Discussion of IPv4 Depletion and Resource Transfer Policies

This wiki page contains discussion of IPv4 resource transfer policies as they may obtain in the wake of IPv4 free-pool depletion. If you have a point you'd like to introduce into the discussion, please either email it to Bill Woodcock woody@pch.net or Tom Vest tvest@eyeconomics.com, or use the "Login" button at the bottom of the page to create an account for yourself, and then email Bill to get the account authorized for write-access to this page.

Our method will be to add new text to the body of this page, discuss it and potential edits in footnotes, and when consensus is reached on an edit, apply the edit to this page and move both the prior text and the footnotes which refer to it to off into a new subsidiary page.

The contents of this page and any sub-pages within the "pch:public:ipv4transfer" hierarchy are free for reuse and adaptation without attribution under the Creative Commons Sampling Plus 1.0 License. None of the contents may be attributed to any individual author without their explicit permission.

Creative Commons License

Frequently-Asked Questions

Does today's limited deployment of IPv6 constitute a failure of private-sector self-regulation?

No. IPv6 offers an expanded range of addresses, and is intended to solve the problem of IPv4's limit of approximately four billion addresses. IPv6 offers no benefits other than additional address space, so there is no reason for the industry to deploy IPv6 in advance of IPv4 depletion, other than experimentation and "gaining experience with it." That said, the transition is not an instant one, particularly for large networks, so many of the largest ones are already several years into their multi-year IPv6 transition processes. The industry would accrue no benefit from switching to IPv6 "early," and in an industry in which many participants feel they must double in size each year in order to keep up with the overall growth of the Internet, the net present value of the time and energy needed to transition to IPv6 is very high. Thus, postponing the transition until the appropriate moment makes better business sense than performing it earlier than necessary.

Does a "market" in IPv4 addresses exist today?

Yes, a "black" market (by definition unregulated and not-officially-recognized) in IPv4 addresses presently exists, just as an unregulated black market in prescription drugs exists. As in any black market, evading regulation comes with a cost, both a higher monetary unit cost, and a risk of resource revocation or civil or criminal penalties.

What do IPv4 addresses cost today?

It's not possible to "own" an IP address, since it's merely a number, and its value lies in its //unique assignment// rather than being inherent to the number itself. Nonetheless, there are costs associated with the labor of performing and maintaining unique assignment, and likewise there are costs associated with evading regulation to use the black market. Today, if you were to divide those costs by the number of addresses held, this is what you would see1

ARIN minimum allocation (/22 for US$1,250): US$1.22\ ARIN maximum allocation (/8 for US$18,000): US$0.0011 ("Flatness" ratio: 1109)

APNIC minimum allocation (/22 for US$1,4842): US$1.45\ APNIC maximum allocation (/8 for US$47,509): US$0.0028 ("Flatness" ratio: 518)

RIPE minimum allocation (/22 for US$5,0163): US$4.90\ RIPE maximum allocation (/8 for US$11,387): US$0.00068 ("Flatness" ratio: 7206)

LACNIC minimum allocation (/21 for US$850): US$0.42\ LACNIC maximum allocation (/8 for US$33,000): US$0.0017 ("Flatness" ratio: 247)

AfriNIC minimum allocation (/24 for US$2,800): US$10.94\ AfriNIC maximum allocation (/14 for US$10,300): US$0.04 ("Flatness" ratio: 274)

Black market minimum block (/24 for ~US$3,000): US$11.72\ Black market maximum block (/16 for ~US$65,000): US$0.99 ("Flatness" ratio: 12)\

What does "convexity" mean in the context of IP address pricing?

Convexity, in this context, means the property that a block of address space has a higher price in the market than would the sum of its parts. Large blocks would have a higher per-address price than small ones. A convex market would be one way of assuring that sellers of address space would prefer not to disaggregate unnecessarily, and to maximize the number of large (and therefore probably scarce) blocks available to those who would like to be able to make large aggregate announcements. Preserving aggregation is one of the conservationist regulatory goals of the RIRs. This would seem to make market convexity inherently desirable.

So how can we cause "convexity" in the market?

Unordered Notes

Bill's basic principles:

Tom's random responses:

  1. Bill: Note that these comparisons can be made easily //within// each RIR, but are not strictly apples-to-apples //between// RIRs, due to differences in policy between regions, differences in recurring billing and "maintenance fee" models, et cetera. Notably, some of these fees may recur annually, depending upon the mode of the transaction. ↩︎

  2. Bill: Conversion rate 1 AUD = 0.94 USD as of March 1, 2008 ↩︎

  3. Bill: Conversion rate 1 EUR = 1.52 USD as of March 1, 2008 ↩︎

  4. Bill: use-based qualification of resource recipients is the basis of the current RIR private-sector self-regulatory scheme ↩︎

  5. Bill: Yes, yes, gross generalization, I know. ↩︎

  6. Bill: The latter, not the former. People qualified based upon need, as need was defined at the time they qualified. ↩︎

  7. Bill: I would say that they should be judged based upon the terms under which they qualified. Which means yes, and no, respectively. They couldn't be speculators if they didn't "buy low" for the purpose of "selling high." ↩︎

  8. Bill: Why do you think such people would exist in any significant number? The purpose of the RIR analysts is precisely to make sure they don't. ↩︎

  9. Tom: I think this response presumes that legitimizing decentralized resource transfers will not affect community policy compliance rates -- i.e., that compliance levels will remain just "as if" there were still a single, unitary delegator for all number resources. This seems quite optimistic, to put it mildly. ↩︎

  10. Bill: I'm not sure about this... It seems like it's common //in the absence of regulation//, but "when alternative suppliers can be excluded" seems to me like a circular definition. ↩︎

  11. Tom: hypothetical example: large "within-industry" player that seeks to secure additional address resources outside of the approved transfer process, perhaps to capture the savings afforded by convex priving. Real world examples: national/terrestrial incumbents that purchase 100% of cables landing on home market that they had previously only had minority interests in; regional terrestrial operators that purchase formerly independent platforms that had once provided means for customers and competitors to bypass them -- in both cases to decrease rather than increase capacity for sale. Granted, it's a circular definition -- but it's also the objective of much real-world business strategy in many if not all commercial sectors. ↩︎

  12. Bill: Only if it's useful, and I'd need to be convinced that the category wouldn't be an empty set, to believe it was useful. ↩︎

  13. Bill: If pricing is "convex" would there ever be a barrier to new entrants that was higher than the barrier to larger players? ↩︎

  14. Tom: This is an interesting concept, but it's hard to imagine it emerging naturally except at the lower extreme, e.g., $(1/24) >> $(2/25). Assuming convexity did stick, it's even harder to imagine what would deter industry players from trying to capture substantial savings by scooping up multiple smaller prefixes in the gray market -- and thereby crowding smaller players out of the market. Finally, even if convexity sticks but does not encourage defection, what mechanism will cause it to relax over time so that it doesn't become a different kind of barrier to entry and growth, e.g., as routing capacity grows? ↩︎