Decoding the happy talk at an IGF plenary session

The Internet Governance Forum has many valuable and productive workshops going on but it has never figured out what to do with its plenary sessions. You bring 1000+ people from all over the world into a single room – people who are interested and involved in Internet governance. What do you do with them?

There has always been a tension in IGF between two camps. On one side are commercial and governmental interests who want to avoid rocking the boat and thus want IGF plenary sessions to avoid airing real debates and policy conflicts. These people want to fill the plenary sessions with happy talk or soothing anecdotes – e.g., how puppies were saved by an ipv6 implementation in a rural village in Africa thanks to an ICT4D grant. On the other side are those who want IGF to provide interactions that will bring out policy conflicts and problems and provide an opportunity for different perspectives to debate these problems (and perhaps even move toward resolving them).

This year's main session on “critical internet resources” (i.e., domain names, numbers and routing) provided a numbing reminder of what happens when the first camp wins.

First, some months ago the session organizers decided that routing was not really an important issue. So what happens in the months leading up to IGF? We have intense debates over RPKI implementation and policy, we have several instances of route hijacks and technical breakdowns in routing. But routing is not discussed, not even mentioned, in the main session.

Second, what is the single most important and high-stakes issue in critical internet resources? The answer is easy: it is the scarcity of ipv4 addresses and the impending depletion of the free pool of addresses. It is becoming increasingly evident – even to the most ardent IPv6 evangelistas – that ipv4 will run out in about 18 months, whereas ipv6 will not be widely deployed for another 5-10 years. That leaves us with at least 3 and possibly 15-20 years when there are no ipv4 addresses and we can't switch to an ipv6 internet.

The IGF plenary session gave us lots of happy talk about how various people are implementing IPv6. What they didn't tell you is that “implementing IPv6” at this stage requires also the use of additional ipv4 address resources. Because at least 99% of the internet is not compatible with IPv6 now, implementation requires running what is called “dual stack;” i.e., both ipv4 and ipv6 at the same time. So unless people are prepared to cut themselves off from most of the internet, the next ten years will require additional IPv4 resources that we just don't have. This means markets for ipv4 address transfers will develop, and that will require major changes in the policies and practices of the RIRs.

Nothing was said about this. Indeed, the dialogue was structured to prevent us from talking about it. The main session emcees had lined up a set of status quo speakers who do not want the public to worry about that stuff.

In short, at the IGF main session on CIR, there was no discussion of IPv4 scarcity, no discussion of address transfer markets, no discussion of the problems inherent in the dual-stack strategy for the transition.

It gets worse. There was supposed to be a discussion of “enhanced cooperation.” What's the biggest issue in enhanced cooperation? It is the renewal of the IANA contract between ICANN and the U.S. Department of Commerce. This part of the program was filled with anodyne but somewhat informative discussions of how ICANN views the world and how it is implementing its Affirmation of Commitments. Several pre-designated speakers were called upon, including ICANN's board chair.

Thankfully, Bertrand de la Chapelle, a French government representative who was recently appointed to the ICANN Board, departed from the script. He opened up a discussion of the IANA contract renewal by making some useful and interesting comments about the institutional arrangements around the IANA contract. The IANA, he said, needed to be primarily concerned with the integrity of the domain name root and not simply framed as an issue of U.S. unilateral control. Responding to Chapelle, I made some comments about the implications of the IANA contract renewal and how some people were advocating separation some of the functions in the IANA contract; e.g., detaching the address root from the DNS root and perhaps even separating out standards and protocols to the IETF. Jeanette Hofmann, a session co-chair, asked some useful followup questions.

But the other co-chair, Chris Disspain of the Australian ccTLD, openly discouraged further discussion by saying “it's a contract and many of the principals don't want to talk about it.” After that, there was no discussion of the political dimensions of enhanced cooperation.

We then went into a discussion of disaster recovery efforts in Haiti. An important topic, no doubt, but not a matter of policy and not involving global governance of CIRs.

As the program wore on, the audience diminished slowly but steadily. Most of the people who remained in the room were doing their email or surfing the web. At one point Disspain said plaintively, “no one's paying attention.”

Emily Taylor's summary of the session made it clear just how much a lack of substance has been institutionalized in the IGF. Her session not only deliberately ignored the discussion of the IANA contract – the only real substantive and unplanned discussion in the entire session – but made no mention of ANY intervention from the floor. The entire “summary” of the session, I suspect, had been written beforehand, based on the pre-designated speakers. And then the kicker: we used to clash and conflict over these issues, she said, but now “the heat is gone.” How nice. 


  1. Anonymous

    “implementing IPv6” at this stage does not necessarily require the use of additional ipv4 address resources, as existing ISP/carriers/SMEs etc can use their existing IPv4 address to dual stack. Plus, as Vint Cerf corrected you yesterday, there are other transition mechanisms besides dual-stacking.
    But blog away, don't let reality get in the way of your hyperbole!

  2. Anonymous

    I enjoy the way you deprecate ICT4D. Such a sense of style while you massacre the cares of others to make your own concerns seems so much more important.

  3. Anonymous

    Internet growth via dual stack does require additional v4 addresses. Even if one relies entirely on NAT (and I suppose you are referring to carrier-grade NAT in your message above) at some point new organizations will require v4 addresses – that is why the RIRs are creating reserve policies for the remaining v4 blocks.
    The dependence of the IPv6 transition on NAT raises several interesting questions, among them whether such NAT'ing will make v6 unnecessary in the long term, or whether the use of NAT by large carriers will make it unnecessary for smaller organizations to convert to v6, as they can retain v4-v6 compatiblity without ever implementing v6. In short, your attempt to posture as technically knowledgeable fails once again. But it doesn't matter so much which of us is correct here – my point was that we were not even able to have that interesting discussion in the main session! And without ornery people like me to bring it up, those topics are deliberately not mentioned by the technical community, which prefers simple-minded v6 evangelism.