[arin-ppml] IGF session on IP addresses

[Owen DeLong]

On Oct 28, 2012, at 14:29 , Milton L Mueller <mueller at syr.edu> wrote:

> 
> On the afternoon of the first day of the Internet Governance Forum, there will be a workshop on "What is the best response to IPv4 scarcity? Exploring a global transfer market for IPv4 addresses." It will be moderated by myself and Geoff Huston, Chief Scientist at APNIC. There will also be a main session discussion of the topic the next day. 
> 

Thanks for posting this.

I will, unfortunately, be unable to attend. However, I will provide you with my position on these issues i below in the hopes that those positions can be represented in the room in spite of my inability to attend this particular forum.

I do think that the title of the workshop includes an inherent assumption that skews the validity of the workshop overall. The question "What is the best response to IPv4 scarcity?" should not automatically be answered with anything including the word "transfer". Indeed, the BEST response to IPv4 scarcity is the deployment of IPv6 such that IPv4 scarcity becomes moot. All other responses are suboptimal mechanisms designed to cope with our failure to fully deploy IPv6 in a timely manner.

For purposes of classifying my positions, they come from my perspectives as:

1.	An individual user of the internet.
2.	A legacy resource holder
3.	An IPv6 resource holder
4.	Operations staff at various ISPs over the years
5.	Director of Professional Services at Hurricane Electric
6.	An elected community leader for Number Resource Policy
7.	An active participant in several RIRs Policy Development Fora
8.	Author of many current ARIN policies.

For convenience, I will refer to context 1 as "Individual" and context 2 as "corporate".

> The workshop will take a unique format. It will be organized not as a series of panelists speaking _at_ an audience, but as a structured, multi-stakeholder deliberation over policy alternatives among peers.  It will test the extent to which the various stakeholders interested in that issue can come to agreement on a set of five basic policy issues affecting IPv4 number resources. The five policy issues we will discuss are:
> 
>    the role of needs assessment in market trades;

In both contexts, my opinion is that needs assessment remains vital in market trades. Absent needs assessment, the motivations and incentives for IPv4 transfers become completely decoupled from the policy goals of said transfers.

The ideal goal of IPv4 transfers is to encourage efficient utilization of the IPv4 address space, where efficient is defined as maximizing the number of hosts and purposes for which numbers can be made available. Please disregard any economist-specific definitions of the term efficient, especially as they use that term related to markets as such definition is utterly and completely unrelated to efficient address utilization as I intend the term above.

If you remove the needs assessment from market trades, you allow incentives for hoarding, speculation, and anti-competitive acquisition to become significant factors in the trading of IP addresses and essentially remove any ability of the RIRs or anyone else to prevent such practices. This will not only lead to inefficient IPv4 address utilization, but it will unnecessarily increase costs for those that need addresses and allow address scarcity to serve as a barrier to competition to an even greater extent than it already does.

>    the status of legacy number block holders;

The following opinion is identical in both contexts:

First, I would like to clarify some terms. There are no such things as legacy blocks. There are only legacy registrations. A legacy registration is a registration issued by one of the predecessor registries prior to the creation of an RIR serving the region in question. In the case of Europe, this would be registrations issued prior to RIPE-NCC's inception. In Asia, it would be registrations issued prior to the formation of APNIC. In the rest of the world, it is registrations prior to the 1997 inception of ARIN. (The current LACNIC and AfriNIC regions were served by RIPE, APNIC, and ARIN prior to the formation of LACNIC and AfriNIC). In limited cases, legacy registration status may survive a transfer.

Legacy registrations are distinct from conventional registrations only in the following ways:

1. They predate the current RIR system.
2. They are, in many cases, exempt from paying RIR fees, at least under current practices.
3. They are eligible to sign the LRSA in the case of legacy registrations in the ARIN region.
4. They do not have an explicit contract for registration services.

They are not exempt from or in any way unaffected by RIR policy. The RIRs are the successor registries cooperating under the MoU, acting jointly as the NRO within IANA (operated by ICANN under the IANA Functions Contract). As such, they are responsible for the registration and related services of all IPv4 address blocks as delegated to them by IANA, regardless of when such registrations were initially created.

Since each of the RIRs serves as a body for the community at large to set number resource policy and no stakeholder is prevented from participating in the policy processes of each of the RIRs, those community set policies are binding upon the RIRs in their ability to operate the registries. Legacy holders have no rights to RIR services outside of those policies set by the various RIR communities through their public open policy development processes.

>    the accuracy of post-transaction records;

Accuracy of registry databases is less than ideal. This is largely due to the lack of contracts with holders of legacy registrations and somewhat sloppy recording processes in the earlier days of the internet. While improving the accuracy of the registration database is a laudable goal, there is no evidence that proves a laissez faire transfer market will actually result in an improvement to database accuracy. Further, the elimination of critical policy constraints on number resource distribution set by the community would be far more detrimental than the current state of registration database inaccuracy.

>    aggregation policies

The less regulated a transfer market, the smaller blocks will be traded and the more disaggregation of routing will occur. Current market restrictions due to policy in the ARIN region limit transfers to prefixes no longer than /24. Even at /24, disaggregation due to market forces may well overwhelm the scalability of the IPv4 routing table to the point that IPv4 is no longer viable. Further deregulation can only further accelerate the dysfunctional aspects of IPv4 routing.

>    market power

I'm actually not sure what this means in context, so I have no opinion on this topic at this time.

> 
> For each of these issues two or three opposing positions are set out, and the workshop will attempt to document who takes which position, why, and whether the group can come to rough consensus on one position or some new compromise position formulated on the spot.
> 

I encourage you to publish the "two or three opposing positions" ahead of time so that you can gather comments specifically on them prior to the actual meeting.

> The panel is scheduled for 14:30 - 16:00 on 6 November, the first day of the IGF, in Conference Room 9. Although the actual remote participation link for this session has not yet been created by the IGF Secretariat, this page provides general information and guidance regarding how the IGF will facilitate remote participation. Baku's time zone is 9 hours ahead of US Eastern time, and 3 hours ahead of Continental European time.

I presume these are local time in Baku. For other readers information, Baku is UTC+4. On November 6th, the day of this forum,
the US will be back to Standard Time (EST=UTC-5, CST=UTC-6, MST=UTC-7, and PST=UTC-8).

Thus, the forum start time will be 5:30 AM EST, 4:30 AM CST, 3:30 AM MST, and 2:30 AM PST.

Owen