Re: [IANAxfer@apnic] Key elements of the transition of IANA stewardship
Dear David,
Thanks for this and please see embedded comments below.
Best,
Richard
> -----Original Message-----
> From: David Conrad [mailto:drc@virtualized.org]
> Sent: jeudi, 11. septembre 2014 02:28
> To: rhill@hill-a.ch
> Cc: ianaxfer@apnic.net; John Curran
> Subject: Re: [IANAxfer@apnic] Key elements of the transition of IANA
> stewardship
>
>
> Richard,
>
> On Sep 10, 2014, at 9:22 AM, Richard Hill <rhill@hill-a.ch> wrote:
> >> I figure the
> >> role of ICANN’s Board is primarily to provide a mechanism by
> >> which interests outside of the RIR world can provide input into
> >> Internet numbering policies.
> >
> > Meaning that you would accept that the ICANN Board could override RIRs
> > policies? For example, on the basis of advice from GAC?
>
> No. The board can refuse to ratify a global policy. That does
> _not_ override the RIR policies. The ICANN Board can suggest
> there are other considerations when reviewing a policy, but if
> you look at the ASO-MoU, in the end, the final step is for folks
> to go to mediation.
The MoU actually contains an arbitration clause, not a mediation clause. It
specifies ICC arbitration in Bermuda. But it does not contain a choice of
law clause, so it is not clear what laws would apply.
If such a situation arose, I suppose that ICANN would argue that California
law applies and that, under California law, the Board has the ultimate
fiduciary duty, so the Board's decision must prevail.
So wouldn't it be better to clarify the matter, to make it clear that the
ultimate authority for IP addressing is NRO?
>This does not impact the RIR policy (which
> has been approved by all RIRs’ memberships).
>
> >> If ICANN did not abide
> >> by the outcome of RIR policies/processes, the RIRs had already
> >> created an entity (the NRO) that was specifically designed to
> >> provide the exact same top-level coordinative service that IANA
> >> staff performs (that is, allocating top-level blocks of numbers
> >> to the RIRs) should ICANN be unable/unwilling to do so.
> >
> > Yes, but, absent the contract with NTIA, ICANN has no
> obligation to defer to the NRO.
>
> Of course they do. ICANN has entered into an MoU with the RIRs.
Yes, but that MoU only specifies that NRO is the ASO. ASO is an organic
unit of ICANN and therefore ultimately subject to the authority of the ICANN
Board.
> Even if they hadn’t, there is RFC 2860.
I think that it is generally agreed the RFC 2860 applies only to protocol
parameters.
> Even if you ignore that,
> the entire mission of ICANN is "is to coordinate, at the overall
> level, the global Internet's systems of unique identifiers, and
> in particular to ensure the stable and secure operation of the
> Internet's unique identifier systems.” If the NTIA contract was
> not there, ICANN’s Board still has a fiduciary responsibility to
> abide by the corporation’s Bylaws.
Yes, that is correct. ICANN has the fiduciary duty to be in charge, meaning
that the Board has the fiduciary duty to be the ultimate decision-making
authority.
As I already said, some people are OK with that, but some are not.
>
> The IANA functions contract in relation to the numbering function
> is entirely superfluous.
>
> > My point is that, if people think that the NRO should have ultimate
> > authority, then the procedural and legal framework should specify that
> > explicity.
>
> And again, you’re viewing the world from a top-down prescriptive
> perspective. The NRO does not have ultimate authority. The
> Internet numbering world simply does not work that way.
You lost me. Somebody has ultimate authority for formally approving
policies. The question here is whether it should be the NRO or the ICANN
Board.
>
> >> The ICANN Board could also declare Pi = 4.
> > The value of Pi is determined by mathematics, so it is what it
> is no matter
> > what anybody says it is.
>
> And the value of an IP address is determined NOT by the board of
> ICANN or the RIRs, but it’s announcement by service providers
> into the routing system.
Sure. But I was referring to the allocation policies, not just the
allocated values.
>Again, the role ICANN and the RIRs play
> is to ensure that addresses are not duplicated. If ICANN were to
> declare 18/8 was now owned by Richard Hill, no one would listen.
>
> > Contrast this to the situation with respect to domain names,
>
> I thought we were talking about Internet numbers.
Yes, but you referred to "a world of top-down national law or treaty-based
legal fiats to define policies", so I thought that you were referring to
telephone numbers. I apologize for misunderstanding, but then I'm not sure
what you were referrring to.
The telephony equivalent of IP addresses are Q.708 SANC/ISPC and E.212 IMSI.
The management of those resources is pretty similar to that of IP addresses,
except that the top-level assignments are made by ITU (instead of ICANN) and
the lower-level assignments are made by national regulators (instead of
RIRs). I don't know of any country that has regulations imposing the use of
assigned ISPCs or IMSIs: they are used on a voluntary basis by operators
because it makes commercial sense.
>
> >> A small amount of power resides in ICANN and the RIRs
> >> by their guarantees of uniqueness and by providing access to
> >> registration data.
> >
> > Again, this is similar to the current situation for telephony, if you
> > replace ICANN/RIRs by ITU/national regulators.
>
> No. National regulators have the force of law.
Yes, if they exercise it. As I already stated, many don't really excercise
it these days, preferring to leave things to commercial forces.
>This power _does
> not exist_ in the Internet numbering world. Instead, there is
> the voluntary cooperative agreement among the ISPs of the world
> that the set of 32-bit and 128-bit values that are used on the
> public Internet are _coordinated_ by the RIRs and ICANN. Neither
> ICANN nor the RIRs can stop anyone from picking a random 32-bit
> or 128-bit number range and announcing that onto the Internet (as
> we have seen on multiple occasions). If ICANN or the RIRs abuse
> that coordinative role, then they will no longer be providing
> that role. This is radically different than the role played by
> national regulators.
In my experience, it is rather similar, in particular for ISPCs and IMSIs,
but I'm happy to agree to disagree.
>
> >> If ICANN (or the RIRs) screw up sufficiently, they’ll simply
> be replaced.
> > Yes, but the replacement is hardly "simple" and would create
> significant disruption.
>
> Not really. What is it exactly you think ICANN does with respect
> to Internet numbering?
ICANN formally approves the RIR policies. IANA allocates the IP address
blocks to the RIRs.
Of course one can set up another entity that does that, but that might take
a little while. And it might take quite a lot of communication before
everybody understands that a new entity is performing those functions.
>
> > I'm not suggesting that
> > IANA would run amok and publish nonsense. I am concerned that the ICANN
> > Board could, absent the contract with NTIA, override NRO/RIR
> > decisions/policies, if the Board felt that it was its fiduciary
> duty to do
> > so.
>
> Again, the ICANN board cannot override RIR policies.
>From a formal, legal point of view, I think they can, absent the NTIA IANA
functions contract. But I'm happy to agree to disagree.
>However,
> even if they could, it would have NO effect. The IANA address
> registries are not operational in nature. They are merely a
> publication of mutually agreed conventions of who has what
> address space. If ICANN were to disregard those conventions, it
> would simply result in the IANA registries being disregarded
> (hint: this is not speculative — if you think otherwise, then you
> must think DEC still owns 16/8).
I'm well aware of the fact that the IANA list of what used to be called
Class A blocks is not up to date. I don't understand why it has not been
corrected, in particular to show that the old DEC block is now owned by HP,
but that is a different matter.
But if ICANN/IANA add no value to the IP address processes, then why are we
having this discussion in the first place? Why doesn't the NRO publish the
correct list of top-level allocations? What is IANA's value-added?
>
> >> Do you believe ISPs will spontaneously decide to terminate
> >> routing service for the IPv4 address space they announce?
> >
> > Of course not. But consider a different scenario. A US Court
> orders ICANN to transfer a certain set of IP addresses from a
>
> > certain country to certain private parties.
>
> Sure. Courts (in the US and elsewhere) can order Pi = 4, but that
> does not make it so.
Again, the value of Pi is a mathematical construct. Who owns what IP
address block is a human (legal) construct. Courts can and do make things
happen to humans, such as seizing their property and giving it to someone
else.
>
> > ICANN puts pressure on the concerned RIRs,
> > asking them to revoke the addresses in question.
SNIP
>
> > The non-US based RIRs refuse to comply.
>
> Unless forced by a court, I feel safe in saying the US-based RIR
> would also refuse to comply unless the RIR’s membership decided
> it was appropriate to do so.
Sure. And this would create an inconsistency, which might or might not
result in practical problems.
>
> > ICANN then considers what steps to
> > take. This leads to a great deal of confusion and uncertainty.
>
> No. If ICANN modified their registry to meet the court demands,
> it would result in ICANN’s registry of address allocations simply
> being ignored. It would have no other impact.
Yes, that's exactly what I am saying: it would create confusion and
uncertainty regarding what registry is authoritative.
>
> > Or consider the scenario where GAC, for whatever reason,
> decides to advise
> > ICANN not to comply with some new NRO/RIR policy. And suppose that the
> > ICANN Board agrees to abide by the GAC advice.
>
> OK.
>
> > Depending on the case, the NRO/RIR might well decide to declare
> that ICANN
> > could no longer carry out the IANA functions role to its
> satisfaction, and
> > set up a new IANA function.
>
> Exactly.
>
> > But think of the mess and the disruption that would be caused.
>
> I’m actually curious. What mess do you think would be caused?
See above: time and effort required to set up the new entity and to inform
everybody that that new entity is now authoritative.
>
> > So I think that, as outlined in my original post to this list,
> it would be a
> > good idea for the NRO to follow the precedent established by
> the IETF and to
> > negotiate a binding contract with ICANN for the IANA functions
> related to IP
> > addresses.
>
> Isn’t that the ASO-MoU?
No. The ASO-MoU merely says that the NRO is the ASO, an organic ICANN body.
RFC 2860 says that the IETF is in charge of protocol parameters. That is
something different.
>
> Regards,
> -drc
>
>