Dialog with Frontline Wireless

Last week I posted my concerns over Frontline’s proposal to the FCC for the rules governing the upcoming auction of the 700MHz “beachfront” spectrum which is scheduled to be freed up in February of 2009 when UHF TV goes all digital.  Frontline proposed that 10MHz out of the 60MHz scheduled for auction be sold as a single nationwide block subject to a special rule which would require the successful bidder to build out a public service network at its own cost and to wholesale this 10MHz to retail providers with a proviso that allows for public safety preemption in emergencies.

At my invitation, Frontline has responded to my concerns and followup questions via an email dialog which is reproduced verbatim below with only some pleasantries removed.  The Frontline participants are former FCC Chairman Reed Hundt who is Frontline Vice Chairman and Gerard Waldron of Covington and Burling who is counsel to Frontline and was Senior Counsel to the House Telecomm Committee at the time the law governing the auction was written.

Although I don’t agree with the logic off all their answers, I’m impressed that they took the time to respond thoughtfully and think they’ve helped advance an important discussion.

Concern #1: In an emergency, ordinary citizens need to be able to communicate with each other and with public safety.  If your mobile phone operator or ISP happens to be a customer of Frontline, then you’re out of luck and out of touch when trouble happens.  How do you call for help if your frequency’s been preempted?

Waldron: Giving public safety ("PS") preemption simply means their packets get to go first; it does not mean a loss of capacity to commercial users.  They may experience some additional latency, but giving PS priority would not shut down the system.

Concern #2: a mechanism for preempting frequencies in an emergency is inherently unsafe.  Suppose you’re a bad guy; you target the control center for preemption or YOU pretend TO BE the control center and do the preempting yourself.  That may be your attack.

Waldron: The proposal is not to preempt frequencies, but to manage the IP network to give certain packets priority.  This proposal is not new and it does not introduce new risks.  In fact, every IP network in operation today has the ability to give certain packets priorities.  In fact, the Pentagon today uses bit prioritization for mission-critical traffic in Iraq and around the globe.  Indeed, most businesses use some bit prioritization.  This approach is both safe and tested. 

Concern #3: Public safety is already slated to get an additional 12MHz of capacity when the UHF frequencies are reallocated in February of 2009.  There is no evidence that it needs any more although there’s plenty of evidence that public safety needs better communication than it has today.

Waldron: It is true that on most days, public safety does not need the full 12 MHz of additonal capacity.  That is why the Frontline proposal is so spectrally efficient:  when PS spectrum is not being used by public safety, it will be used by the public.  But during a time of emergency, they need the full 12 MHz and then some.  That is why the Frontline proposal would give them access to additional spectrum that could never be set aside just for them, but helps everyone when it is part of a national shared network.

Concern #4: Frontline proposes that the bidder be required to accomplish a “build-out of 75% of the population of the United States by the end of the fourth year, 95% by the end of the seventh year and 98% by the end of the tenth year.” It’s not only because I’m in Vermont that I wonder what happens to the uncovered parts of the population during this period.  Only Frontline can build to use the special 10MHz so it goes unused in rural areas for a long time to come.  Do public safety agencies in those areas wait patiently for “free” infrastructure to be built or do they meet the needs of their constituencies by building infrastructure needed only until Frontline shows up?

Waldron: Let me clarify two points about the Frontline proposal, as reflected in our latest Reply Comments.  First, we propose to get to 99% of the pop. in year 10.  Second, we make clear that if some PS entity wants to go ahead, they are welcome to and then that local network will be integrated into the national network.  So PS agencies do not need to wait patiently if they have the resources to move forward on their own.  If they do not, then they would have to wait, but let's be clear:  that is the same position they are in today, so they would be no worse off and in the long term will be much better off. 

Concern #4: The “build-out at no cost to taxpayers” IS at the cost of taxpayers; it’s just disguised.  Consider other potential bidders for the 10MHz under the proposed rules.  Each of them considers the revenue they’ll get from being able to wholesale the frequency when it’s not in use and the capital cost of the build-out.  They subtract the cost of the build-out from what they’re willing to pay for the spectrum.  The likely result is that the auction – which results in revenue to the federal government – nets less than it would have had there not been a build-out requirement.  There’s no one to make up the difference except us taxpayers; ipso facto, we pay for the build-out either way.

Hundt: First, if we as Americans want a national interoperable broadband network for public safety - which many do and some don't I suppose, but assume that we do -- then the incremental cost of providing that network on top of, so to speak, an existing commercial network is by far the lowest cost solution. Verizon says an independent public safety network could cost up to $62 billion over 10 years; our guess is the incremental spending by Frontline or a Frontline clone might be $1-2B. so that's about a $60b savings for the taxpayer, if you want such a network.

Second, as the auction revenue, it's not at all clear that an Open Public Safety network will attract fewer bidders willing to spend less money in the auction than, say, Verizon would bid. Verizon, one presumes, will bid for the spectrum something equal to the impact on Verizon if Frontline bought the spectrum, spent the $7B or so necessary to build it (maybe up to an additional $2B beyond that over time), and then used it not just for public safety but also to provision national roaming for Vz competitors. That impact would be a defensive calculation: what's it worth to Vz to keep Frontline from existing. Vz has no announced or easily inferred plan actually to use this spectrum; they have plenty and have no self evidence or announced way to use this spectrum too for incremental gain. So it appears the spectrum provides no economic gain to Vz. Vz would bid a sum equal to avoided decline in net present value of future prospective profit. Who knows what that number is? Whatever it is, that's the Verizon bidding number. By contrast, under the Frontline plan you and anyone else can launch Frontline efforts to raise money and bid. The price paid will be, presumably, a price that still permits the winner to make a reasonable profit by pursuing an open and a public safety conjoined business model. That price may well be as high or much higher than the Verizon price; there's no reason to assume it will be lower.

Third, what is the significance of the price of the auction? What the spectrum is truly worth to Americans is measured by the welfare gains stemming not just from the sale but also from the use of the spectrum, which can and would be in the hands of Frontline measured in the billions. The network build out spend of $7B or more, all by itself, compares favorably to the zero Vz would spend to build out the network. As I said about a million times when I launched the auction regime now coming to an end due to lack of remaining spectrum to sell, there's only one good reason to sell spectrum: it then goes to the firm that values it and wants to use it. That was back in the day when spectrum caps assured that firms would buy what they needed and not more, not warehouse, not bid to deny entry. In today's era, when that strategy is possible, it remains the case that the impact on competition and in this case public safety is far higher than any alleged, and possibly completely erroneous, differential in auction price. For instance, suppose for discussion's purpose that Vz should be willing to pay $4b to foreclose competition and a new entrant should be willing to bid $3B to launch a business that will lead to innovation, billions more of network build, and a 10% drop in retail price (I'm making all this up for a heuristic). The last of these translates to a savings of $5 per sub per month per 80% penetration, or $1.2 billion -- and that is in one month. So should the government be happy to have grossed an extra billion in the auction and passed up an order or two in magnitude more in welfare gains?

Followon discussion:

Evslin: Reed and I have had a long term respectful disagreement over whether public safety packets should be prioritized on the wired Internet and I think this carries over to here.  My contention is that the Internet performs as well as it does in emergencies because there are NO vulnerable centralized control facilities such as those that would be needed to authenticate which packets get priority and to turn priority routing on or off.  I believe that a centralized facility is both vulnerable to targeted attack or to simply becoming a drag on an otherwise distributed network during times of emergency.

Nevertheless, I appreciate your clarifying that packets will be prioritized rather than frequencies preempted.

Also appreciate the clarification that public safety entities MAY build out while waiting.  But the economics and politics are daunting since they aren’t equipped to get offsetting commercial revenue and will have a very short period to get a return on assets they deploy.  Seems to me it would help the proposal to allow other entities to build using these frequencies where Frontline isn’t planning to go in the early years – perhaps with some sort of arrangement for a buyout of facilities that are built to a certain specification.  Without a provision like this, from a rural perspective, 10 valuable MHz stay fallow for up to ten years AND the economics for meeting public service needs in the interim are prohibitive because the payback is so short.

Waldron: On the prioritization point, every IP network built today has the capacity for prioritization.  If that vulnerability exists, it exists today and is an artifact of the IP network.  This proposal therefore produces no new vulnerability, and so the concern you raise is a criticism of every 700 MHz spectrum block (since they all will use IP networks) and there is nothing unique to the E Block and Frontline proposal.  We may disagree whether the concern is valid, but I just want to make sure that it is wrong to suggest this is a new vulnerability unique to the E Block.  For that reason, I do not see that as a fair criticism of the E Block proposal, but rather a commentary on IP networks more generally.

Hundt: I agree w gerry. Having worked with several security companies I have a working knowledge of issues and believe they are common to all, and solvable best with a national licensee as we propose.

Those buildouts are mandatory. If someone wants service sooner, our incentives are to provide it.

Evslin:  Many local IP networks have prioritization schemes.  End-to-end, the Internet famously doesn’t (that’s why Reed proposed one in his Atlantic article).

Granted that Frontline is technically a subnet and not the Internet, it will still be big enough to be important in its policies.

As described, “ordinary” ps traffic won’t be prioritized.  Therefore, in an emergency, some central control has to decide in what regions priority is needed and instruct the appropriate routers.  I believe that poses three dangers:

1. the central authority could malfunction in an unprecedented emergency and have perverse effects;
2. a malevolent actor might successfully masquerade as the authority and disable the net just when it is most needed or as part of an attack;
3. the delay imposed on non-public safety traffic in an emergency might be enough to cause all non ps packets to be dropped or delayed so substantially that they can’t be used for applications like VoIP.  Very hard to know before the emergency.  The result is what happens with the PSTN:  People can’t reach others who might help them or even reach first responders at least if their ISP uses Frontline.

There are related technical problems such as how are ps packets authenticated?  How is 911 kept working over VoIP since the callers packets are not from a PS entity?