Communications Insights: Trends and Cool Stuff

Yesterday, Prof. Tim Wu of Columbia Law School published an Op-Ed in the New York Times on the subject of bandwidth.  In the article, he compares bandwidth to oil in terms of its percentage of the average household’s expenditures and in terms of the cartels which produce it.  He says:

Like energy, bandwidth is an essential economic input. You can’t run an engine without gas, or a cellphone without bandwidth. Both are also resources controlled by a tight group of producers, whether oil companies and Middle Eastern nations or communications companies like AT&T, Comcast and Vodafone. That’s why, as with energy, we need to develop alternative sources of bandwidth.

While Prof. Wu’s might be right in his conclusions, I have to take exception to some of the points he makes along the way, particularly regarding cost of bandwidth.

To begin with, the price of oil is based, to some large measure, on the cost of its production and not necessarily the cost of its consumption.  The cost of production includes the cost to pump the oil out of the ground, refine it, and distribute it.  The cost of consumption would include the societal cost of pollution such as global warming caused by greenhouse gasses.  Here in Europe where a gallon of gasoline exceeds $9, most of which is tax, the retail price may better reflect the cost to society not only production but consumption as well.  The cost of the production of bandwidth includes both network CapEx and OpEx.  The cost of its consumption includes the negative effects of congestion felt by competing would-be users at times of peak use.  It is effectively zero, when use is non-rivalrous.  The price of bandwidth, as well as other resources subject to high negative externalities, should reflect the cost its production and consumption.  This maximizes the benefits which society obtains from the resource.

This is precisely why, contrary to Prof. Wu’s assertions, the FCC is working on such ideas.  FCC: OSP Working Papers #41, #42, and #43, on which I am a proud coauthor/collaborator, look at precisely these issues.  We designed and tested a system which instead of assigning spectrum in static blocks, would co-ordinate use of the spectrum to an efficient optimum.  Beyond the overly simplistic bandwidth dipstick, the FCC work also modeled other dimensions of performance, such as latency, and could be extended to include jitter, reliability, robustness, etc.  Nonetheless, the economic congestion protocols we developed would allocate bandwidth in real-time based on willingness to pay when there is congestion and it would be free otherwise.

To be fair, a business model which relies solely on congestion-based prices for its economic logic would not be sustainable.  Imagine an airline which would allow its passengers to fly for free, unless more passengers show up than there are seats, in which case it will charge all the passengers based on their willingness to pay.  One would expect such an airline to have very small planes, or, more likely, very large seats in its planes.  Without a way to ensure rivalry among its passengers for its capacity (i.e., seats), such an airline would surely go out of business.  Thus, a sustainable price for bandwidth must reflect the cost of both production and consumption of the resource.

Insight: I continue to tout these papers in my blog because they are important, cutting-edge work.  We sought to lay the groundwork for a better system which incorporates the best of the licensed and unlicensed approaches to spectrum access.  This system would be, to use Eli Noam’s words, would be “open, but not necessarily free.” As such, it would maintain sufficiently low barriers to entry, which would make it sufficiently difficult to obtain monopoly rents.  It would be nice if Prof. Wu would put forward some of the available solutions to “alternative supplies of bandwidth” in addition to pointing out the problems.

For more than nine decades, lawyers, engineers, and economists have argued that radio spectrum regulation is needed due to the fact that without some form of intervention, it is impossible to exclude or limit the use of a common resource such as spectrum.  Without exclusion, users consume the spectrum without regard to their usage’s impact on the benefits obtained by other would-be users.  They, therefore, tend to overuse the spectrum, causing interference to other users.  This reduction in social welfare due to overuse is referred to as the Tragedy of the Commons.However, we can now observe from the debate surrounding the TV White Spaces that the ability to exclude certain users is not sufficient to remedy the Tragedy of the Commons. A relatively small number of over-the-air TV households are able to use these spectrum bands without regard to the costs their use imposes on the rest of Americans.  Indeed, according to the most recent FCC statistics, in 2005 only about 14% (See Appendix B, Table B-1) of US TV households receive their TV over-the-air. The remaining 86% get no direct benefit from this spectrum.

The National Association of Broadcasters is now opposing tests the FCC is currently conducting which will measure the impact of unlicensed use of the White Spaces on digital TV reception. In order to protect digital TV receivers, potential White Space users must be excluded, and the NAB is throwing its weight around to ensure that outcome.  According to a quote from NAB spokesman Dennis Wharton, “We’re not going to be engaging in threats or anything, but about 70 members of Congress have already sent letters in expressing concern.” Well, as I wrote in a previous entry on Cool Stuff, at least one of those 70 letters is total bunk. Nonetheless, the cost to all of society of affording interference protection to this minority must also be considered.

Insight: If the NAB’s argument is accepted without scrutiny, the 14% of TV households will prevent the other 86% of US TV households (plus the TV-less households) from using those radio frequencies for broadband Internet, baby monitors, new forms of low-power broadcast, and other RCS (really cool stuff).  This lost benefit will not be compensated.  The exclusion of certain competing uses is necessary but not sufficient to ensure that society reaps the maximum benefit from the radio spectrum.  A means through which spectrum users can bear the costs they impose on others by excluding them is also necessary.

My article Unlicensed to Kill: a Brief History of the FCC Part 15 Rules has just gone to press and will be published in the journal Info.  I originally gave the paper at The Genesis of Unlicensed Wireless Policy conference organized by Tom Hazlett at George Mason University Law School.  (Yes, dear reader, Tom Hazlett hosted a conference on unlicensed.  The Seventh Seal is broken and the End of Days is truly upon us.)

The conference looked at the origin and evolution of the FCC’s Part 15 rules.  There were several interesting takeaways.  Most of these are lessons which we already know, but all too often take for granted.  Keynote speaker Michael Marcus reminded the audience that people frequently act in their short term interest, in a way in which they foreclose long-term opportunities for themselves.  Dr. Marcus described the regulatory battles of the 1980s during the FCC’s rulemakings where cordless manufacturers fought fiercely, opposing certain rule changes.  These rules now enable most of the cordless phones these manufacturers now sell.  The closing keynote, Dewayne Hendrix pointed out how spectrum policy the cognitive dissidence spectrum policy faces in affording interference rights.  We allow licensees to “whine” about interference when they use decades old technologies which do not have the ability to reject unwanted signals which more modern gear does.

Insight: In the US, there is no such thing as unlicensed spectrum.  Rather, and this is an important distinction, the FCC allows low power operation on a sufferance basis, proved the devices cause no harmful interference and accept all received by them.  Operators have a right, but not a vested right to continued operation.  The FCC has historically viewed the radio energy emission from these devices as not rising to a level sufficient to call “spectrum”.  This has left me wondering if there is no such thing as spectrum at all.  Spectrum is a legal and engineering construct to control for an immutable fundamental physical property.  When multiple electromagnetic waves, used as carrier waves to transmit information are incident in time, harmonic in frequency, and alight on the same reception antenna, they degrade one another’s ability to transmit information.  Next generation radio policy will focus more on solving the coordination/congestion problem, and not on “spectrum” per se.  (I also gave a really cool PowerPoint.  (Click to start, click to advance each slide after the animation stops).

In a previous Cool Stuff, I wrote about the study which I completed demonstrating the social value from reallocating some of the Digital Dividend spectrum for broadband mission critical public safety communications. The European Parliament seems to agree.

Yesterday, the European Parliament’s Industry Committee adopted a report urging that the EU should ensure a set of harmonized, EU-wide rules on how to allocate radio frequencies that will be freed up when analog terrestrial television broadcasting ends in 2012. The report was an own-initiative report authored by Italian liberal MEP, Patrizia Toia and was adopted in Committee with 41 votes in favor, 1 against, and 1 abstention. A plenary vote is scheduled for September. Further, the amendments to the report argue that approximately 100 MHz of the Digital Dividend could be reallocated to mobile broadband and other services such as public safety services, radio frequency identification (RFID), and road safety applications, without preventing broadcasting services from flourishing.

Insight:  While the transition from analog to digital terrestrial television should be complete in Europe by the end of 2012 (nearly 4 years after the U.S. is scheduled to complete its transition), decisions on how to reallocate the approximately 75% of the highest quality spectrum which will be released cannot come fast enough. Mission critical broadband communications networks require long lead-times to plan and deploy, and the services they enable are nothing short of lifesaving.  Public safety and security users urgently need an additional allocation of a approximately 30 MHz for these purposes.  The Industry Committee correctly urges Member States to release their Digital Dividend spectrum as quickly as possible, follow a common methodology, and develop national Digital Dividend strategies by the end of 2009.

I have been listening to a bunch of excellent presentations for the first two days of Supernova2008. Rather than rehashing what each speaker has said, I have been trying to formulate a theme. Not an easy task. I have noticed a few reoccurring themes: social activity, intellectual property, management of information, and marketing; all good network-related themes. I spent most of the second day at the Open Flow Track. Much of what was discussed was is integrating systems: Connecting the connections. That is to say that the internet has provided connectivity and access to persons and applications. The essence of Web 3.0 is making sure that your Flickr works with your Dopplr, with your, dare I say, Napstr.

Insight: The rich and lively discussion in the Open Flow Track seemed to focus more on engineering and business practice questions in terms of getting APIs to work together and making sure that privacy, security, and trust are respected according to applicable law and good business practices. I still found myself searching for more a fundamental concept. A more fundamental question which was present but perhaps not fully articulated was how to describe this continuum of “openness” vs. “closedness” (not a real word). So, here I get to like to wax poetic for a second. Eric Raymond, a pioneer of Linux and the open source movement, gives us a particularly literary book title and syllogism, “The Cathedral and the Bazaar”. Raymond sees the cathedral as representing a system of architecture which is, “carefully crafted by individual wizards or small bands of mages working in splendid isolation - no beta version.” It is a centralized, coordinated approach. Open source architecture he likens to “a babbling bazaar of different agendas and approaches.” It is decentralized with varying standards and rules, but is not anarchy. Both approaches seem to work in creating stable systems, though they may be suit to different types of applications.

It was widely agreed that there should be a general preference for openness. I agree, but to my mind that there is a choice between openness and closedness. This choice implies a tradeoff. And, if there is a tradeoff, there is by necessity some optimization. What the optimum is will depend largely on your point of view and social optimum does not necessarily equate one-to-one with a private optimum. At the very least we can have a rational discussion as to what the relative merits of the tradeoff are and where the different optima may lie. In sum, do we want a world that looks more like the Cathedra or the Bazaar, or is there an entirely new form of architecture that we should consider?

I finally just had a chance to read the FNPRM for the rules governing the C Block in FCC’s recently concluded 700 MHz Auction and it says “handsets”! Why is this significant?

For the C Block, comprising 22 MHz in the upper 700 MHz band, the FCC created special open access provisions. The FCC will require licensees to provide a platform that is open to third party devices and applications. Specifically, licensees must allow customers, device manufacturers, third-party application developers, and others to use any device or application of their choice on their networks in this band, subject to certain limited conditions. Licensees may not “lock” handsets to prevent their transfer from one system to another, or to other services that compete with wireless service providers’ own offerings. The FCC concluded that these rules were justified because it did not find “that competition in the [mobile] marketplace is ensuring that consumers drive handset and application choices, especially in the emerging wireless broadband market…. it is easy for consumers to differentiate among providers by price, most consumers are unaware when carriers block or degrade applications and of the implications of such actions, thus making it difficult for providers to differentiate themselves on this score.”

Insight: Beware the law of unintended consequences. Here it is not the proverbial monkey wrench, but a pair of bolts, literally, which could bring the FCC’s policy to a grinding halt. Implicit in the service rules is the assumption that the band will evolve to resemble the next generation of the current mobile market in the US. But assumptions like this never last. I wonder if it is possible for clever operator could escape the open handset requirement by providing fixed services. The 700 MHz spectrum is particularly well suited to a variety of applications, one of which might be fixed broadband. In rural and suburban areas fixed wireless broadband could be an effective competitor to wireline. Presumably these areas would be sufficiently competitive that the FCC’s finding would not hold (remember it is limited to handsets, and not even service plans). Thus, if the licensee is bolting “pizza boxes” to the side of homes, would this type of network not be subject to the open access provisions? It’s unclear, but something to think about.

Anyone who knows me for 2.5 seconds knows: 1) that I cannot leave well enough alone and 2) that I cannot stand Microsoft operating systems.  So, I am curious to see what happens when you run network neutrality full speed into 18th Century copyright law.

Network neutrality is a broad, sweeping concept intended to maintain the open and interconnected characteristics of the Internet.  One of the central principles of network neutrality is the freedom to attach any “legal” device which does not harm the network and run any application over it.  I am not quite sure what an illegal network device is, perhaps a digital Kalashnikov (though if the Supreme Court reinterprets the Second Amendment, even that might not apply).

Enter the Hackintosh.  A Hackintosh is a PC which, with a few a modifications, can run the Mac OS X (called OSx86 for the IBM/Intel 8086, 386, 486, 586, etc., architecture). This is apparently not hard to do since Apple started using Intel chipsets. Moreover, the modifications are made to the bios and harddrive of the underlying computer, and not to the Mac OS. However, it is Apple’s contention that it is violation of its copyrights to run OS X on anything but one of their machines.

So, if a computer is attached to the Internet, would Apple’s prohibition violate the principle of network neutrality?  Well, it turns on whether the Hackintosh is a legal device. I am not an expert in intellectual property law anymore, but to my mind it would be illegal tying and bundling to require that OS X could only be run on Apple machines. Can you imagine if Sony sold me a record (vinyl or shellac) and in the liner notes stated that I do not have permission to play it on a Victrola, only a Sony turntable?

Next, let’s consider the DMCA.  The Digital Millennium Copyright Act makes it a crime to circumvent technical prevention measures (TPM) in order to illegally copy copyrighted electronic materials. Running OS X on a machine other than a Mac does not necessitate an illegal duplication.  Under the first sale doctrine, one can buy a valid copy of OS X Tiger on eBay for about $75.FN Apple cannot say that it is an illegal copy to put the OS on a computer - that’s purpose the software was sold for.  Further, they cannot restrict the device with which you read the OS, back to the Victorla… could GE prevent you from reading a book published by NBC Universal under a Sylvania light bulb?

So, let assume that Apple does not sell their operating system (but they do), rather they license a complete device, called a Macintosh. Presumably, then the operating system is part of a useful article, and not a writing. It would therefore not be copyrightable. So, Apple would not have any valid copyrights to be violated by duplication and modification of the OS.  Further, since duplicating it is not illegal, the DMCA does not apply. Loading OS X on a machine other than an Apple might be breach of contract, but not a copyright violation.

Insight: I am not sure that Hackintosh is a legal device, but it probably is not an illegal device. I am curious to hear the opinions of the network neutrality and copyright experts. So, in this high speed collision between the broad, sweeping principles of network neutrality or arcane copyright law which survives?

Google’s Larry Page spoke at a recent New American Foundation event, calling for “Wi-Fi on steroids” for the TV White Spaces.  Every time I hear this, I cannot help but think, “Oh great, a radio that is hyper-aggressive, muscle-bound, and impotent.  Why would I want such a technology?!” All joking aside, I approve the sentiment, but a little more careful analysis is need.There is here a unique window of opportunity to allow new uses of the TV spectrum which is currently inefficiently used.  For the past 9 decades, the FCC has regulated high power uses of the radio spectrum, such as broadcasting.  The FCC has also for the past 7 decades permitted low power uses, with increasing success.  The TV White Space presents the opportunity to permit medium power uses of the spectrum - something between Wi-Fi and TV.  However, neither the high-power of low-power paradigms seems to fit.  Licensed approaches typically allocate use to a single entity which makes decisions about use.  As a result, much remains unused at any given time.  The rules created are hard to change and do not afford much flexibility in terms of decisions regarding use by the licensee.  In contrast, unlicensed approaches strictly limit the radio energy which a device can radiate into the ether.  By controlling the emissions, the rules limit the possibility of harmful interference.  These rules create a much more flexible set of permission, but due to the stringent power limitations ranges of the radio devices can be extremely short.  What is needed is a new form of coordinating spectrum uses for medium power applications, which holds the benefits of both approaches while minimizing the potential downsides.

Insight:  Fortunately, some of the FCC’s best and brightest have been working this issue.  In a previous Cool Stuff, I wrote about my FCC Working Paper, which lays out ideas for the implementation of economic congestion etiquettes which would allocate spectrum use in real time to its highest monetary value uses.  This approach could significantly improve the value society receives from the use of the radio spectrum, without the need for dangerous pharmaceuticals.

There is no generally agreed upon definition for Network Neutrality and deviations or violations against the principal.  Network neutrality is something of a catch-all phrase that has come to reflect a number of potential behaviours that some have considered to be anticompetitive.Network neutrality implies that all IP packets should be treated more-or-less the equally, and the debate reflects concerns that they might not be in the future - that a network operator might somehow apply different treatment to IP packets (or datagrams) associated with different services, applications, destinations or devices.

Insight:  The term network neutrality is a loaded one. It implies that any deviation from its principals is not neutral, and in some way unfair. In many instances, it is perfectly fair to provide preferential treatment to some network uses, say premium customers or emergency services. Well, the highly-paid lobbyists have returned with their own term to redraw the chalk lines on the level playing field. In recent discussions involving the US FCC, the term network management practices has come to take the place of the term network neutrality.  While the semantics of this term are more objective, it does not accurately describe the nature of the problem.  Network management refers to a much broader scope of activities including business practices, sales and marketing, security, fault tolerance, and capacity planning. Since the issues present in this debate represent only a narrow slice of network management, what term should be used?  How about false and deceptive advertising or anticompetitive practices?  Somehow that lacks the sexiness of Net Neutrality.

Telecommunications network operators have traditionally charged on a unit basis for messages sent and received. Dating back to the heyday of telegraphy, operators charged by the word. With the rise of the Bell System telephony was charged on a per minute basis. The 1984 Divesture of AT&T saw the rise of two part tariffs, to separate costs related to access and long lines networks. However, the basic premise was a per minute charge (oftentimes including a connection or first minute charge). This arrangement was simple and allowed for easy comparisons of unit charges, dividing total revenues by total minutes of use (MoUs). This price per minute is an easily comparable metric - a voice minute has changed little since Alexander Graham Bell famously whispered, “Watson, come here I need you.” Regulators could employ price per minute performance as a gauge on the efficacy of policies, and consumers could understand the value of what they are getting.

However, now, driven by competition in mobile telephony, communications carriers have offered an alternative set of arrangements (the same is now true for fixed line carriers and even some data networks). These have been variant of the flat rate plan, mixing flat rate and per unit pricing plans. Under such “buckets of minutes” plans, customers pay a flat rate for a huge number of minutes whether they use them or not. If their usage is over the allotted amount, then they pay on a per minute basis at a rate much higher. The per-minute charges are not intended to be used on a routine basis; rather, they are set at high or punitive levels, so as to enforce the need to upgrade to the next higher band as the customer’s usage increases over time. These plans are most appropriately viewed as representing banded flat rate arrangements. There are different bands, representing different numbers of total MoUs per month. In the United States, these banded flat rate plans have been well accepted by consumers and industry. They probably track underlying costs of mobile telephone service reasonably well, to the extent that usage-based average incremental costs in the United States are primarily a function of air time. Customers appreciate the predictability and the relative simplicity of the plans. So long as the consumer does not exceed the maximum number of minutes in the band, the consumer will tend to think of the plan as being purely flat rate.

The introduction of bucket plans, however, frustrates any calculation of per minute unit price. Plans of this type usually incorporate a per-minute charge for minutes, but it would be a mistake to analyse a banded flat rate plan as if it were a simple two-part tariff. In calculating per minute charges it is unclear whether to utilize actual minutes of use consumed or plan minutes sold, but not used. The results would be quite different. The former would yield a higher price, but would not indicate the value “left on the table” by the consumer. While the latter would tend to under represent the actual cost by making implicit the value of the option of further minutes of use at zero incremental price.

Insight: The appropriate way to regard bucket plans, and hence, to value per minute unit prices, is as a permutation of a financial call option. A call option affords the right, but not the obligation of use at a specified price, called the exercise or strike price, on or before a specified date. The intuition begins with a graphical analysis, see inserted figures. The graphic on the right displays the payoff of a financial option at various prices. The x-axis represents the price of the underlying security and the y-axis represents the potential profit or loss. The value of the option stays flat until the market price exceeds the strike price. At this point, its value increases and becomes positive once it exceeds the cost of the acquiring the option. Now consider the figure on the left. It shows the cost of two bucket plans. The x-axis is still the independent variable: minutes of use. The value (or price) of the plan is flat until exceeds the flat-rate portion of minutes. It then increases at a linear per minute rate, until at some point it is preferable for the customer to take the larger bucket. These two graphics look virtually identical, suggesting that the option valuation method holds promise as an effective metric.

The value of options can be captured by a complicated formula developed by Fischer Black and Myron Sholes in the early 1970s (the two shared the Nobel Prize for this observation). The option value is based, inter alia, on the current market price, the strike price, the variability of market price with regard to the strike price, the risk free cost of capital, and the maturity of the option. By finding equivalents for these variables in bucket plans, we can easily derive per minute values for under bucket plans.  I realize that this may be a bit of an oranges-to-tangerines comparison. There are certain items not present in the derivatives markets in these service plans. Nonetheless, such a calculus would allow for straight forward comparisons of the price of services plans across firms, countries and over time series where such plans might not exist.