[arin-ppml] ARIN-prop-176 Increase Needs-Based Justification to 60 months on 8.3 Specified Transfers

[Owen DeLong]

On Jul 3, 2012, at 2:44 PM, Milton L Mueller wrote:

> Sorry, Owen, 
> I wasn't aware that you were calling for a statistical comparison of the 1 year vs. 2 year, I only saw Jimmy Hess's message, which seemed to be calling for "substantial statistical evidence" favoring 5 years before making a change. It was unclear to me how one could ever get statistical evidence about a 5 year period. 
> 

I posted the first call for evidence and suggested that we needed evidence that 5 years would not be harmful before considering such an extreme change.

I would also posit that experience with the 2 year period and some evidence that going to 3 years would provide a benefit without causing undue harm is also necessary.

> How long was 1-year in effect? How long has 2-year been in effect? Based on my rough understanding of the periods, it would seem that increasing the time horizon has already greatly increased the pace and number of transfers. However, there are confounding variables, namely the growing scarcity of IPv4 as time passes and the depletion of the free pool. 

One year was in effect as the free-pool allocation period for many years and served us well. It was the original metric for transfers enacted in the 2009-1 draft policy when it was activated, so I would say somewhere between 2 and 3 years.

Two years went into effect very recently (the very latest revision of the NRPM).

To the best of my knowledge, there have not been any transfers yet since the change, so I'm not sure how you could make such a claim about the increased pace.

> Regarding the experiment with no needs assessment at all, I guess if you view an experiment with the absence of needs assessment as equivalent to releasing hordes of murderers and rapists into the community unfettered you would indeed be opposed to such an experiment. My point was simply that the difference between a 5 year horizon and no needs assessment at all might be small enough that we needn't bother with a specified time period, if we were actually willing to conduct an "experiment." 

I would agree that 5 years bears little difference from eliminating needs basis. The community has resoundingly and repeatedly said no to eliminating needs basis, so I would suggest that on that basis you have just made a very good argument against a 5-year time frame as well.

Owen

> 
> --MM
> 
>> -----Original Message-----
>> From: Owen DeLong [mailto:owen at delong.com]
>> Sent: Sunday, July 01, 2012 11:22 PM
>> To: Milton L Mueller
>> Cc: jeffmehlenbacher at ipv4marketgroup.com; Scott Leibrand; arin-
>> ppml at arin.net
>> Subject: Re: [arin-ppml] ARIN-prop-176 Increase Needs-Based
>> Justification to 60 months on 8.3 Specified Transfers
>> 
>> 
>> On Jul 1, 2012, at 4:17 PM, Milton L Mueller wrote:
>> 
>>> 
>>>> -----Original Message-----
>>>> I understand your desire to sit tight and assess statistical evidence
>>>> before suggesting a longer justification period is required. My
>>>> concern with such a strategy is the decided lack of comprehensive
>>>> transfer market data.  We have only the ARIN Specified Transfer
>>>> Listing Service
>>> 
>>> [Milton L Mueller] As a social scientist, it seems obvious to me that
>> asking for statistical support for a longer time frame, without allowing
>> an actual, real-world experiment with the longer time frame, is asking
>> for the impossible.
>>> 
>>> The only way to gather "statistical evidence" on the impact of
>> changing the time frame for needs justifications is to allow a different
>> needs assessment time and see how the change affects the quantity and
>> type of transfers authorized. No empirical conclusions can be drawn
>> about the relative merit of a 24-month and 60-month period by looking
>> ONLY at the statistics generated by a 24-month period.
>> 
>> And we have now allowed a different policy (24 months instead of 12) and
>> I want to see a statistical analysis of that change before making yet
>> another modification. I'm not sure how you think your argument applies
>> more accurately to 24->60 than it does when I propose it for analyzing
>> the change from 12->24 before applying another change from 24->60.  Can
>> you please clarify what I am missing here?
>> 
>>> That is why I view the request for statistical evidence as a tactic
>> designed to delay or defeat Jeff's proposal.
>> 
>> A very interesting conclusion. Inherently, it would delay Jeff's
>> proposal since we are seeking to wait and see what happens with the last
>> change before applying Jeff's proposed additional change, but, I fail to
>> see how your above statement makes that an unreasonable approach or why
>> delaying Jeff's proposal is somehow an inherently bad thing to do as
>> your statement above seems intended to imply.
>> 
>>> If one really wants to do an experiment, it would probably make more
>> sense to conduct a limited experiment with no needs assessment at all,
>> and see what happens. If one discovered a significant increase in the
>> number of approved transactions, and/or a massive increase in what
>> appeared to be speculative acquisitions, and if either result could not
>> be explained by other variables, it would support the conclusion that
>> the current time horizon constricts the number of transactions in a
>> specific way.
>> 
>> Other than the magnitude of the consequences, how would this differ from
>> an experiment in cost-cutting prisons where we simply released every
>> prisoner that signed a written promise to obey the law and reversed the
>> policy only after we found many of the criminals breaking the law again?
>> I don't advocate this kind of destructive testing.
>> 
>>> In the absence of such an experiment, the only empirical data that
>> might support or refute the change would be a survey of all prospective
>> buyers in which a statistically significant sample of them stated
>> unambiguously that they would participate in the transfer market if the
>> needs period were extended to 60 months; or that their planning horizon
>> for acquiring IP addresses was closer to 5 years than to 2 years. Such a
>> survey would be very difficult to conduct, and you would still be
>> dealing with stated preference rather than revealed preference. But it
>> would be potentially informative.
>> 
>> I'm not sure that preference is in question here. I'm sure that those
>> seeking to acquire resources through the transfer market would prefer
>> maximum leniency in the process. I'm willing to accept that as a given.
>> 
>> The question at hand is whether allowing that would be damaging to other
>> areas of internet policy and to what extent. Looking at the impact of
>> the switch from 12 to 24 months for some period of time before making
>> yet another modification seems prudent.
>> 
>>> The bottom line is that those calling for statistical evidence have
>> one of two choices: either they must agree to conduct an experiment, or
>> they can abandon the claim that they oppose the proposal for lack of
>> statistical evidence and admit that they just don't want it to happen,
>> regardless of evidence.
>> 
>> Given sufficient evidence that it would not harm the community to do so,
>> I would not oppose the proposal. Until such evidence becomes available,
>> yes, I oppose the proposal. Whether or not such evidence is or can be
>> made available does not change my desire not to harm the community. As a
>> general rule, it is not unreasonable for the community to expect those
>> desiring a change to bear the burden of proving that the change is more
>> beneficial than harmful to said community. It is my opinion that those
>> supporting this policy have not yet met that burden.
>> 
>> Owen