[arin-ppml] Draft proposal that needs some wordsmithing

Owen DeLong owen at delong.com
Fri May 6 15:57:59 EDT 2011


On May 6, 2011, at 11:13 AM, Mike Burns wrote:

> Hi Owen,
>>> Because prior to exhaust there was no other mechanism to ensure addresses were allocated and used efficiently.
> 
>> If a market was the best mechanism, we could have adopted policy to auction
>> IPv4 addresses off to the highest bidders before exhaustion. Exhaustion would
>> likely have come much earlier in such a case, but, it was possible, so, your
>> assertion that no other mechanism existed is false. No other mechanism was
>> chosen, but, it did exist.
> 
> I think arguments about who would have kept the auction money would have ended that option pretty swiftly.
> 
> 
>>> Prior to exhaust, without such a mechanism, I could have walked up and asked for a /2.
> 
>> What prevents you from doing so post exhaustion in an unmoderated market if you think
>> that purchasing all (or as much as you can) of the available address space will have
>> a sufficient negative impact on your competitors as to be worth the investment?
> 
> Nothing prevents me from asking. In the prior case, I would have been turned down for lack of need.
> In the post-exhaust case, I would have been turned down for lack of money (or supply, obviously).
> I doubt that there will be a /2 for sale any time soon, but maybe in the End of Days for IPv4.
> (When the value approaches zero.)
> 
Sure, a /2 is a ludicrous example in both cases. (ARIN never had a /2 to give you in the
pre-exhaustion world, either).

However, at /6, /8, or even /12, this still remains meaningful and gets progressively more
realistic as a way to engage in anti-competitive behavior.

>> The system that was devised and implemented was a needs analysis which simply made allocations according to demonstrated need.
>> The ensured that at least at the start, addresses would not be frivolously wasted.
> 
>> It also ensured that address registrations did not become an artificially valuable
>> commodity. Exhaustion will force this artificial valuation on such registrations, but,
>> removing the justified need restriction from this forced market will not help
>> the situation as it would still have the same effect of allowing waste. Admittedly,
>> the degree of frivolity may be more a matter of perspective at that point, but,
>> waste through capture for competitive advantage is still waste as far as I am
>> concerned.
> 
> Is it your contention that the current system has not allowed waste, or just that a private market will allow for more waste?
> 
It is my contention that the definition of waste is often in the eye of the beholder,
but, if you define waste as addresses which are not assigned to machines or
assigned to machines without a purpose considered useful by at least some
person and related to the address being used to provide or consume a service
over the network, then, it is my argument that the amount of waste allowed by
a market with justified need would be less than that allowed by an unrestricted
market.

> 
>> We know the needs requirement was not a perfect way to ensure efficiencies.
>> We know that from the number or allocated and not advertised space, if nothing else.
> 
>> No, we do not. Allocated and not advertised DOES NOT mean underutilized.
>> There are many legitimate reasons IP resources may be unadvertised while still
>> fully utilized (or more accurately, utilized but not visible in any routing table to which
>> you particularly have access).
> 
> I am of the opinion that there is a lot of waste in legacy space, and only slightly less waste in RSA space.
> Everyone here can make their own decision on that using their own experience to guide them.
> This is all space that has been allocated outside functioning market environments.
> 
I have no reason to believe that a market without justified need requirements would
somehow magically reduce waste vs. a market with the requirement that recipients
have justified need for their resources. After all, any recipient with justified need for
the resources they are receiving is non-waste almost by definition, where this
remains largely undefined for a  market with no such requirement.

An unrestricted market depends on the price of the address to reduce waste. However,
unless you consider that tying addresses up to keep them from benefiting your
competitors is not waste (I believe it is waste), there is no reason to believe that
such an unrestricted market would not create exactly that type of transaction.

>> A market will not be perfect either, but unlike the prior needs analysis, we seem to be judging the free market by the exceptions.
>> 
>> Since the misdeeds (which you claim will be "exceptions") in the market have the
>> potential to overwhelm the market, where no such risk existed in the needs
>> analysis, I think that is  a legitimate approach.
> 
> The stewards of APNIC decided otherwise, and your assertions, like mine, about the dangers of allowing free individuals voluntarily engaging in trade, are unsupported.
> 
The APNIC community decided that abandoning needs basis might serve their
community better. The other four RIRs have elected to preserve needs basis.

My statements about the dangers of allowing unrestricted trade are well documented
in other markets where they have occurred. Can you point to examples of completely
unregulated markets where such misdeeds have not occurred over any substantial
period of time?

> 
>> 
>>> 2) Why would any organization with need for unique IPv4 addresses
>>> choose to not have those addresses recorded in the database which
>>> guarantees their value in order to escape stating their need? (i.e.
>>> What class of organization with legitimate need would be hurt by
>>> having to demonstrate that need before receiving addresses?)
>>> 
>> 
>> An aggregator buying unroutable bits to aggregate to a routable size?.
> 
>> I see no reason this couldn't be done by an organization with need just as effectively
>> as by some random aggregator intending to resell the result.
> 
> What if his need is to buy snippets of space in order to aggregate it into sizes acceptable to the network operator community and then sell them?
> How would he describe that need to ARIN?
> 
That isn't a need supported by policy. If he needs an aggregate acceptable
to the operator community for his own use, OTOH, I think he could easily justify
those multiple purchases to ARIN under current interpretation of 8.3.

I don't see where the reseller provides any benefit to the community over the
end organization being the one to do the aggregation.

>> Somebody who has a different view on the IPv6 transition timeframe and has a longer planning horizon for IPv4?
> 
>> Then they come to the market multiple times.
> 
> Each time involves a cost of waiting, an uncertainty cost of the addresses in the future which some companies may find unacceptable, supply uncertainty,as well as transactional  and deaggregation costs.
> Is good stewardship to result in increased costs for consumers of  ip address space?
> 
Yes... The future of IPv4 is uncertain and will get progressively more uncertain as time progresses.
This is the reality of betting your business on continued availability of IPv4.

Providing greater or longer durations of certainty to some organizations at the cost of denying it to others
is contrary to good stewardship.

Good stewardship is keeping the cost playing field relatively level for all players over time.
At least to the greatest extent possible.

>> A reseller of vanity addresses, like 100.100.100.100?
> 
>> I see no reason to promote or create these. They offer no meaningful benefit to the
>> community at large.
> 
> Is it ARIN's role to decide this? Doesn't the history of the Internet suggest allowing volunteer private organizations to make these kinds of decisions?
> 
It is the community's role to decide this. As a member of the community, I have offered
my opinion. Others may offer theirs. At the end of the day, ARIN policy should reflect
the collective judgment of the community.

There is no history of volunteer private organizations making these decisions with
respect to number resources on the internet and in places where number resources
have been allowed to be managed in such a way, careful regulation has been required
in each case to avoid substantial problems that have occurred in its absence. It is a
complexity which I do not think we should take on, give its limited value to the community
as a whole.

>> A wholesaler of addresses who caters to those who need instant availability (needs analysis takes time)?
> 
>> My last needs analysis took less than 24 hours. The average of my last 5 needs analysis is less than 36 hours.
>> That's real-time, not resource analyst or my hours.
> 
> Microsoft got turned down recently from APNIC for a temporary allocation for some technical symposium in Australia.

So?

> Why can't they look to a wholesaler for rapid, and maybe temporary deployment?

In the APNIC region, under their policies, presumably they can. In the ARIN region, they cannot because policy
prohibits such wholesalers. What policy does, however, allow, is "matchmakers" who could serve the same purpose
without possessing the address registrations in the interim. I believe the STLS term for them is "Facilitators".

Given the ability to have facilitators, what need is there for wholesalers?

> Surely you can understand that transfers will likely increase in number, and ARIN's needs analyses could take longer?

I don't think that transfers will increase beyond current allocation/assignment request rates (modulo normal growth)
unless the market is becoming excessively liquid which would indicate that we have some level of failure in policy.

> But no such wholesaler can exist if you require him to demonstrate need.

Nor is he needed or useful in my opinion.

> Think of a Quik-E-Mart for IP addresses. Are you certain there would never be a call for that kind of convenience?

The visual of purchasing a Big-Gulp full of IP addresses at the local 7-11 is just too much for me to take seriously.

> What if there are supply problems on STLS? A wholesaler with inventory on hand is often a valuable resource in times like that, even though he is going to make a buck on the deal.
> 
How is a wholesaler with inventory different from a facilitator as defined in STLS in any meaningful way?

The facilitator can still make a buck on the deal. If there's a supply problem, it's because there isn't a supply,
not because policy prevented people from providing supply.

>> A speculator, who could have a positive role in free markets?
> 
>> ROFL -- The concept of speculator in the same sentence with "positive role" amuses me.
> 
> Suggested reading for when you arise from the floor
> http://mises.org/daily/320
> 
After a long winded explanation of how he can cloud the definition of speculation to mean virtually
any form of investing, he finally gets to the point of claiming that there is a benefit to the market if:

	1.	Prices adjust quickly
		-- In the case of speculators this almost always means "increase"

	2.	There is greater liquidity
		-- Though he makes no mention of how, exactly, speculation actually increases liquidity

	3.	The market fluctuations are minimized
		-- Which is interesting to attribute to speculators after attempting to attribute case 1 to them as well.

In my observations of various markets, speculators have had a destabilizing influence with a general
tendency to increase prices, often far above and beyond rational levels.

Yes, the speculators at the end of that process when the bubble bursts usually get hurt badly, then, we
the taxpayers get to pay to bail them out, too.

>> An organization that does not want to undergo an ARIN analysis for fear it will lead to a review and recovery procedure?
> 
>> An organization which has reason to fear this is an organization which probably shouldn't
>> be getting additional resources from the community.
> 
> They would be buying them from the rights holder, not getting them from the community.
> 
The resources belong to the community. The current resource holder is holding them in trust.
They are not property.

> 
>> An organization from another region?
> 
There is no policy to support inter-regional transfers currently and your proposed policy
would not inherently create them.

>> You say this as if it is somehow a benefit.
> 
> I was asked by Chris why anybody would transfer addresses and fail to have them registered, and these were just examples.
> I don't think this is a benefit, but I support a global free market for IP addresses, so they can flow to wherever they are needed most, as measured by their price.
> In this way I feel I am extending the definition of community wider than the region of abode.
> 
I do not support using price as the sole measure of the need for addresses. There are many cases where I think
that, for example, that providing services to <insert 501(c)3 organization> is a far better use than insuring that
<insert large monopolistic telco> can make sure that their competitors feel the pinch of address exhaustion
well before they do.

Measured by money, <insert large monopolistic telco> is almost certain to win.

>> A buyer of a /24 who thinks an ARIN needs analysis isn't worth the expense?
> 
>> Again, not seeing the benefit to the community in providing this person the opportunity to take
>> that /24 out of the hands of some more deserving organization with documented need.
> 
> You miss my point, they may have need, but for a small transaction, having to negotiate ARIN hurdles could be viewed as unworth the effort. Again this is in response to the request for examples of potential buyers who would not take the steps to register.
> 
For a small transaction, the cost of a needs analysis is also small. As someone with rather
extensive experience producing needs-basis justifications for submission to ARIN (and
some experience with other RIRs), I can say with certainty that the needs-basis justification
for a /24 is very simple and does not provide a significant cost to the equation.

>> Microsoft? They didn't seem to want or need a needs analysis until ARIN began negotiating with them after the original asset agreement with Nortel had been negotiated.
>> 
> 
>> This, also, strikes me as an indication that removing needs basis would have a negative impact
>> on the overall outcome.
> 
> ARIN ran in and got the transfer reflected in whois through shenanigans with the needs justification, in my opinion. If there were no needs requirement, I think Microsoft would have asked ARIN to make the updates to whois as the normal course of business, but without knowing how accomodating ARIN would be on a needs analysis, they pointedly left ARIN agreements out of the first negotiated asset sale with Nortel.
> 
Your opinion is noted, but, I don't concur with the characterization you make of the situation.

I think you state a number of assertions in that paragraph with no factual basis to back them up.

> 
>> I don't pretend to be able to able to identify all the types of transactions for which an ARIN needs analysis seems an unnecessary intrusion into a transaction between two private entities.
>> 
>> What you call unnecessary, I call vital to the overall interests of the community.
> 
> It was vital when there was no other free and voluntary mechanism to ensure efficient use, but I am trying to show that the needs mechanism is now outdated and poses a problem for whois.
> 
Money does not insure efficient use by any definition of efficiency that I find acceptable.

Measuring efficiency by money is sort of like measuring electrical consumption of
a house by measuring the temperature rise in the upstairs bedroom. Sure, the electrical
consumption in the house will definitely contribute, but, you won't get anything near
an accurate measurement.

>> The point is that many prior transfers have taken place, particularly with legacy space, that have not been reflected in whois.
> 
>> Hopefully as these can be identified, the space can be revoked and reallocated
>> to organizations that comply with policy. The original legacy holder has some
>> protections. A third party as a result of an unauthorized transfer should not have
>> any protections in this regard.
> 
> Microsoft was the third party here. Addresses were transferred to Nortel from their acquisitions, the original legacy holders here.
> By your definitions, since ARIN was not involved, these transfers were unauthorized.

Yep.

> And yet a bankruptcy judge saw no problems with Microsoft buying them from Nortel.

I agree that it is unfortunate that the bankruptcy judge did not see fit to consider the
community with greater weight.

> I haven't stressed this, but the legal facts as I see them are that legacy holdings can be transferred without ARIN notice or needs requirements.

As I see it, this remains unclear. Each of the cases so far has carefully not decided this
particular point.

> By removing needs requirements for transfers, we bring policy in line with developing law, and this is sure to reduce future conflict.
> 
I would rather preserve conflict and try to develop the law in a more useful direction.
I guess you can consider this part of my right to petition the government for redress of grievances.

> 
>> One of the problems relates to the requirement for a needs analysis.
> 
>> No, the problem is the belief that community resources can be transferred outside
>> of community policy.
> 
> For legacy addresses, these transfers have occurred in reality, it's time for belief to catch up.
> 
As I said, as these blocks which are being used by entities other than their legitimate holders
are identified, ARIN should reclaim them and reissue the resources to organizations within the
policy process.

That is reality. It is time for ARIN to start becoming more aggressive about identifying them
in my opinion.
> 
>> If a holder of legacy space acquired through an asset sale approached ARIN to reflect that transfer, ARIN would not update whois without a needs analysis.
> 
>> As it should be.
> 
> You are arrogating needs requirements over whois accuracy by taking that stance.
> 
No, I am taking the stance that the better way to ensure whois accuracy is by identifying
blocks being hijacked by organizations to which they were not issued and reclaiming them
and providing them to members of the community in compliance with policy.

> 
>> In addition, the requirement for ARIN to do a needs analysis and the potential for review and recovery on either the buyer or seller increases the FUD factor in the market.
> 
>> Only for those attempting to circumvent policies constructed by the consensus of the community.
> 
> If I want to buy and the seller want to sell, and we have reached an agreement on price, then having to undergo an audit before we can process the sale, or being subject to one thereafter,  is most certainly an added uncertainty.
> 
So what. You're trying to trade in community resources that are held in the trust of the
community for a particular purpose. The community has a right to audit your use of them
and ensure that it is in compliance with the policies of the community.

The presence of police on the highway adds an element of uncertainty to my ability
to drive at any speed I prefer. I don't necessarily see that as a bad thing, even though
it was very inconvenient on several occasions in my younger years.

>> For a market to function efficiently, Fear, Uncertainty, and Doubt need to be assuaged, and this proposal does that.
>> 
>> I have tremendous fear and uncertainty about the effects of this proposal. I doubt that it will function as
>> advertised. Indeed, I believe this proposal increases each of those things from my perspective.
> 
> So you see how FUD works to prevent action.
> 
More often, I see how it is used to provoke incorrect or counterproductive action, such as the PATRIOT
act.

>> I copied liberally, almost entirely, from the APNIC policy to allow needs-free transfers. The rationale which was most effective in that regions's deliberations may have been the concern that by imposing the needs requirement, transactions would be more likely to occur outside the system, leading to a decay in whois reliability.
>> 
>> That is the argument Geoff used which appears to have had sway in that region.
> 
>> Geoff has repeatedly made that argument in the ARIN and RIPE regions (and I'm not sure
>> that he has not made it in LACNIC or AfriNIC as well). So far, it has not been found to be
>> convincing outside of APNIC.
> 
>> By structuring my proposal in this way, I am trying to get people to consider whether the original and laudable needs requirement should be maintained when keeping it could lead to whois degradation.
> 
>> This question has been asked and answered as part of the debate around 2008-2, its successor
>> 2008-15 (IIRC) and the boards reconstruction of that into 2009-1. You are welcome to ask the
>> question again, but, I'm not inclined to believe the answer has changed.
> 
> Things have changed since then in terms of continued failure to transition and the MS/Nortel deal, and APNIC reaching exhaustion and approving their new transfer policy.
> 
Uh, APNICs current transfer policy was in place prior to 2009-1. If you're talking about their inter-regional
transfer policy, that's new, but it doesn't really support a removal of needs basis from the ARIN region.
If anything, it makes it even more vital.

I don't see how the continued failure to transition differs from the expectations that were considered at the time we were debating 2009-1.

Finally, I don't really see that the MS/Nortel deal changes anything other than indicating that we may have been insufficiently specific
with the restrictions expressed in 2009-1 (NRM 8.3) and might need to tighten the language so as to place better constraints on staff's action more
in line with intended policy.

As I said, you are more than welcome to ask the question again.

> 
>> My argument is that  proper stewardship recognizes the existence of a market which will fulfill the original stewarship role of ensuring efficient use, and we can direct our stewardship best to policies which help to ensure whois veracity.
> 
>> My argument is that the market alone is not a good steward and a regulated market is necessary
>> to ensure the vital interests of the community.
> 
> And I counter that the vital interest in address-use-efficiencies are better offloaded to the market, and the vital interest in maintaining whois veracity is retained.
> 

Yes, your argument that it is better entrusted to the market is precisely the point where I think we have the strongest disagreement.

Owen

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.arin.net/pipermail/arin-ppml/attachments/20110506/a7d8a55c/attachment.html>


More information about the ARIN-PPML mailing list