Final Report (May 2003), available at http://svartifoss2.fcc.gov/prod/ecfs/retrieve.cgi?native_or_pdf=pdf&id_document=6514285082 (finding no significant interference from low-power FM transmitters). In a property regime, the high-power broadcasters might be bale to sell low-power rights, but it is hard to imagine they would do so broadly. Incumbents do not have a good track record of voluntarily allowing in competitors, even when they are compensated. The general failure of the “unbundling” requirements of the 1996 Telecommunications Act shows this dynamic well.
 Though in theory a spectrum property owner might be able to enter into an underlay agreement the property regime would not provide the necessary tools to make such an agreement work. See supra note 185, 217.
 Coase, The FCC, supra note 17, at 29. Coase though such situations called for “special regulations” directly allocating spectrum uses, alongside the property rights mechanisms that would prevail elsewhere.
 See Kwerel & Williams, supra note 54.
 See Levin, supra note 26, at 94-95; Mueller, supra note 26, at 25 (“It is often difficult to monitor the actual output pattern of a transmitter without knowing the antenna heights and location, power, input, and transmission method.”)
 “This discussion should not be taken to imply that an administrative allocation of resources is inevitably worse than an allocation by means of the price mechanism. The operation of a market is not itself costless, and, if the costs of operating the market exceeded the costs of running the agency by a sufficiently large amount, we might be willing to acquiesce in the malallocation of resources resulting from the agency’s lack of knowledge, inflexibility, and exposure to political pressure.” Coase, The FCC, supra note 17, at 18. See also Coase, supra note 103; De Vany et al, supra note 26, at 1508 (acknowledging that the costs of negotiating and exchanging spectrum property rights may be quite high relative to their value).
 Bell Labs in fact first demonstrated the basic technical approach of cellular telephony in the 1940s. However, such a network would not have been commercially viable at that time.
 See Jon M. Peha, Spectrum Management Policy Options, IEEE Comms. Surveys (1998) at 6 (“It is therefore quite unlikely that a block of spectrum would emerge nationwide for cellular unless it was created by a single company.”)
 In the real world, it took Craig McCaw and other entrepreneurs years to string together national cellular telephony networks, even though the spectrum was already designated for that purpose and each geographic market had only two cellular operators.
 See Michael A. Heller, The Tragedy o f the Anticommons: Property in the Transition of Marx to Markets, 111 Harv. L. Rev. 621 (1998); Hunter, supra note 70. Benkler applies the anti- commons argument to wideband underlay. See Benkler, Some Economics, supra note 4. Benjamin takes issue with Benkler’s claim, asserting that if the problem is that owned spectrum bands are too narrow to accommodate ultra-wideband, the initial allocation can simply use wider bands. See Benjamin, supra note 98. This deus ex machina approach ignores the reality that whatever property rights system is adopted will have some initial license parameters based on foreseeable transactions at that point in time. Wider bands may allow for more wideband underlay, but future variants of UWB may call for even wider transmission bands. Furthermore, the wide bands of property rights would make other types of underlay more difficult, such as interweaving in guard bands between frequency blocks. Bigger blocks means fewer border areas between owned blocks. No matter how good the regulator is at assigning initial property rights, it will not be able to anticipate all the subsequent possibilities. It faces precisely the informational problem that Coase first identified as the fatal flaw in the current government licensing approach.
 Faulhaber and Farber call these “indirect transaction costs.” See Faulhaber & Farber, supra note 3. Benkler calls them “administrative costs.” See Benkler, Some Economics, supra note 4.
 See Faulhaber & Farber, supra note 3.
 See id.
 The fact that property rights can accommodate an easement doesn’t say anything about the choice of whether to designate frequency bands for exclusive or unlicensed use. That choice, as Benkler explains, depends on the relative economics of two competing industry structures and the network architectures they generate. See Benkler, Some Economics, supra note 4.
 See supra note 78.
 See supra note 91.
 If, as is likely, the licensed system operates at high power and the easement is for low-power devices, the high-power system will increase the computational complexity for the unlicensed network to find the efficient transmission path, thus increasing its costs. See Benkler, Some Economics, supra note 4, at 40
 See supra note 87; see Benkler, Some Economics, supra note 4, at 39.
 See Faulhaber & Farber, supra note 3.
 See supra note 65.
 See Faulhaber & Farber, supra note 3.
 See New America Foundation & Shared Spectrum Survey, supra note 101.
 See Benkler, Some Economics, supra note 4, at 55.
 See De Vany, supra note 218. Broadband broadcasting, which I refer to as wideband underlay, is an ironic term. Broadcasting is broad in its reach to a wide audience who receive the same programming, but it does so by sending a high-power signal through a relatively narrow wireless channel.
 De Vany, like other property advocates, was too sanguine that such subdivisions would occur. This is the private park scenario raised by Faulhaber and Farber, which I address above.
 Strictly speaking, subdivision of property rights is only partially an open-boundary form of communication. The subdivision can be no larger than the bounded property right itself. This is another case of the inherent tradeoff in an exclusive-rights regime: some activities must be prohibited through correlative duties as the price of assigning a right.
 private microwave reallocation.
 If the government paid off the aggrieved users of the band, the result would be similar to a buyback or compensated taking of spectrum that had been turned into private property.
 This is not to say that normative and economic values cannot be connected. For example, Tom Hazlett argues that property rights, by better aligning economic incentives, will better serve First Amendment interests than the current governmentally managed approach. See Thomas Hazlett, Physical Scarcity, Rent-Seeking and the First Amendment, 97 Columbia Law Review 905 (May 1997). Coming from the other direction, Benkler constructs an economic argument for commons approaches, based on comparative institutional patterns of information collection and preference articulation. See Benkler, Overcoming Agoraphobia, supra note 30, at 76.
 Coase, supra note 25, at 32-33.
 See Patrick Gelsinger, Intel Keynote Transcript, Intel Developer Forum, San Jose, CA, interview with Bob Metcalfe (Spring 2003), at http://www.intel.com/pressroom/archive/speeches/gelsinger20030221.htm.