[ppml] Policy Proposal: Resource Reclamation Incentives

[Owen DeLong]

On Jul 3, 2007, at 7:39 AM, Leo Bicknell wrote:

> In a message written on Tue, Jul 03, 2007 at 10:07:06AM -0400,  
> Martin Hannigan wrote:
>> AT&T? Ford? Level(3)? Apple? Haliburton?
>
> Two of three already have signed RSA's (I assume).  AT&T and Level(3)
> both have a large number of blocks they got by buying companies
> that formed post-ARIN, and so they got the RSA's with them.  If you
> in herit an RSA, does it cover your legacy space as well?
>
Even if an organization has signed an RSA covering other resources,
I'm not sure that means they would be willing to bring their legacy
space under that agreement.  There is a difference between an
organization being willing to sign an RSA for new resources and
an organization being willing to give up perceived rights in resources
they have held since before ARIN existed.

> I haven't seen anyone make a formal argument it's property.  Legacy
> owners seem concerned about two things:
>
> - Fees.
> - That they not be subjected to the policy requirements for the space.
>   That is, 80%, 50%, or whatever utilization numbers.
>   - Specifically, that their space cannot be revoked for failing to  
> use
>     it.
>
While nobody has made that argument formally (except Martin
at this point), it might not be as easily dismissed as we would
hope.  I don't agree with Martin on the property status (it's hard
for me to imagine a law which provides for the ownership of
integers), but, it's not hard for me to imagine lawyers convincing
a judge that IP addresses are property.

I think that for most legacy holders, the "Fees" issue is a matter
of principle or a minor issue.  I think that the real issues from
the legacy holder perspective are:

	-	I don't get any benefit from signing an RSA
	-	It subjects me to policies that could force me to renumber
	-	It subjects me to policies that could change at any time
	-	It potentially limits my options in terms of what I can do
		with my address space
	-	It costs me money

GIven the first bullet point, any one of the latter bullet points
is pretty much a no-brainer to say no to the existing RSA.
We really don't have a way to change the first bullet point, so,
all we can do is work on reducing or eliminating the negativity
of the others.
> Note, if you want to transfer your block to someone legally (via
> ARIN), the transfer guidelines page clearly states that the new
> owner must sign the RSA and abide by all policies.  That is, if
> there is an implied contract with legacy space it's non-transferable.
>
Right... I fully agree with this.

>> Reclamation is not as simple as it looks. *I don't support
>> Owen's policy*, but I support the intent. Amnesty is a one
>> shot deal, not a stop at the drive through convenience store
>> dropping off a token /20 for a bottle deposit refund every
>> now and then.
>
> While Owen's policy has "Reclamation" in the title, it doesn't
> appear to me it actually encourages reclamation that much.  Sure,
> it allows it to happen, but we already have an amnesty program that
> allows it to happen.  Rather, it appears to me he has three intents:
>
I think it provides some encouragement towards reclamation.
I think that gentle reclamation efforts allowing people to return
address space in whatever size chunks they are willing to and
on whatever timetable they are willing to is more likely to result
in reclamation than policies which attempt to force the issue.

While I can understand Martin's desire to have reclamation
happen as a one-shot deal, I don't think that as many organizations
will sign on for it under those terms.  For the ones that would,
my policy doesn't really provide any advantage to returning
things a /20 at a time.  If you return 10 /20s day 1, you get 20
years of fee waivers (if you are subject to fees).  If you return
a /20 every other year, then, if the policy doesn't get changed for
10 years, you get the same result, but, you face the risk that the
policy could change along the way.
> - Entice people into the RIR system by giving them more favorable
>   terms.  (Sign an RSA, pay fees.)
Yes.

> - Turn in your existing bucket of disjoint small netblocks for
>   one large, aggregateable netblock.
Sort of.  My policy allows this, but, for the most part, these
situations would fall under existing 4.7.  My policy only applies
in cases where the new aggregate netblock is smaller than
the sum of the existing netblocks being turned in.

> - Returning address space reduces your fees to provide some incentive.
>
Correct.

> So if I have 16 disjoint /24's in the swamp taking up 16 routing
> slots I can turn them back in for a /20 and take up one routing
> slot.
>
That would fall more under 4.7 than my policy.

If you had 20 disjoint /24s, then, replacing them with a /20 would
fall under this policy and get you 1 year of fee waivers.
> Sure, there's a carrot in there to reclaim space, but I can't imagine
> anyone thinks this will get someone to return a /8.
>
I don't think it will get someone to return a /8.  I do think it  
might get
some /8 holders to return a /12 or a few /16s or some /20s, etc.

> Rather than call it the "Legacy Outreach and Partial Reclamation"
> policy I think it might be better termed the "Legacy Outreach and
> Aggregation" policy.
>
I think Aggregation is more the purview of 4.7.  I did want to maximize
aggregation as part of my policy, but, it is not a primary goal.

Owen