[arin-ppml] Against 2013-4

Milton L Mueller mueller at syr.edu
Mon Jun 3 09:50:16 EDT 2013

I would have to oppose most of the statements in this proposed revision of the RIR principles. 
While it is a good idea to update a document written literally a generation ago for a different IPv4 world, it seems obvious to me that a lot of the thinking that went into this 2013-4 is an attempt to rigidify obsolete thinking rather than to update things. This is a backwards-looking revision that has little support in the real world. 
It seems to me that the proposer of this policy thinks that fundamentally nothing has changed in 25 years.

Here are some examples:

> -----Original Message-----
> Policy Statement:
> Section 0: Principles and Goals of the Internet Registry System
> 0.1. Efficient utilization based on need (Conservation)

This represents confused thinking. Conservation as a principle does NOT necessarily mean needs-based allocation. There are many ways to conserve number resources without the fiction of technical needs assessment. For example, pricing or fees graduated to the number of addresses used is a form of conservation. Market trading, in which you bid a scarcity-based price for number blocks, is a form of conservation. Technical needs assessment as currently performed for IPv4 is another form of conservation, but it is arbitrary and often discriminates against specific technologies. I would not want to see this "update" used to rationalize obsolete methods of conservation. 

IPv6 allocations as I understand them are not actually based on demonstrated need in the same way as IPv4. Whatever your position on the needs assessment debate in IPv4, any "principle" regarding conservation should be formulated in a way that is FORWARD-LOOKING, flexible and leaves the door open for the community to change its definition of the best way to avoid wasting number resources. In a nutshell: efficiency and needs assessment are NOT equivalent. This draft clearly doesn't get that. 

> Policies for managing Internet number resources must support fair
> distribution of globally unique Internet address space according to the
> operational needs of the end-users and Internet Service Providers

Again, "fairness" and "operational need" are distinct concepts. Operational need may justify giving all the available resources to a large provider, which some may find unfair. This statement is essentially meaningless in that it introduces to potentially conflicting standards. 

> operating networks using this address space. The registry should prevent
> stockpiling in order to maximize the conservation and efficient
> utilization of the Internet address space.

I cannot ever support such an economically vague statement. It would authorize a huge expansion in the role of RIRs. "The registry should prevent stockpiling" means what exactly? What does "stockpiling" actually mean in an IPv6 world, anyway? If I use 1/1000th of a /48 assigned to me am I "stockpiling"? Is there any evidence that "stockpiling" is a problem? If so, where is this evidence? 

> 0.1.1. Documented Justified Need (Needs Based)

This section attempts to codify and make permanent a set of policies that were developed in the final death throes of IPv4. What a waste of time. 
The idea of authorizing intrusive "accounting of resources" is precisely the opposite of the way we need to be going, both in IPv4 and IPv6. We should let the market allocate transfers of the fully-allocated IPv4 numbers, and current policies, which give organizations blocks based on the number of networks they claim and some fill ratio, for IPv6. There should  be flexibility in the methods of conservation used. There is no need to specify concrete methods and practices in a principles document. That is just a mistake. 

> All Internet number resource requests are subject to audit and
> verification by any means deemed appropriate by the regional registry.

Reveals a scarcity mentality appropriate to the last decade. 

> 0.2. Hierarchical aggregation (Routability)

I agree with the comments Bill Herrin made about this earlier. " Policies for managing Internet number resources must facilitate scalable routing." Scalability is what we care about, not necessarily hierarchical aggregation or anything more specific. Remember, these are supposed to be long-lasting principles, not a codification of what we are doing now, and not a specific set of policies. Let the community set the specific policies flexibly going forward.

Also agree with Bill that we should have an explicit principle that "ARIN does not set Internet Routing Policy"

> 0.3. Uniqueness (Registration)

This aspect of the proposed revisions really went off the rails. 
First, uniqueness should be valorized as the single most fundamental and important principle of stewardship, the one to which all the other principles are subordinate. It is the most important justification for having a registry. 

Second, the purpose of uniqueness is NOT to aid law enforcement but to maintain the technical integrity of the address space. Law enforcement is, as the term implies, guided by law and if the world's governments want to require certain forms of identification they can do so, subject to due process, legislative checks and balances, constitutional constraints, and so on. We don't need to be doing that job; indeed it is illegitimate and dangerous for us to attempt to insert an address registry into that role. The person who wrote the uniqueness aspect of this document doesn't seem to understand this. 

> 0.4. Stewardship
> It should be noted that efficient utilization and hierarchical
> aggregation are often conflicting goals. All the above goals may
> sometimes be in conflict with the interests of individual end-users or
> Internet Service Providers. Care must be taken to ensure balance with
> these conflicting goals given the resource availability, relative size
> of the resource, and number resource specific technical dynamics, for
> each type of number resource. For example, efficient utilization becomes
> a more prominent issue than aggregation as the IPv4 free pool depletes
> and IPv4 resource availability in any transfer market decreases.
> Conversely, because the IPv6 number space is orders of magnitude larger
> than the IPv4 number space, the scale tips away from efficient
> utilization towards hierarchical aggregation for IPv6 number resources.

This section is also inappropriate for a principles document. It purports to tell the current community, as well as all future deliberations for the next 20-odd years, how to make policy tradeoffs. That is the kind of thing that should be left to the community itself. 

More information about the ARIN-PPML mailing list