[arin-ppml] IPv4 Transfer Policy Change to Keep Whois Accurate

[Tom Vest]

On May 12, 2011, at 1:26 PM, Mike Burns wrote:

> Hi Owen,
>> 
>>> I still don't see the connection between my proposal to drop needs requirements for transfers and the participation rate of DNS whois or the UK land office.
>>> I may be missing something obvious, though.
>> 
> 
>> I believe he is arguing that if you turn address policy into a free-for-all (as in your proposal), like
>> DNS, it will decrease, rather than increase whois accuracy. I hadn't thought of this consequence, but,
>> now that Tom and Paul have brought it up, it does make sense.
> 
> 
> I am still missing the connection between removing needs requirements for transfers and having that decrease whois accuracy.
> If you can connect the dots, you will go a long way in convincing me that my proposal is flawed.

Hi Mike, 

Again, apologies for the delayed response. I tried to clarify the connection in my message of May 13, 2011 12:04:39 PM EDT (specifically, @ bullet points 3 and esp. 4). But that was a long message that also covered several other reasons why the so-called "needs" (which IMO should be "capability") requirement is important which are completely independent of its relevance to whois accuracy, and perhaps I wasn't clear enough on the specific one that interests you.

Basically, the need/capability test facilitates the ongoing maintenance/preservation of whois accuracy because it assures that each subsequent allocation/assignment (and in the future, each transfer transaction) will trigger the same kind of "moment of controlled disclosure" that occurs when a new entity joins an RIR and/or requests an initial allocation. As a group, network operators -- and esp. growing ones -- undergo the sort of internal changes (e.g., reorgs, relocations, new sites, new non-M&A commercial partnerships, et al.) that can trigger changes in their external contact details fairly frequently. For all sorts of reasons that are mostly banal (oversights, procrastination, impatience with "bureaucracy," miscommunication, someone else's job, thought they were already informed, etc.), the RIRs don't always "get the memo" at the time when such changes occur -- or even afterward, during subsequent "casual" interactions. Absent other countervailing factors, such changes would cause the overall quality of registration data to degrade progressively over time, with the more dynamic/faster growing networks typically leading the way down. 

What prevents (or at least substantially mitigates) this progressive decay is the policy-mandated needs/capability test requirement. That requirement assures that each subsequent interaction between registrants and the RIR that could materially alter the distribution of IP number resources *will not* be "casual" in the above sense, but rather will (typically) involve some presentation of documents which illustrate the existence and size of the new addressing requirements. The review of such materials -- which frequently include invoices for new network-related assets or similar documents that show buyer address and other contact info -- provides a formal opportunity for RIR and registrant representatives to make sure that they're on the same page with respect to all current contact information.

So, to put this explicitly in the context of your proposal:

The exhaustion of the unallocated IPv4 pool is not going to reduce the frequency with which address registrants undergo the sort of internal changes that can make some or all of their current whois contact details outdated -- if anything it might make those changes happen more frequently. Thus, in order for whois data quality to be preserved going forward at (at least) current accuracy levels, the current practice of making each subsequent address-related transaction subject to a mandatory needs/capability capability review must continue.

In order for your proposal to have *any chance at all* of causing a net improvement in whois data quality, the number of future Pv4 transfer seekers that are 

x: (NOT current RSA signatories + NOT willing/able to undergo a need/capability test + ARE certain to self-maintain the quality of their own whois information at historically unprecedented  high levels in perpetuity)

...would have to exceed the sum of other kinds of future address seekers, including those who are:

y: (NOT current RSA signatories + ARE willing/able to undergo a need/capability test + NOT certain to self-maintain the quality of their own whois information at historically unprecedented high levels in perpetuity)

z: (ARE current RSA signatories + (n/a) + NOT certain to self-maintain the quality of their own whois information at historically unprecedented high levels in perpetuity)

Logically, the universe of potential future address seekers is complete characterized by (x + y + z) as described above, plus two other groups that, hypothetically, wouldn't be affected either way by your proposal:

Incorrigibles: (NOT signatories + NOT willing + NOT self-maintaining) 

Saints: (ARE signatories + (n/a)+ ARE self-maintaining)

[Note: I say "hypothetically" above because I actually believe that the adoption of this policy would undermine an existing community "norm" of whois participation that currently contributes to "irrationally" high data quality across current registrants -- and as a result your policy would cause average levels of whois "self-maintenance" to decline across all groups. But that possibility is not factored into this analysis]

Assuming that "saints" and "incorrigibles" would be equally represented across both current ARIN members/RSA signatories and future address seekers (and excluding any possible "normative" affects), your proposal would only be net positive at the point where ((non-Saint, non-Incorrigible x)) exceeds (non-Saint, non-Incorrigible (y+z)). Given the size of the current ARIN membership, the only way this pans out in your favor if 90%+ of current members and 90%+ of future address seekers actually fall into the "Saint" or "Incorrigible" category.

But of course, this assumption would also mean that your policy (and all policies, more-or-less) are almost complete irrelevant.

Happily, I believe that  those demographic assumption are grossly inconsistent with both RIR administrative experience and with the documented record of RIR community-policy interactions over the last 20 years. 

Hence, I am opposed. 

TV


>>> My whole goal is to increase accuracy in Whois, and I am not relying on any financial or price mechanism for that increase.
>> 
> 
>> And now there is evidence that your proposal would likely have the opposite effect.
> 
> What evidence?
> 
>>> I have not argued that pricing will increase registration, I have argued that pricing will ensure productive use.
>> 
> 
>> Which also remains in dispute and unproven.
>> Owen
> 
> Hence the word argued.
> 
> 
> Regards,
> Mike