The real question is really "why isn't most of the spectrum like this?" - people sharing it rather than it being sold off to private holders to be owned
To understand the thinking find a copy of the 1959 paper "The Federal Communications Commission" by Ronald Coase. See https://www.jstor.org/stable/724927 (pay-walled). It's the thesis behind the modern property-rights spectrum regime, and has bled over into many other federal regulatory regimes.
As interpreted by many modern, mostly conservative, regulators and scholars, the idea is that by giving private property rights in spectrum you incentivize not only more efficient usage through price signaling, but you incentivize research & development into more efficient technologies.
There's plenty to critique, especially as it applies to spectrum. See, for example, the famous Salon.com article summarizing David Reed's arguments: "The myth of interference" (2003), https://www.salon.com/2003/03/12/spectrum/ But if you want to understand your adversaries' thinking, Coase's paper is the place to start.
> incentivize not only more efficient usage through price signaling, but you incentivize research & development into more efficient technologies.
As with software patents, I think there's an argument to be made that this is no longer true. Digital modulation schemes are leagues more efficient than older analog techniques, both in terms of power and bandwidth. Spread spectrum schemes in particular are even reasonably resistant to being plopped on top of each other.
The spectrum ownership scheme allows established players to sit around on their asses using loud, inefficient modulation schemes with huge dead zones in between channels because they don't need to do any better than that.
I think radio technology would be better across the board if spectrum were open, and if people had to compete by using better, more efficient, more interference-immune modulation schemes, or risk getting stepped on by someone else. If we had a system like this, the noise floor would just slowly increase in heavily used parts of the spectrum, rather than what we have now, which is a whole bunch of legally mandated quiet zones punctuated by huge narrow carrier spikes.
The technology to do this exists... you can go buy a broadband SDR and put together a system like this in GNU radio fairly quickly for experimental purposes, but it's not really legal to actually use this anywhere.
The thing is, that the QoS and power efficiency of e.g. LTE isn't attainable if you have uncontrolled interference. It's a general issue with far-side crosstalk.
Interference is largely a property of the specific scheme. 5G schemes are better than LTE when comparing apples-to-apples (e.g. aggregate bandwidth on the same bands in the same geographic area), whatever comes after 5G will continue moving the ball forward, and the theoretical limitations on how much further we can keep kicking the ball are still out of sight.
Thus the question isn't what regulatory regime is best for squeezing the most out of LTE, but rather what's the optimal pace of radio communications evolution and how to achieve it. That's a drastically different type of question that almost certainly has a drastically different answer, or at least a much more complex answer.
The ideal spectrum allocation scheme would be a by-the-minute rental.
Whenever anyone wants to use spectrum, their device automatically bids and gets allocated whatever size slice of spectrum and transmit power they ask for. Then the device maker or the user pays the bill at the end of the year, perhaps with some minimum free amount each year per person, which could cover all except the biggest spectrum users.
If you didn't like the government allocation scheme, you could rent your own block and do your own sub-allocations, for example some kind of packetized system with microsecond long packets wouldn't want to have to contact a government server for every 1 microsecond allocation.
That gives a strong incentive to use efficient allocation schemes.
An "ideal" scheme is once again derailed by reality, where predictability is valued; where antennas and signal processing equipment are physical objects, that cost money to procure and install, and are subject to engineering limitations which limit their flexibility.
Remember, the free market is only efficient when transaction costs are low.
As a quick example, I’ve been doing a bit of work with land-mobile radios lately (LMR). We rent a very small slice of spectrum from the government (12.5kHz wide, IIRC), and all of the radios use that frequency. They work anywhere (within the range of our license anyway). They run independent of a repeater or any kind of fixed base station.
If I’m out in the middle of nowhere with my team and we’re using these radios because there’s no cell coverage, a) how would a given radio “bid” on a piece of spectrum, and b) how would the other units in the talk group know to listen to a particular frequency for my message?
That's a neat problem. You may hear N other stations, but they may not hear each other at all. Its very difficult to know whom to tell about coordination information, or even how to calculate it. Solve that, even heuristically, and make a $B
Coase actually argues in that paper, in true UChicago fashion that the spectrum should be completely unregulated in a free market for private efficiency. He’s actually arguing against the FCC. As he also touches on in The Nature of a Firm. Read Section 5 of the paper you’ve linked.
Yes. But the explicit premise in that section is that there are exclusionary property rights in the spectrum protected by traditional property and contract law, and that contiguous ranges of spectrum can be subdivided and sold exactly analogously to a piece of land. For example, he explains that while there are technical interference issues at the edges of spectrum bands whereby the user of one range could incidentally interfere with the user of another range, the same type of problems occurs with land (e.g. nuisance) and are similarly manageable with a traditional property rights regime.
In Coase's view the government shouldn't be auctioning spectrum for limited times or with strings attached, as we commonly do now. (What the FCC does now is a quasi-Coasean spectrum regime.) In Coase's view spectrum usage would be most efficient by selling all spectrum (from 0-infinity) in a one-time auction to the single highest bidder. That bidder could (and, crucially, would in Coase's view) then subdivide and sell the spectrum according to market demand.
Coase's logic is solid; it's the premises that are problematic. First, it ignores transaction costs, or at least assumes they're de minimis. Even before the Law & Economics field took off (which is the field Coase most influenced), it was understood that one of the purposes of legal rules was to minimize transaction costs. Sometimes a market is made more efficient by specifying a fixed rule, and even more often by specifying a fixed default rule, then by permitting perfect contractual freedom. Of course, this begs the question of when you can know whether to intervene. But that's the only legitimate question, not whether you should ever intervene. In Coase's time his argument was more practical as widespread independent and even individual broadcast usage wasn't a commercially realistic scenario; transaction costs would always be born by economic behemonths and therefore would be de minimis relative to their price. But that's not Coase's actual argument; his actual argument is categorical and absolute, and that's how most modern proponents (IME) apply it as well.
Secondly, the notion of spectrum rivalrousness is contentious. If something isn't rivalrous, a property regime is unnecessary and usually strictly inefficient.[1] Without getting into the science (and I'm not an electrical engineer so I'd just make mince of things), as a practical matter spectrum usage is only partly rivalrous (rivalrous mostly deriving from the technical environment, not from the fundamental character of spectrum, as with physical goods and real estate). Your ability to enjoy maximal channel bandwidth between two physical points is limited by how sophisticated your transmitter and receiver, and ultimately by computational power, not by coincidental usage, per se. (Not unless someone is actively trying to screw you.) As WiFi has born out, we're all in an arms race for buying and using the latest, greatest gear. In one sense this is just a convoluted price signaling scheme--the people most in need of bandwidth just pay for fancier equipment--that might be less efficient than a simple property regime. But arguably in practice this process plays out slowly enough (especially with moderate power limits, carve outs for emergency use, etc) that these transaction costs are far less than the transaction costs we see in a market where we expect Verizon's or AT&T's spectrum usage rights to trickle down to ad hoc groups and individuals of spectrum users in the same way William the Conqueror's lands in England were slowly subdivided until every Englishman could build his own castle.
Nobody will ever sell me rights to spectrum for use within a 10-meter or even 100-meter radius, even though it's possible and would open up a tremendous wealth of opportunities. Cordless phones (POTS, not cell) and WiFi could only have happened with unlicensed (i.e. unowned) spectrum, in contradiction to Coase's argument.
[1] Exception--if food wasn't rivalrous because we all had free Star Trek-style replicators I imagine obesity would be an even worse problem than it is today, we'd be more sickly, and so we'd probably want to impose costs to maximize aggregate wealth. (See soda tax.) And if we did that we might want to make those costs transferable (e.g. carbon credits) rather than centrally figuring out the most efficient price for each individual.
That’s a mistaken view. Coase argues precisely that the market would self-regulate by broadcasters determining the spectrum interferences with each other and drafting financial agreements to ameliorate those interferences. Not by having an initial legal monopoly, which was abhorrent to the Chicago School. He believes that the natural institution of private property in the sense of contract through the traditional court system is enough of a regulation and no special regulation (e.g. FCC) is needed.
Section 5 sophisticatedly exposits the specific case on how the free market solves frequency interference in admonishment of the special regulation. Excerpts of his discourse:
“All property rights interfere with the ability of people to use resources.What has to be insured is that the gain from interference more than
offsets the harm it produces. There is no reason to suppose that the optimum situation is one in which there is no interference”
“The operator whose signals were interfered with, if he had the right to stop such interference,would be willing to forego this right if he were paid more than the amount by which the value of his service was decreased by this interference or the costs which he would have to incur to offset it.” He goes on to talk about the case of deliberate interference. This is about deliberate spectrum preference overlap not accidental interference.
And Coase disambiguates his advocation if it still isn’t clear: “What this analysis demonstrates,so far as the radio industry is concerned, is that there is no analytical difference between the problem of interference between operators on a single frequency and that of interference between operators on adjacent frequencies.”
“It has been the burden of this article to
show that the problems posed by the broadcasting industry do not call for any fundamental changes in the legal and economic arrangements which serve other industries.”
You’re mistaking market efficiency, transactional costs and externality. It’s argued that legal rules are necessary to instead reduce externalities by market failures, which existed long before Coase. Coase instead contributed to the limits of cost savings about transaction costs provided by the formation of business and maintenance of markets in The Nature of the Firm (aside from his free market advocation of the spectrum), which became Coase’s theorem.
I’m educated as an Electrical Engineer; it’s clear to us spectrum rivalry isn’t contentious but true. Even amongst a single broadcaster within his own spectrum he runs into constant constraint and trade off by his own bandwidth.
I’m not arguing for or against the ideas of Coase, only correcting that public presentation of his ideas as an interventionist. Especially since Coase didn’t think externalities were unfortunate, instead he thought that was a “dirty word”, and seldom found a case for a government all to eager to specially regulate.
> I’m educated as an Electrical Engineer; it’s clear to us spectrum rivalry isn’t contentious but true.
But it's rivalrous in a distinctive way--it's a function of contemporaneous and evolving limitations of the existing state of technology and usage patterns. To quote a famous IEEE Spectrum article, "Moreover, interference is not some inherent property of spectrum. It's a property of devices. A better receiver will pick up a transmission where an earlier one heard only static. Whether a new radio system "interferes" with existing ones is entirely dependent on the equipment involved. Consequently, the extent to which there appears to be a spectrum shortage largely depends not on how many frequencies are available but on the technologies that can be deployed. Many regulations intended to promote harmony of the airwaves have instead, by putting artificial limits on technology, created massive inefficiency in spectrum utilization." https://spectrum.ieee.org/telecom/wireless/the-end-of-spectr... What that's referring to are techniques like MIMO, beamforming, beam steering, etc.
As David Reed is quoted and paraphrased as saying in that Salon article: '[We have stupid radios not because we haven't figured out how to make them smart but because there's been little reason to make them smart. They're designed to expect signal to be whatever comes in on a particular frequency, and noise to be everything on other frequencies.] 'The problem is more complex than just making smart radios, because some of the techniques for un-confusing the receiver are best implemented at the transmitter, or in a network of cooperating transmitters and receivers. It's not simply the radios. It's the systems architecture, stupid!' https://www.salon.com/2003/03/12/spectrum/
We're only beginning to scratch the service of the channel discernment improvements that can had with MIMO, beamforming, beam steering, and other techniques, but we're limited by the larger ecosystem of devices and regulations, which is largely an artifact of a system which assumes a fixed interference burden across vast geographical areas. The channel capacity of a band of spectrum is, in the absence of coherent interference, point-to-point as a theoretical matter because non-coherent electromagnetic radiation does not interfere. To reiterate, interference as typically experienced is a function of the inability to differentiate a signal which is present or which could be present if the transmitting and receiving systems were more sophisticated, and (according to the theoretical work) this is ultimately a computational problem.[1] Thus "spectrum scarcity", at least in the context of P2P wireless networks, is a function of technology, not an intrinsic, fixed property.
Taking that context into account--that interference and thus the degree of rivalrousness for any band of spectrum is a complex, dynamic function that dissipates as technology progresses--we can end up drawing different conclusions than from the technological presumptions behind Coase's argument. For example, as someone else mentioned, maybe we should be promoting technological development by letting people freely interfere, which can be done by imposing no rules whatever, not even a property regime. Absent power limits, that's pretty much what we do with WiFi. Similarly, a lot of theoretical radio techniques are first explored and applied by the military precisely because of the adversarial context, and they're consistently fruitful because the physical limits of what's possible are still quite distant.
> I’m not arguing for or against the ideas of Coase, only correcting that public presentation of his ideas as an interventionist.
I never had the sense that Coase was characterized as interventionist, at least not in the context of FCC reforms. If refuting that is your goal, fair enough. IME, however, in popular discourse anything "free market" is generally considered non-interventionist, and as compared to FCC regulations contemporaneous with his paper that's a fair characterization. But from my perspective (and AFAIU of many others, especially in the SDR community), establishing a private property regime is still rather interventionist, no less than it is with copyrights and patents, relative to the baseline of nothing. Nobody claims taxi medallion systems are non-interventionist, even when medallions are a freely transferable property right (e.g. NYC, sub-class of medallions in SF, etc). Uber would certainly disagree.
> But it's rivalrous in a distinctive way--it's a function of contemporaneous and evolving limitations of the existing state of technology and usage patterns.
> Whether a new radio system "interferes" with existing ones is entirely dependent on the equipment involved.
With this definition then there is no other type of rivalry. This is true for any resource with tragedy of the commons, where an externality is encountered. This goes for land, agriculture, pollution, oil, etc. The limit here is always the efficient production from some resource using technology. This is part of the ceteris paribus of rivalry.
For what you're discussing, for non-interfering waves with the same spectrum, you would need a QAM or some other modulation scheme, with more granular phase detection, which requires exponential more energy, thus capital, per degree resolution of phase detection at the same distance. Thus, the rivalry always exists because enterprises prefer the method leading to the cheapest production with the same product.
> maybe we should be promoting technological development by letting people freely interfere, which can be done by imposing no rules whatever, not even a property regime. Absent power limits, that's pretty much what we do with WiFi.
As I'll continue discussing in your other comment this is exactly what Coase advocated for.
> establishing a private property regime is still rather interventionist
It is in the sense you're discussing with the FCC but not in the Coase's sense where private property is in the strict abstract sense by enforcement of contracts between consenting individuals. Coase's point is that it shouldn't be specially regulated differently than the enforcement of any other private contract. Unless for specific cases which can only be known through experience
> Not by having an initial legal monopoly, which was abhorrent to the Chicago School.
In the Chicago School monopolies are impossible; or rather, monopoly pricing is a fiction.
I can't find a direct top-down subdivision quote (my copy isn't searchable; he might not literally say it at all[1]), but efficiency of the single initial auction winner scenario can be logically inferred from what he explicitly says. For example, on pp 20-21 he says that in a property regime broadcasters would be able to assemble larger bands through free market purchases, subject only to their ability and desire to pay. Efficient aggregation and subdivision of spectrum is a built-in presumption.
AFAIU, applying Chicago School principles, the price at which a single owner of all spectrum would subdivide and sell a portion would have no (or at least de minimis) monopoly profit component. Another way to look at it is that ownership of any contiguous range of spectrum, just as with a contiguous piece of land, is by definition monopoly ownership. Breaking contiguousness might require a premium, but only proportional to the economic value from contiguous usage. Breaking contiguousness of all spectrum would require a premium proportional to the marginal additional usage value, which for all spectrum is pretty much nil. What usage of all spectrum has a value greater than the aggregated value of using two, smaller portions? More likely there's greater value in the latter than the former. At some point subdividing becomes inefficient, but certainly at that point we're no longer dealing with a single, contiguous band of spectrum. The initial, profit-maximizing owner has already subdivided most it, and probably no longer owns any of it--our hypothetical bid winner was probably a specialized broker banking on extracting value by subdividing it.
Of course, judging by your comment, we both believe that monopoly profits arereal. Minimally they're approximately equal to the transaction costs required to buyout or break the monopoly, which is approximately equal to the cooperation costs for victimized buyers, which in practice would be born by a prospecting broker. (What we intuitively think of as "monopoly profit" are the opportunity costs to buyers having to pay a premium over the price paid in a competitive market. But we don't include those costs in the buyout/breakout scenario because they're recovered after the fact in the hypothetical and cancel out.) AFAIU, according to the Chicago School these transaction costs are de minimis, especially with large, liquid financial markets. This is why they're such fans of financial market deregulation, derivatives markets, short sellers, private equity, etc. But many people, including some Chicago School luminaries after 2008 (e.g. Greenspan), aren't so sanguine about those costs.
And this explains why AT&T and Verizon don't behave like the efficient economic actors in Coase's world--they won't subdivide or sublease their spectrum down to the levels required for WiFi-like usages because the costs of all of us cooperating to collectively bid is too high. Actually, that's not strictly true--the costs of cooperating can be trivial if we move beyond simple market theory. All we need to do is democratically choose, directly or indirectly, for the FCC to expand unlicensed spectrum, marginally rolling back a regime we previously democratically created.
Yes, I get that Coase in particular was appreciative of transaction costs in other contexts, such as in the theory of the firm. But Coase, and people in general, often make implicit assumptions about relative quantitive costs that become assumptions about qualitative costs that in turn become judgments about whether transaction costs are substantive. The transactional costs of subdividing and transferring spectrum among TV, radio, military, and other large, traditional broadcasters undoubtedly looked de minims in 1959. Certainly they were lower than the costs of FCC central planning, and that much I can't disagree with. But 40 years later the environment had completely changed.
[1] I distinctly remember seeing illustrations of this arguing-to-the-extreme concept, but the illustration itself was probably from another author as a gloss on Coase's work.
It can instead be logically inferred that it must not be what he's arguing.
Coase: “Or, alternatively, if this operator had the right to cause interference, he would be willing to desist if he were paid more than the costs of suppressing the interference or the decrease in the value of the service he could provide if interference were barred. And the operator whose signals were interfered with would be willing to pay to stop this interference an amount up to the decrease in the value of his service which it causes or the costs he has to incur to offset the interference.”
Why would the operator have to pay to stop this interference if he owns that spectrum? If I go onto your land do you have to pay me to get me off of it? No, you have legal protection protecting that property right against non-consensual trespass.
Similarly, we can then conclude the operator must not have a property right of that spectrum if he must pay to prevent trespass, otherwise the operator would be able to file a criminal injunction to prevent the interfering individual from continuing to disrupt his signal at little cost--as you would do for a land trespass.
You're confusing the specific bill he mentions in the preceding section about the legal etiology of the FCC, which had proposed to do exactly what you're mistaking as the argument of Coase; where Coase thinks only that that specific bill shouldn't have been dismissed so trivially and as a resolve to the current situation where the FCC has already sold partial licenses--but leaves it thoroughly short of an advocation as he develops the case of no special regulation for broadcasting rights.
I think that in practice the idea of privately owned spectrum really came into its own with the original scheme used for analog cell phone service. The incumbent phone companies were given their own exclusive spectrum and the competitive cell phone company (originally there was only one) got their own exclusive spectrum. This A band, B band system actually worked fairly good compared to the sort of thing happening with the landline monopolies.
Obviously the system couldn't be expanded to a market with more than two entities without causing a lot of inefficiency. The regulators did this expansion anyway and that led to the insanity we have today.
As an operator of large wireless networks I prefer to purchase licensed spectrum over using shared bands every single time. It's hard to overstate how much more reliable and performant the network can be when there is some assurance you won't be interfered with.
I would concur, but there are not many non-licensed bands. So all sorts of traffic is more congested in those. If there were more unlicensed bands I think it would probably improve those use cases.
To an extent, but that entire band is only 75MHz wide. There are only 2-4 proper channels, and it has twice the range of 5-6GHz, which puts you in competition with an order of magnitude more hot spots.
The microwave at my house would reliably murder my wi-fi connection whenever you turned it on, which was great fun while playing real-time multiplayer games.
This is one of the things I have trouble understanding too. It’s a complex issue, but it seems like it would be best opened to the public where next generation radio technologies can evolve. I don’t think 5G/LTE will go away but there may be new technologies which are more open and integrate within the existing frameworks. I imagine some kind of public mesh net/CDN that goes out to LTE/5G only when necessary to access information not in the local context.
I think fears are overstated and used as a pretense to sell off spectrum to private entities, which can provide private investors a handsome reward for their monopoly on the spectrum. Consider "junk" unlicensed WiFi spectrum (i.e. 2.4ghz). Normally private owners would argue, "We can't allow anyone to broadcast, it'll be overcrowded. We need a gatekeeper to solve this." In reality, yes it can be overcrowded in dense areas, so packets still get sent, but at a lower rate. This creates demand and people start gravitating to less crowded, unlicensed spectrum (i.e. 5.2ghz) for crowded indoor use scenario. It's more bandwidth and better attenuation with your neighbors. Consumer radios are also certified to only produce so much power on the spectrum which adds additional safeguards. So the free market and a dash of radio certification/regulation ends up providing an efficient decentralized communication medium.
With other shared mediums, we police the spectrum (e.g. light, sound, etc...). If a user starts blasting out FM, or too much 2.4gz and so on, it can be detected, traced and prosecuted. It's 2020, we don't need FM, we need high speed IPv6 radios and improved routing tables/algorithms. Lift FM channels into the digital realm and let's get on with it.
The reason why WiFi has poor performance in dense environments is because there is no central coordinator for spectrum usage. Stations pick a random frequency and start talking. If another station is talking when you want to send a message, your station will wait indefinitely for the channel to be free. The result is that you get random latency spikes; your VoIP call just gets dropped, you get disconnected from your online game, your video call freezes up. There is no concept of determining what's best for everyone; anyone sending a message checks if the channel is clear and goes about their business.
This all sounds somewhat fair, but the problems really start when not all the stations can hear each other. Station A can hear Station B and Station C. Station C starts sending a message to station A. Station B can't hear Station C, so starts sending a message to station A. The result is that both messages are lost and the effective channel capacity goes to zero. This is obviously a big problem.
The fundamental problem is that radio is very much half-duplex. Only one station can use a certain amount of spectrum at a time; if someone is using it, you can't. It causes problems everywhere; with WiFi, with air traffic control, etc.
Centralized frequency allocations get around this with better protocols. All the stations know about each other, and can be told when they can transmit by a centralized coordinator (TDMA is an example of such a protocol). As a result, each station has a predictable amount of airtime, and therefore guaranteed bandwidth and latency; no time is wasted with two stations talking over each other. (There are other sharing schemes that are more interesting and popular than naive time-sharing; CDMA, spread-spectrum, and GPS's "Gold Codes" are all interesting reading.)
The easiest way to implement this is not to write a standard and then ensure that everyone buys a standards-compliant device, but rather to give chunks of spectrum to companies, and have companies distribute devices that work they way they want. It's easier for the regulatory agency ("pay us money and do whatever you want"), and doesn't cap innovation. Verizon can collect your phone tomorrow and give you a new one if they want to use whizbang new technology. I don't think that's a bad thing, but it is of course a compromise. (Faster innovation, less government interference with technology... but at the cost of making it impossible to start a cell phone provider on your own.)
I guess my TL;DR is: it's not that simple. You could let everyone do whatever they wanted. That's how it was in the early days. People used spark-gap transmitters that generated so much noise that there was effectively only one half-duplex channel for the entire planet to use. Regulators saw that that wasn't going to scale and invented spectrum allocations, and here we are today. It's not perfect, but technology is improving how we used the unlicensed bands. (I am not as up to date as I want to be on WiFi specifically, but if you look at the amateur radio community, it's routine these days to communicate around the world with 5-10W of power. 20 years ago, it would have taken two orders of magnitude more power and 20x the bandwidth to do with voice what we do with FT8 today. People are still idiots, though, and generate plenty of QRM. It only gets worse when people without licenses start operating the radio stations.)
You already can't do whatever you want, even in ISM bands (... a WiFi AP with 5 kW transmit power would be really bad for everyone. It's also fairly illegal.) Having some rules forcing devices to cooperate (... perhaps more intelligently than WiFi currently does) would be welcome... but there is a difference betweeen "having to be nice and coordinate with each other" and "needing explicit permission from a centralized authority to do anything whatsoever".
Actually, the proposal already contains this: "An automated frequency coordination system would prevent standard power access points from operating where they could cause interference to incumbent services." We could absolutely have central authority assigning time slots in congested areas in a fair way, while staying decentralized everywhere else. A bit like the way air traffic control works.
But there are already protocols (DECT I'm looking at you) that design all this stuff in, that can happily share channels and do exactly this sort of thing
Well US DECT only has 5 channels (everywhere else it's at least twice as many)
DECT base stations are capable of detecting when two (or more) stations are on the same channel and sharing transmission slots between them.
With 10 channels and 12 full duplex slots/channel you can handle 120 people sitting on top of each other, spread them out a bit and then you can handle many more - probably a small number of hundreds if done right
