My third FCC Technical Advisory Council meeting (3-hr. video archive here) was the most exciting yet. The TAC’s Critical Legacy Transition working group, studying the legacy public switched telephone network, recommended that the Council advise the FCC to set a concrete date to sunset (shut down) the Public Switched Telephone Network (PSTN). (!) The working group recommended the year 2018 as a starting point for lively discussion.
The working group’s co-chair Adam Drobot emphasized that the specific date is a straw man — the important point is that we should advise the FCC to pick a date and begin the substantial preparation work required to ensure that alternatives are available. In particular, the proposal (slides 4-18) included adding a non-trivial set of critical-infrastructure features (emergency services, universal service, qos, metrics for performance, reliability, reachability) to the IP-based Internet and/or mobile/cellular systems, as well as synchronizing the National Broadband Plan’s universal coverage goal with the proposed sunset date for the copper-based phone system.
We then spent thirty minutes trying to understand exactly what was being proposed. The high order bit is that it is not really a choice, much less a radical one. With fewer and fewer subscribers willing to pay for land lines (cutting the cord is real), the PSTN has lost its required economy of scale. So we better find a way to more cost-effectively meet the needs of the remaining PSTN users.
The motivating objective of the industry is to remove as many regulations as possible and let the copper-based phone business fade away gracefully, rather than have government regulations try to prop up a dying industry. In other words, the U.S. regulatory framework should catch up with the reality that the Internet killed the phone business years ago, so the TAC ought to advise the FCC in the direction of stewarding a transition. ATT proposed to the FCC two years ago that carriers should not be legally required to provide copper-based phone service anymore. Still unknown is how much new regulation is required to bring the Internet and/or the wireless and mobile worlds up to the same public interest obligations as the PSTN. Or if it is possible to do at all.
Dale Hatfield pointed out that full deregulation at the highest level would leave market power issues at the layer below, so we have to clarify exactly what problem we’re trying to solve. Dave Clark echoed this sentiment, noting that what succeeds will depend on who invests in which outcome, so for example, streaming video will make more of a difference to the resulting architecture, because economic forces will drive it that way.
One of the recurrent questions was regarding the ramifications for the national and international telephone numbering system. The answers ranged from “We’re in the ITU, we’ll work it out.” to “Numbering systems abound in the marketplace; and table lookups are cheap; let the market work it out.” Walter Johnston of the FCC reminded us that phone numbers and IP addresses are not just integers, but tokens for a whole set of agreements (protocols and politics). Ironically, the next topic on this TAC meeting’s agenda was the IPv6 transition, where letting the market work it out has not been auspicious for Internet numbering systems.
Many have blogged about the inevitable PSTN demise, including Om Malik, Avaya, and Tom Evslin (member of the FCC WG proposing the transition) who has been thinking about it for a while and is also impressively capturing public reaction to the pre-proposal.
There were many open questions and requests for a more rigorous definition of the problem being solved, but no disagreement that we need to update the set of antiquated regulations covering what constitutes the future of our communications infrastructure. It seems a little soon for another Telecommunications Act, until I read parts of the 1996 (U.S. Telecom) act, at which point it seems way overdue. The Working Group committed to releasing its current draft proposal and to present the next round of thinking on this topic at the next FCC TAC meeting in September.
Just as exciting was the next topic: IPv6 (I am a member of this working group). John Brzozowski of Comcast presented (slides 19-37) background on IPv6, some information on the current state of IPv6 readiness, and some potential recommendations for setting national objectives, benchmarking standards, and other government policies to support the transition. Harold Teets of Time Warner Telecom offered a reality check by asking for clarification of actual IPv6 traffic numbers, to which John responded that traffic levels were “less than 1%, probably less than 0.5%”. (I found no more than 0.3% in a recent survey of published measurements. Still, basically zero, as many have been warning for years.)
Interestingly, in light of the previous PSTN discussion, we were immediately asked why we did not propose a specific government-led transition date at which the IPv6 Internet would be supported by all service providers. Several people clarified that it was not really the jurisdiction of the FCC to set a flag day (except for turning off the phone network, apparently), so the IPv6 transition would be more of an evolution than an event. Dave Clark again reminded us that the Internet is operating in an inherently international context (and with an unsatisfying “multi-stakeholder” substitute for accountable and transparent governance of naming and numbering), so it is not clear what such a flag day would mean without global coordination and support.
In the absence of government regulation, Peter Bloom of General Atlantic asked what the forcing function for IPv6 would be, where we would see a tipping point toward IPv6 deployment. John suggested that the obvious forcing function is economic — when it gets more expensive to support the alternative than (for Internet service providers) to switch over to IPv6. I disagreed — such an economic model would work if providers were all converting to independently functioning widgets, but that’s not the “alternative” available — the new widgets only work if everyone else converts. Incumbent service providers have two alternatives in the time horizons that frame their business decisions: (1) support a bunch of complex IPv4 plus NAT technology that vendors are already building; or (2) support a bunch of complex IPv4 plus NAT technology that vendors are already building and support IPv6, at considerable extra cost, for an indefinite period of time. What would you do?
It is not just that there is no forcing function, but we now have an actual counter-forcing function. The RIRs’ self-regulatory experiment in management of Internet numbering systems has recently driven off a cliff by sanctioning the purchase and sale of IPv4 addresses as if they were real estate. As an example, I mentioned that Microsoft paid millions of dollars (it turns out only $7.5M, or $11.25 per address) to acquire some IPv4 addresses in Nortel’s bankruptcy proceedings. Institutionalizing a valuable market in IPv4 addresses is a reliable recipe for removing any incentive for IPv4-holders to invest in upgrading to IPv6.
Then there was another lengthy discussion (slides 38-53), accompanied by a dense technical report that should be released soon, on optimizing shared access to spectrum. Potential recommendations including developing better metrics for technical efficiency of spectrum use, best practices for improving receiver standards, particularly in spectrum selectivity and sensitivity, identifying successful examples of spectrum sharing, encouraging small cell deployment, and eliminating barriers to infrastructure deployment.
Finally the Broadband Deployment working group reported briefly on its charter to improve the inconsistencies in state and local government permitting processes that pose obstacles and counter-incentives to infrastructure investment. We learned that there is already another FCC Intergovernmental Advisory Committee that might be instrumental to these objectives, so this thread only took a few minutes.
What was most astonishing was the matter-of-fact acknowledgment that our current Internet (IP) numbering resource allocation system has reached its architectural limits, with no current plan for overcoming the problem, sandwiched between a daunting admission that all communications infrastructure will be converging to IP and we had better hurry up and prepare for it, and an ambitious objective of dramatically improving the way we share spectrum resources. There seems to be some unconnected dots in this picture — a lack of recognition that IP addresses are the “spectrum” for the Internet, and their efficient and equitable allocation is yet another critical-infrastructure feature that will need integration into whatever kills the phone system. It is not a problem the FCC can solve on its own.