At the RIPE 66 meeting in Dublin, the bold, long-overdue proposal to eliminate needs assessments for IPv4 number allocations got an interesting response. RIPE-NCC is the Regional Internet Registry for Europe, and almost everyone in Europe supported the proposal. Vocal opposition came from two policy tourists from North America: ARIN Board member Bill Woodcock and ARIN Advisory Council member Chris Grundemann.
The author of the “No Need” proposal, Tore Anderson, argued that because the free pool is depleted, RIPE’s conservation principle no longer applies. In other words, there is no longer any reason to use technical needs assessments to conserve the allocation of the resource. If someone has more addresses than they need, they can sell them or transfer them to someone who does need them. There is no reason to interject a complicated and expensive bureaucratic process to determine whether someone “needs” the addresses before they buy them.
Chris Grundemann argued passionately against Anderson’s ideas, both on the floor at RIPE 66 and in a blog post. But his argument took an odd form. He claimed that the need for conservation has not been eliminated by the depletion of the free pool. He wrote, “Now that almost all of the public IPv4 address space has moved from RIR pools into the ‘wild,’ there is arguably a much greater need to practice conservation. The loss of the RIR free pool buffer does not mark the end of ‘the lifetime of the public IPv4 address space.'”
They are both right. Indeed, the arguments do not contradict each other at all.
We do need conservation. But we don’t need administrative needs assessments.
Grundemann’s attack on Anderson’s proposal was framed as a defense of traditional needs assessment, but he missed the target. He believes that the need to conserve the IPv4 address space justifies continued application of traditional needs assessment practices to IPv4 transfers. But the leap from conservation to needs assessment is a non sequitur. Needs assessment is not the only form of conservation. It is not even the best form of conservation. The market price system acts as a perfectly efficient and effective form of conservation. It makes people pay the actual scarcity value of the resource. Technical needs assessment is a poor substitute for it. Indeed, market prices conserve more rationally and more fairly than the administrative processes. As the resource becomes more scarce, the price goes up. It doesn’t impose arbitrary time horizons on business planning. If the organizations in the internet industry want to extend their time horizon and guarantee themselves of the resource over a 5 year time span, they can do so, provided they are willing and able to pay the price for a 5 year supply. If they want to be more careful and acquire only a 1 year supply, they can do that. Needs assessment, in contrast, imposes an arbitrarily fixed time horizon – e.g., 2 years – on the definition of “need.”
So in order to defend needs assessment, Grundeman not only has to argue that conservation is still needed. He also has to prove that there is something fundamentally wrong about using market prices to ration the resource. That’s where he fails. It’s excusable, because neither Grundemann nor Woodcock are economists* and both are ill-prepared to make coherent arguments about that topic. Neither of them presented a convincing case about the nature of the number market. Instead, we were given vague references to “bad actors,” “stockpiling” and “hoarding.”
There are three responses to this argument.
First, there’s the dirty little secret that Grundemann and other defenders of the status quo repeatedly ignore. The biggest cause of “stockpiling and hoarding” is not and never has been market allocation of IP addresses. The biggest cause is the existing system of common pool governance. That system had already coughed up 41% of the address space by 1991, just as the Internet was getting started. That system privileges large ISPs who can deal with the uncertainty and bureaucracy of needs assessments because they have long experience playing the ARIN game and filling out complicated forms to get addresses. That system provides organizations who hold allocations no incentive to return them, whether they need them or not. That system imposes so many costs and restrictions on the acquisition of addresses that it encourages everyone to ask for and keep more than they need, whenever they get the chance. In fact, only the emergence of transfer markets has started to rectify the historical misallocation of IP numbers, by giving holders of too many addresses an economic incentive to give them up and sell them to someone who does need them. Currently, about 35% of the address space is unused – in other words, there is already a ton of “stockpiling and hoarding.” Needs assessments, by making market transfers more difficult, helps to freeze this situation in place. And yet this is the system Grundemann defends and resists changing.
Second, Grundemann spends so much of his time in the ARIN policy playground that he seems to have forgotten how the real world of business works. Probably 95% of the resources that currently go into the production of Internet service are allocated by the market. Routers and equipment, telecommunication services, office space, etc., etc. – all are allocated on a “buy what you think you need” basis. How and why are addresses different? If Chris’s company wants to rent office space in Denver, no one performs a needs assessment before they will sell it to him. They do not say, “hey, Chris, you are asking for 12 rooms, but we think you only need 10.” Does anyone really think the market for real estate should operate like that? Office space is fixed in supply in the short term, just as IP number blocks are; there is no economically relevant distinction there. We might add that Grundemann’s salary, like that of nearly all other network managers, is determined by the market. If he wants to switch jobs there are no administrative agencies who have to make an advance determination that the company hiring him really “needs” him. Don’t you think its better we don’t do that? What is the reason – other than “we’ve always done it this way and at ARIN we’ve turned this way of doing things into a religion” – that IP numbers should be handled differently now that the free pool is occupied?
There is a third and final argument why we don’t need needs assessment: fears about speculation and hoarding driving the market are completely without merit. This is pretty obvious once one examines the economic characteristics of the IP address market. Louis Sterchi, one of the participants in the RIPE debate, has written a blog post noting that speculators do not enter these kinds of markets, and given the fact that a huge portion of the space is already held by many, many large and small organizations who are using and will not give up their numbers, it would be impossible for any speculators to corner the market and have a strong impact on price.
Independent academic research has shown that the transfer market has worked well at moving number resources from underutilized holders to people who need them. Speculators are not present. While many of these transfers have been subjected to the arbitrary intermediation of RIR needs assessment, one major transfer – the Microsoft – Nortel deal – was sealed in bankruptcy court prior to ARIN approval. No real needs assessment was involved (yes, I know we will get an obligatory objection from ARIN in our comments section, but who cares – the evidence is conclusive). That deal, like all the others, involved real users of the address space.
The debate over RIPE policy proposal 2013-3 now moves from the public meeting to RIPE-NCC’s Address Policy Working Group (APWG) mailing list. Let’s hope reason prevails.
* Woodcock’s ignorance of economics is is painfully obvious, but that never stops him from expressing opinions. In the RIPE-66 transcript he refers to the Chicago School as a school of economics not respected or taken seriously, apparently overlooking the fact that 3 members of that school are Nobel Laureates.