[arin-ppml] Q2: on Address Transfers - Overkill on the freezeperiod?

[Robert E. Seastrom]

"Milton L Mueller" <mueller at syr.edu> writes:

>> -----Original Message-----
>> From: Robert E. Seastrom [mailto:rs at seastrom.com] On Behalf Of Robert
> E.
>> Complete disagreement here.  Restrictions on the transferor and
>> transferee are different sides of the same coin, but restrictions on
>> the transferor are closer to the source and more likely to catch
>> people's attention.
>
> Closer to the source of what? I quite literally have no idea what you
> mean.

Closer to the source of the goods that are being shlepped around, in
this case blocks of IP addresses.  They come from IANA to ARIN to an
initial allocation, and thence under the post depletion scenario, to
be transfered to another entity under ARIN approval.

> From what I've seen of your argument, you still aren't refuting my
> contention that you can catch speculation more effectively by
> restricting post-transfer sales than by slapping categorical
> restrictions on people who can transfer. 

I haven't seen any substantiation of your claim that downstream
regulation alone is more effective than a combination of downstream
and upstream regulation as called for in 2008-2.  Pardon me for
rejecting it out of hand under the circumstances.

Note that I am advocating *both* conditions on the transferor *and*
the transferee.  Please present your less-is-more argument.

>> > As Scott noted, the only thing you want to do with the source is
> make
>> > sure that they don't sell their assignment and then come back to
> ARIN
>> > for more for free.
>> 
>> Or go back into the market, driving up the prices needlessly.
>
> This comment makes no sense to me, and seems to be unrelated to the
> point about someone selling addresses and then restocking from the free
> pool. 
>
> What do you mean by "go back into the market, driving up prices?" If
> more supply goes into the market, prices tend to go down. If onselling
> by transfer recipients is restricted for two years, they don't go back
> into the market, for two years. 

We aren't proposing that the transfer market be open until we've run
out anyway, so don't jump to the conclusion that there will be any
free pool to replentish from.

Sorry for the imprecise language by the way - "go back into the
market" should have been "go back to the market" or "go back into the
marketplace"; it is referring to the behavior of the customer, not the
destiny of the netblock.  Should have been clear from the context that
I wasn't referring to netblock churn, rather the entities involved,
but thanks for the opportunity to clarify that.

>> I have a huge amount of faith in ARIN's analysts and believe that they
>> do an exceptionally good job today of making sure that requests that
>> are blatantly fabricated don't slip through.  I do not make the error
>> (as you seem to have) of assuming that everything is black and white,
>
> Huh? It is the advocates of centralized needs assessment that believe
> everything is black and white. I believe that nothing in economics is
> black and white; everything is a trade off and a question of relative
> cost compared to close substitutes. You on the other hand seem to think
> that there is some fixed, divinely ordained relationship between an
> organization's network and the amount of addresses it needs. 

I would say there is absolutely a relationship between an
organization's network design and purpose, and the amount of address
space it needs, just as there is a relationship in rough terms between
the span/purpose/traffic on a bridge and the amount of steel it needs.
Determining what resources are needed to meet desired objectives and
criteria (as well as the cost tradeoffs involved) is at the core of
any engineering discipline.

There are no "close substitutes" for IPv4 space.  IPv6 doesn't even
come close.  When (if) we reach a tipping point where there is a whole
lot of widespread IPv6 deployment and v6 "just works" to the extent
that v4 does today (including rollout of transitional technology),
*then* that argument could be made.

> Black and white folks could say that _no one_ "needs" IPv4 addresses any
> more, because they could implement NATs or switch to IPv6. An economic
> perspective says, it all depends on what you have to pay to implement
> those various options. And since my consumption of more v4 addresses
> stops you from consuming them, there ought to be a price system to use
> to value those options. 

There is nothing counter to this in 2008-2; indeed, the mere existence
of 2008-2 would seem to suggest that the authors and contributors,
myself among them, "get it" on these points.

>> Creating a climate in which the incentive to fib on one's applications
>> is increased rather than decreased, is a step away from goodness.  I'm
>> not for it.
>
> That climate is HERE, my friend. It is caused by the increasing scarcity
> of v4 addresses. It will continue regardless of the fate of transfer
> policies. That was my point. 

There has always been an incentive to fib to ARIN (tempered by the
refresh rules - you don't want to get so much space that it lasts you
long enough that you're back into small allocation land) and it hasn't
changed that much over the past 10 years in absolute terms.  When
exhaustion is looming larger on the horizon relative to the average
lifetime of the management team at an ARIN member, *then* you're going
to see a big uptick.

That's not what I was addressing though, when I said "creating a
climate".  The scalar that represents the
current-level-of-BS-incentive is irrelevant moving forward - what
matters is the vector that represents ramp-up of BS-incentive that a
policy creates or reinforces.

> However, you and Scott did give me an idea. If you do want to impose
> restrictions on transferors, then simply require that they cannot
> transfer _any addresses they received from ARIN_ in the past year/2
> years. 

Which means you can get a bunch of addresses for the ostensible
purpose of expansion or perhaps aggregation, move into those, and put
the old ones on the market?  Sorry, no deal.  If I'm not mistaken, we
hashed that one over in Reston.

                                        ---Rob