The US Commerce Department has issued what it calls a set of “Internet Protocol Numbering Principles” December 3rd. In this post, we contrast the NTIA statement with the results of the Baku IGF workshop on IP numbering. The contrast shows that either the NTIA is out of touch with reality in numbering policy, or is deliberately making a meaningless statement in order to sidestep some increasingly intense, consequential politics.

The last time the US Commerce Department issued a set of principles, it was for the Domain Name System (DNS); it came in 2005, at the height of the World Summit on the Information Society’s controversy over the legitimacy of ICANN and US control of the DNS root. The Bush administration principles basically extended a middle finger to the world and said “the DNS root is ours and we aren’t letting go of it.  But we will work with you to address concerns about your control of your country code.

For better or worse, the 2005 DNS principles were meaningful and had impact. If the current NTIA is attempting to replay those glory days, they have failed miserably. The IP Numbering Principles are little more than an anodyne description of the status quo. They say that Regional Internet Registries (RIRS) are responsible for IP address policy development (a fact, not a principle); that ARIN is the RIR for North America (another fact) and that the USG “is supportive of the policies, processes, and procedures agreed upon … through ARIN” (not really a principle, either); that the USG encourages IPv6 adoption; that multistakeholderism is in general A Good Thing.


But wait: there appears to be a bit of meat in the last principle:

Consistent with the policies developed through the current multistakeholder model, the USG believes that all IP numbers are allocated for use on a needs basis and should be returned to the numbering pool when no longer needed.

For a brief moment – just a flash – you think there is something there. But after reflection you realize no, there isn’t. Indeed, anyone familiar with the real situation in IP numbering is likely to burst out laughing after reading this “principle,” as it is so completely out of touch with reality. Forty two percent of the IP address space was not allocated on a needs basis. The RIRs have no contractual or legal authority to reclaim that space. You can say that it “should” be returned to the numbering pool, but it won’t be, and everyone knows that. Even for that portion of the address space that was allocated on a needs basis, it is well-known that holders of address blocks have no incentive to return unused blocks and do not in fact return them in 99% of all cases. Moreover, the ability of RIRs to reclaim even that allocated space is weak, given the drains on their resources such a course would entail, the potential for litigation, and the risk of generating animosity among their members.

The infeasibility of reclamation, much less of voluntary return, is why the RIRs developed market transfer policies – and why IPv4 transfers are evolving into a market worth hundreds of millions of dollars per year. Yet there is no mention of market transfers in the NTIA principles.

The role of needs assessment in the aforementioned transfer market is a major source of global controversy (although the NTIA does not show any hint of being aware of that, either). Thus it might be useful to bring to their attention, and to the attention of the wider IGP blog-reading public, the report of the workshop held at the Baku Internet Governance Forum on that very topic. A document with the full, official IGF report on the workshop is attached to this post here: Report on IGF Workshop 76

The NTIA has expressed its undying support for multistakeholder policy development processes. It stands to reason, then, that it should take notice of the outcomes of a multistakeholder workshop on the issue.

IGF workshop #76, “What is the best response to IPv4 scarcity? Exploring a global transfer market for IPv4 addresses,” experimented with a new method of deliberation. Instead of a few selected panelists giving talks to an audience and then answering questions, it featured an open discussion guided by a framework. The framework, which is included in the full report, set out 5 policy issues related to market trading of IPv4 number blocks. The three key issues we actually had time to discuss were:

A)    The role of RIR needs assessments in transfers
B)    The status of (uncontracted) legacy block holders
C)    The accuracy of post-transaction records

For each issue, several policy alternatives were defined, and were intended to be used as the basis for discussion. The policy options could be – and were – modified during the session. At the end of the discussion a straw poll was held to see how many participants agreed with one of the articulated alternatives, and whether anyone had changed their mind.

The workshop discussion succeeded in clarifying some of the policy options proposed in the workshop, but did not result in consensus. Instead, the discussion and the straw polls demonstrated how divided participants were over some of the policy options. The discussion and debate did lead to modification of the policy options considered. In issue A (needs assessment), several new policy options were added, including the idea that if needs assessments were meant to prevent monopolistic forms of hoarding or unproductive speculation, that it might be possible for other entities, such as national regulators, to undertake that function rather than using the RIR as a mechanism to enforce market behaviors.

The results of the straw poll for needs assessment show that 9 participants wanted to end needs assessments altogether for IPv4 transfers; 6 supported retaining them in their current form; 6 would like to search for another party to enforce the policy objectives and remove that function from IP address registries. Two (2) participants abstained and 2 went for none of the above. Thus, most participants supported some kind of change in our approach to needs assessments, but there is no consensus on what that change should be. It would be nice if the NTIA showed some awareness of the nature of this debate.

Issues B and C were collapsed into the same discussion, as it was agreed that the policy options for both issues were essentially the same. The issue is whether holders of legacy address blocks that are not under contract to an RIR need the approval of an RIR to sell their number block to another organization. This led to intense discussion of the role of the address registry. The discussion could be said to have produced a strong consensus on the importance of a common, comprehensive registry that maintains the uniqueness of all IP number allocations and assignments. The differences arose over whether the registries should be allowed to use that essential function as leverage for imposing other policies on legacy holders, or enforcing certain market behaviors in the context of an aftermarket for addresses. Several participants noted that the IP numbers had no utility without a uniqueness-maintaining registry, and that all entities described in the registry had a common interest in the maintainence of the registry function. This led to the modification of policy option to include a proviso that those parties who held addresses that were described in the registry should be required to pay their “fair share of related registry costs.”

The straw poll for Issue C revealed even sharper division than on Issue A. Unfortunately, the number of votes was diminished by the fact that the operators of the remote participation process failed to submit the poll to the remote participants. At any rate, only 3 of the 11 participants in the poll supported alternative C3, that “RIRs should not update records unless receiving party signs a contract and conforms to RIR policies.” Five (5) supported C2; that “RIRs should update legacy transactions based on legal proof of transfer, but legacy holders should pay a fair share of related registry costs.” No one (0) supported the view that legacy transfers should be updated regardless of whether the parties pay a fair share of registry costs. There were 3 abstentions.

The economic impact of needs assessment policy is clear. Our study of the transfer market showed that stricter needs assessment criteria for allocations from the remaining free pool in ARIN is driving many organizations to the market place to buy addresses, even though “free” addresses are still available from ARIN. It also showed that organizations may be buying addresses in order to have stronger property rights in their address assets, and that RIR efforts to deny or undermine those property rights acts to stifle the market.

It may be that the NTIA is being cagey. They may be fully aware of the controversial nature of the needs assessment issue and the status of legacy holders’ property rights. It may be that they deliberately chose not to address those issues. We know that people within ARIN were lobbying it, looking for a much stronger statement on the property rights issue. If so, all they got was a rather weak statement about what “the USG believes” rather than a statement of policy or law.


1 thought on “NTIA’s meaningless “numbering principles”

  1. Milton –

    You reference your report of the workshop held at the Baku Internet Governance Forum as a potential resource in discussing this topic, and then go on to include counts from various polls; could you discuss the validity of drawing conclusions from that report based on the relatively small sample size of participants? (This is particularly important given the objections to the proposed methodology expressed at the workshop, and the fact that as a result, some folks left the room rather than be included.)


    John Curran
    President and CEO

Comments are closed.