[arin-ppml] Argument Space / solving for 3(x=reject all)

Tom Vest tvest at pch.net
Thu Sep 4 11:02:18 EDT 2008


(note: the diagram seems to be offline at the moment, so I attached  
below a dated screen capture that I took last night)

Hi Chris,

Great tool, thanks for sharing  this! Now that it's out there, I'd  
like to complicate it a little more ;-)

I think it might be possible to solve for 3(x=all fail) in a way that  
is arguably more "fair," much easier to adopt, revise, and if  
necessary *reverse*, and that would be highly robust against external  
challenges. As such, I think it merits some consideration as an  
alternative to the standing resource transfer proposal.

Start with your node focusing on "additional incentive". It may be be  
that the only additional incentive that some operators need is  
assurance that returning *some* address space and incorporating *some*  
IPv6 over time would not cause them to suffer a major competitive  
disadvantage relative to other operators. If that "additional  
assurance" would be sufficient to motivate enough people to get a  
policy proposal adopted (and/or to attract more support than a  
transfer proposal), then perhaps something a bit like what Stephen  
suggested, i.e., "per-object instead of per-org fees" might do the  
trick.

Imagine that at the end of the free pool, annual renewal fees start  
incrementing per /32, but that ARIN offers a "bounty" equal to 100% of  
the new fees to any signatory that agrees to voluntarily return some  
scale-sensitive quantity of IPv4.* For the sake of "simplicity" I'm  
going to illustrate using $1 per IP, rounded up to the nearest  
classful bit boundary for the renewal fee, and define the address  
return required to earn the "bounty" as equal to the next smallest  
classful address block, but many numbers/ratios would probably work  
equally well.

So, for example, the mechanism I'm imagining would oblige a /8 holder  
to return one /16 per year in order to qualify for the bounty, which  
would also effectively make this approach revenue neutral. ARIN would  
never have to handle one penny more than it does today. To make the  
effects consistent across the smaller end of the address distribution  
spectrum, maybe additional fees and bounties are phased out for  
members that retain a /20 or less, until all members are down to  
roughly equivalent sized IPv4 endowments.

*less any added administrative costs, which should be modest.

What would this approach accomplish?

NEW-1: Incremental, inevitable, but (more) predictable effects - No  
matter what, everyone is facing the same reality of doing more with  
less IPv4, or much much more expensive IPv4, and/or perhaps with some  
IPv6. No one should be imagining that making a windfall on the  
transition, or pushing most or all of the costs of transition onto  
future new entrants are sustainable options; they aren't. By the same  
logic, no one should be significantly harmed by parting with 1/256 of  
their existing IPv4 reserves every year, especially if everyone is  
facing the exact same constraint. Given the mechanism's scale- 
sensitive uniform effects, even operators who grudgingly support the  
idea while still hoping/expecting to NEVER make a full transition to  
IPv6 might find comfort in the knowledge that they could have hundreds  
of years to be proven right. See also New-5, below.

NEW-2: Recovery of liquid legacy IPv4 address space - Nothing in this  
approach requires or assumes that ARIN members will return address  
space that they received directly from ARIN, and everyone is already  
assuming the emergence of some kind of gray market (at least) under  
any/all future scenarios. If quietly purchasing IPv4 in a gray market  
looks like a better deal than returning RSA-covered addresses and  
perhaps adopting some IPv6 on the margin, then nothing would  
explicitly prevent people from doing that. Thus legacy/surplus address  
space holders are not absolutely precluded from capitalizing on their  
early efforts / good fortune, but sales do not have to be formally  
condoned, and the whole system does not have to be jeopardized in  
order for that option to be preserved.

3(a= reject): Nothing short of militarization of the process is likely  
to eliminate all speculation/profiteering, but the policy would define  
an implicit "official price" for IPv4 that could help to establish a  
firm ceiling on speculative pricing.

3(b= reject): By leaving ARIN in place as the sole official mediator  
for IPv4 "recirculation" -- not transfers -- the risk of full  
privatization (intentional or unintentional) is reduced to zero.

3(c= reject): IPv4 recovered by ARIN could be warehoused permanently,  
e.g., to assure an eventual return to address space homogeneity  
somewhere down the line, and to send a signal to non-participants that  
IPv4 *will* be obsolete sooner or later (reinforcing New-2).  
Alternately, the proceeds could be put to other uses (e.g., made  
available for subsequent allocations to those willing to pay the same  
$1 per acquisition and renewal fees, with the proceeds returned as  
dividends to the entire community, or selectively and proportionately,  
e.g., for accelerated IPv4 returns). Also:

New-3: Preservation of industry openness - Regardless of how the  
majority of recovered IPv4 is disposed of, enough will always remain  
available -- ideally through 2008-5 style transitional allocations --  
to clearly demonstrate to all internal and external stakeholders that  
all segments of the industry will remain open to new entry in  
perpetuity. Should also help wrt...

New-4: Mitigation of antitrust risks - In the absence of (New-3) large  
IPv4-based service providers will be perpetually at some risk of  
antitrust action. The proposed policy might conceivable redirect some  
of that risk in the RIR's direction, but it seems to me that given the  
pro-open access orientation, a community consensus-supported approach  
like this would probably provide the strongest possible defense  
against any/all antitrust concerns.

3(d= reject): This approach should help to squelch any bets/ 
competitive strategies that an IPv6 transition will never happen. Once  
people get over the psychological hurdle that IPv6 really *is* coming,  
and understand that the transition is not going to be complicated by  
radical uncertainties,  high risk second guessing, or any other new  
competitive traps, expectations about the future will be aligned in  
ways that might accelerate the pace and reduce everyone's pain of  
migration.

3(e= reject): This approach would assure that at, over time,  
progressively more growth will be accommodated by IPv6 rather than  
through de-aggregation. Doesn't solve the IPv6 routing scalability  
issue, but it does preserve the RIR as a potentially self-sustaining  
administrator for any ongoing/future number resource-related needs  
verification, as maintainer of the registration database, provider of  
whois, and a potential anchor for resource certification, et al. --  
and as a viable mechanism for continuing policy deliberation in the  
event that future routing scalability requirements require the same  
kind of coordinated action that helped to mitigate the last such crisis.

New-5: Clean, easiest possible reversability - Unlike the transfer  
proposal, if this approach turns out to have perverse unanticipated  
consequences, it could be terminated or even reversed with a  
(relative) minimum of disruption.

Potential Downside: if IPv4 is permanently warehoused, then any  
potential revenue arising from IPv4 sales would be foregone. If IPv4  
prices are expected to be modest (e.g., modest enough to avoid  
antitrust scrutiny), then perhaps any loss would be equally modest.  
Alternately, the foregone one-time sales revenues could be thought of  
as investments toward a better, NAT free (or NAT-vonlutary-only)  
future. If those future payoffs are deemed to be insufficient,  
returned address space along with cash proceeds from ongoing IPv4  
"recirculation" could be redistributed to community members, but this  
would probably impose substantial new administrative burdens and risks  
on ARIN. However, the second option is neither required nor recommended.

It may sound crazy, but I think it might be among the least crazy  
options open to us at this point.


Reactions? Perhaps others would like to walk 2008-2 through the same  
kind of explanatory process?


TV, speaking for self alone

On Sep 3, 2008, at 3:46 PM, Chris Grundemann wrote:

> Great idea David, I agree 100% that defining / mapping the argument
> space will help everyone to better understand their own position as
> well as the positions of others.
>
> Maybe I am making this more complicated than it needs to be but for me
> to really wrap my head around the entire argument space, I had to take
> a graphical approach. For those interested, it can be found here:
> http://odin.chrisgrundemann.com/Do_I_Support_A_Liberalized_Transfer_Policy.jpg
>
> I am open to suggestions on how this drawing could be improved,
> utilized, discarded, and/or converted to text to complement the list
> you are generating in this thread.
>
> ~Chris


-------------- next part --------------
A non-text attachment was scrubbed...
Name: Grundeman_Argument_Space_080903.jpg
Type: image/jpeg
Size: 186879 bytes
Desc: not available
URL: <https://lists.arin.net/pipermail/arin-ppml/attachments/20080904/79aa8160/attachment.jpg>


More information about the ARIN-PPML mailing list