[ppml] Policy Proposal: Expand timeframe of Additional Requests

Paul Vixie paul at vix.com
Thu Aug 16 15:35:13 EDT 2007

> ... IP numbers have no value to a hoarder unless they can sell them.  So the
> hoarder is going to obtain as much as he can then as soon as IPv4 runout
> happens he's going to try to sell as much as he can as soon as he can -
> since the value of IPv4 will be highest right after runout.

i wish that i agreed.  even leaving aside the fact that IP addresses aren't
property and that any market in them is a black market until and unless the
community generally agrees to loosen the ties that bind us to RFC 2050, i'm
struck by my utter disbelief that hoarders will dump.  when you've got a
pile of stuff that costs you nothing to hold onto (no upkeep, so this isn't
like pork bellies), you will trickle it out into the market at a pace designed
to keep prices high.  if we postulate N hoarders and demand Q per interval T,
each hoarder will have the quota of (N/Q)/T, knowing that if they exceed this
they will not maximize their profit.  see the RAM chip market for examples,
except that in RAM chips the value of your hoard goes down with innovation --
having a lot of 256K chips is a burden when 512K chips come out, etc.  IPv4
addresses, if the hoarders played their cards right, could increase in cost
over time, since IPv6 is such a dark horse.

> This will simply have the effect of delaying IPv6 switchover for the rich
> organizations that can afford to buy from hoarders.  In other words,
> hoarding does not really take IPv4 out of circulation, since the hoarder has
> to see it put in to circulation to get paid.

a hoarder has to limit the amount put into circulation to get paid "the most."
IOW, they would only dump in the months before IPv6 appeared to be viable, and
that could still be "never" if the community doesn't get some coherency on it.

> And the side effects would be terrible.  Almost certainly it would prompt
> court cases and the like that could possibly produce caselaw in some
> jurisdictions of considering IP addresses property, thus ARIN might lose
> control of the assignment authority.  That is why we would be foolish not to
> do what we can to make the environment for "ip v4 sales" to be very
> difficult for a hoarder to operate in.

on this point, i agree, with the proviso that ARIN's authority to allocate
would become moot if there is nothing to allocate, and ARIN derives its
authority from the community, who may yet cohere on the topic of address
property rights.  RFC 2050 describes a "right to use" on the "basis of need".
perhaps if there's nothing left to allocate, the community will decide to
expand this to a "right to use and/or dispose" which is the technical
definition of property in most of the world.  (dispose doesn't mean
incinerate, it means do something that means you don't have it any more, like
sell it.)  that's a global community decision not a regional community
decision.  i don't know if it would be possible to rev RFC 2050 in today's
world, but i expect some RIR bylaws somewhere might also have to be amended.

also, on this point, the community seems thus far quite divided.  folks who
expect to keep needing more new IPv4 addresses after central pool depletion
seem to favour the idea of making it "legal to buy them" (and therefore,
"legal to sell them").  folks who expect to be able to renumber or to use NAT
or to become more efficient seem to favour the idea of making it "legal to
sell".  on the other hand, folks who know that the highest and best use of
property is to buy low and sell high, speculate, and subdivide, are concerned
about the digital divide (can afford addresses vs. not), globalization (a
village ISP in africa might be able to make the equivilent of a year's profit
by selling their IP address block in new york), routing table growth due to
subdivision, and probably other things i'm not thinking of at the moment.

> > There have been approximately 100 legacy /8s assigned.  This is more
> > than the total allocations made to all the RIRs, combined.
> I keep waiting for someone to point this out when we get into these
> legacy holder flamefests.  Sigh.

would the ARIN community agree to pay higher fees to build a cash hoard that
ARIN could then use to "buy back" legacy space which would then be given to
IANA to hand out on a "right to use, based on need" basis?  (i'm not proposing
this and i'm definitely not speaking as a trustee when pondering it, but i am
curious to hear some answers.)

> ... no matter what you do, whether you stretch out or shrink the time before
> IPv4 runout, the fact is that runout will happen eventually and behaviors of
> ISP's will change.  Even if hoarding and IP sales come into play, all it
> will do is put off the day of reckoning.

yea, verily.  tell it, brother.

> > This does not appear to have impacted the uptake of IPv6.
> IPv6 really hasn't been introduced.  IPv6 is currently nothing more than a
> set of technical specifications without an implementation for the MAJORITY
> of operating systems in current production use.
> Once we see all Windows operating systems that are older than Vista drop
> down to 10% of the production installs in current use, we can consider that
> IPv6 has been introduced.

if you tell folks they can't have anything more than web access unless they
upgrade to Vista, most will say that's fine, web access is all i need.  let's
not focus on the client platform in these discussions, that's a CPE/NAT-PT
question for which v6ops at ietf is more suitable.

More information about the ARIN-PPML mailing list