[arin-ppml] On whether morality can be the lone argument against a transfer market
tvest at pch.net
Wed Oct 1 15:10:39 EDT 2008
On Oct 1, 2008, at 11:53 AM, Scott Leibrand wrote:
> You argue that the main benefit of the last system was that it
> allowed new entrants into the market.
No I don't. I argue that the system that may now be in its last days
worked for everybody. It made de facto global IP transit -- the one-
stop shopping option for all customers and all non-DFZ routing service
providers -- possible... and possible without closing the market to
new entrants. That was a huge benefit for everybody, and huge benefit+
for those who were/are able to sell domestic IP transit, and a huge
benefit+++ for those able to sell international IP transit, which
really was/is a historically unprecedented market opportunity.
Technically, IP transit was/is a by-product of the "economies of
scale" that were made possible by the combination of
+CIDR -- which provided the basic tools.
+ top-level aggregation -- which the RIR community-system provided,
and kept flexible over time as technology improved and RIR community
+ filtering -- which was only commercially feasible because of the top-
level aggregation made possible by the RIR system.
+ open entry for new routing service providers, which the arms-length
RIR processes also enabled, and which effectively made aggregation and
filtering "justifiable" and thus palatable to most direct
stakeholders, as well as to the few indirect stakeholders/outside
observer who knew that the system existed and understood the basics of
how it worked.
That last "open entry" bit was what effectively made new markets --
e.g., aspiring new RIR communities -- want to opt in to the growing
system, and what encouraged national regulators, who were generally
neither accustomed nor willing to abide that level of extra-
territorial service delivery, to sit on their hands.
> How do you propose that we allow new entrants to the market after
> IPv4 exhaustion?
I will spare the list another recital of details I've already shared
I propose that we allow new entrants to the market after IPv4
1. Preventing Pv4 exhaustion from happening for new entrants, through
mechanisms like 2008-5 at minimum -- if not more vigorous initiatives
like the scale-sensitive "contribution scheme" that I've suggested in
2. Gradually reducing the non-substitutability / critical demand for
IPv4 over time, e.g., through a commitment to rapid and ultimately
universal dual-stacking of all "important" public-oriented Internet
resources -- if not more vigorous initiatives like the scale-sensitive
"contribution scheme" that I've suggested in the past.
> I would think that a transfer policy would help new entrants, as
> they would still be able to get their own PI space, through
> transfer, rather than being stuck with PA space from whatever
> provider still has addresses to rent.
Is that really realistic? The current demand for transfer markets is a
combination of some mixture of (a) incumbents who are apparently
willing to spend a lot to delay or perhaps avoid IPv6 altogether, plus
(b) surplus IPv4 holders who are not willing to part with their
surplus resources unless/until someone pays them a lot to do it. As
soon as a market is legitimized, they might or might not be joined by
(c) other incumbents that are willing to cannibalize their own
critical/production IPv4 assets, despite the uncertainty of the future
IP addressing base, in order to secure a quick, one-time payout. They
will almost certainly be joined by (d) new speculators who will be
looking for every opportunity to buy early/cheap and hold out until
they can sell at a much higher price.
Speculators are an intrinsic part of every market -- so those who
imagine that they can somehow be avoided in this case should probably
"get over it," as they say. But even if you manage to create the first
speculation-free market in human history, I doubt that it will matter.
Give (a+b), why should anyone assume that the "reservation price" for
IPv4 transfers will actually be attainable for new entrants, much less
be able to pass the anti-trust sniff test -- with or without any added
> The only other approaches I've seen any support for are rationing
> policies, whereby a block of addresses is reserved to allow new
> entrants to get at least a little bit of space.
If "rationing" means "more than you want," then we've been doing it
for about 15 years. If it means "more than you need," then we haven't
-- but only because we've been actively defining and redefining what
"need" means on a rolling basis, in response to real changes (most of
which we also *caused*) in the wider world. Given that long history of
community awareness of and responsiveness to past real-world changes,
the failure to even consider policy actions that might help to reduce
the actual need for IPv4, for incumbents and especially for new
entrants, is not confidence inspiring, to say the least.
> I think such a policy is a good idea, but it only helps the smallest
> new entrants, or those who choose not to compete for customers
> requiring IPv4 space.
One key rationale for the reservation policies is to make it at least
theoretically possible for successive generations of new entrants to
internalize their own IPv6-IPv4 translation requirements. That way,
maybe -- i.e., if the reservation is large enough to last for many
decades and/or many many thousands of new entrants -- fewer current
surplus IPv4 holders will hold out in anticipation of perpetual demand
and ever-escalating prices for IPv4. Maybe, in combination, the
greater (albeit limited) potential for IPv6-based market entry, plus
the reduced certainty of big payoffs for continued strategic hoarding,
will actually help to cause the "need" for IPv4 to diminish, after a
long long time.
> Therefore, I think a transfer policy is also needed to allow new
> entrants to acquire IPv4 space and compete in that market.
In an all-IPv4 Internet with no clear IPv6 migration strategy, IPv4 is
the ultimate infrastructure bottleneck. It will be just like the last-
mile facilities access plant, only ubiquitous, omni-directional, and
impossible to bypass.
If you're suggesting that market transfers should be approved because
they will enable future new entrants to buy in and actually compete in
the wholesale-IPv4-mandatory services market against incumbents who
got theirs effectively for free, then I strongly suggest that you
think about how that notion actually has worked out in practice, e.g.,
in the facilities-based Internet access business itself.
> Tom Vest wrote:
>> This thread is a complete red herring.
>> The question is not whether or not transfers are good or nice, but
>> rather whether they'll work, and whether they'll be sustainable,
>> and at what cost to everyone, participants and non-participants
>> Many people thought that CIDR, aggregation, and route filtering --
>> the elements that made the RIR system "work" (i.e., rationalize
>> demand for address resources and routing system capacity) -- were
>> evil, because they reinforced customer-level "provider lock-in."
>> That fact probably made it a bit easier for the largest operators
>> at the time to sign onto the idea of RIRs in the first place.
>> However, arguably, this "immoral" feature was counterbalanced by
>> the fact that the system preserved openness at the routing service
>> provider (including self- provider) level. So incumbents got
>> something out of the deal, but so did everyone else -- or at least
>> everyone else who was willing to buy a couple of routers and hire
>> an engineer. That openness helped to assure that even those who
>> didn't want to opt into the routing game would benefit indirectly
>> from the continuous churn and competition that new entry *alone*
>> Set aside the question of whether transfer markets will work or
>> not. Ignore doubts about their technical sustainability. Even if
>> the rosiest scenarios on these two counts come true, transfer
>> markets will still undermine the interests of both "customers" and
>> all future aspiring new entrants -- both of whom can look forward
>> to a future with all of the downsides of PA and then some, and
>> none of the benefits.
>> For that reason alone, the community should be seriously
>> considering other approaches.
>> You are receiving this message because you are subscribed to
>> the ARIN Public Policy Mailing List (ARIN-PPML at arin.net).
>> Unsubscribe or manage your mailing list subscription at:
>> Please contact info at arin.net if you experience any issues.
More information about the ARIN-PPML