[ppml] Markets, pricing, transparency, 2008-2 / 8.3.9
On Mar 18, 2008, at 11:11 AM, Cliff Bedore wrote:
> Tom Vest wrote:
>> On Mar 17, 2008, at 10:20 PM, Cliff Bedore wrote:
>>> Regarding a "market" for IPv4 addresses, I still have not heard any
>>> disadvantage to ARIN staying completely out of the market part and
>>> just have a
>>> policy that allows one more method of acquiring addresses.
>>> As I understand current policy, there are basically two ways to get
>>> 1) Ask ARIN with appropriate justification
>>> 2) Buy/merge with a company and get the addresses as a part of that
>>> transaction (again with some justification to ARIN
>>> It seems to me that all that 2008-2 needs to do is add a third
>>> once the
>>> IANA free pool is gone.
>>> 3) Party 1 offers to transfer to Party 2 some block of addresses.
>>> Party 1
>>> has the right to offer and Party 2 justifies the right to receive
>>> addresses. No market, no ARIN involvement with SEC, just ARIN
>>> approving use
>>> of a block of addresses.
>>> Why is this not a simple way to handle the end of life address
>>> management of
>>> v4 addresses? ARIN gets their fees just like now so they remain
>>> viable but
>>> have no interaction with the "market" and its associated problems.
>> Basically this is the simple way, but it will not work -- technically
>> it cannot work. This is because the arrangement you describe requires
>> the same incentives to operate in multiple, simultaneous, but
>> contradictory ways -- more or less like this:
>> 1. First, eliminate the single source mechanism for address
>> delegation. Henceforth anyone may potentially be a buyer or seller of
>> address resources, as they see fit.
> We're not eliminating the single source. Like the title for your car
> from the DMV, it goes through one and only one place.
You got the title to your car from the DMV?
Gosh, I thought that chain started with something (i.e., a "pink
slip") that came from the car seller and/or financer.
You have to *report* title transfers to the DMV after the fact, after
which they replace the old one with an updated one.
But you have to report it first.
But we all know how much everyone loves to take time off from work/
fun and go visit the DMV, right?
I bet we'd all go visit them religiously even if all licensing and
vehicle registration were voluntary, and police did not exist.
It kind of makes one wonder why all of that taxpayer money is wasted
on police in the first place ;-\
You live in California right?
>> 2. Next, allow market forces to govern the address delegation
>> -- i.e., engage everyone's well-honed instincts to spend less and
>> profit more, and allow the overall distribution of address
>> resource be
>> determined and legitimated by that market process.
The point is not that markets or the profit motive are bad.
The point is that once you subject some system to those motives, you
have to assume that they will be applied consistently and vigorously,
not just enough but not too much, not just in ways that you think are
>> 3. Having empowered everyone this way, and fired up those competitive
>> juices, also make sure that under certain defined circumstances,
>> everyone always ignores both their new sovereign prerogatives and
>> their newly engaged, otherwise-unrestricted market instincts, and
>> voluntarily accept extra costs, restrictions on when and under what
>> circumstances they may act, etc.
> Doesn't happen in what I describe.
No, I agree it doesn't.
>> 4. Make sure that they follow these rules and pay these extra
>> costs in
>> 100% of the required circumstances. Also assure that everyone takes
>> the extra effort to inform some now distant, formerly critical
>> institution of the results of their actions -- for whatever reason.
> Doesn't happen in what I describe
No, I agree it doesn't.
>> 5. This is a decentralized, market-driven system, but unlike every
>> other market in human history all rules are purely voluntary, and no
>> other enforcement mechanisms will ever exist. Make it work anyway.
> Not decentralized. Not a market. Simply one more way in which
> addresses may be obtained from/through ARIN
"Decentralized" means the individual transactions that affect the
quality of the "central registry" and the overall (next-level)
distribution of address resources happen outside of the direct
purview of the maintainer of the central registry. Don't blame me
man, I didn't make the word up.
>> If that illustration doesn't work, maybe this one will:
>> Basically the arrangement you describe would be like vehicle
>> in the US. We all know that, if we can afford it, auto insurance
>> to protect us from somewhat from extreme risks. We also probably have
>> a vague sense that the insurance we pay for also helps to protect
>> other drivers too -- although that's hardly a motivating factor
>> for us
>> to buy in. But some people simply can't afford it, and some are
>> forgetful and let their coverage lapse, and some people just like to
>> live dangerously, damn the consequences for themselves and
>> everyone else.
> Not like insurance. Much more like the DMV.
Yes, like the magical, police-free, self-enforcing DMV.
>> The Insurance Research Council estimates that 15-16% of all vehicles
>> on US roads at any given time (c. 2007) were uninsured. Lucky for US
>> drivers, there is no Department of Homeland Insurance that might be
>> called in to rectify the situation in case uninsured, anonymous
>> motorists start crashing into important things, or if that compliance
>> rate drops too low. We do not enjoy the same luxury.
> People hijack addresses now for various reasons. I even had one of
> hijacked years ago. But it got fixed within the existing system.
Well, if we can import some of that auto-enforcing DMV magic, I'm
sure that successor central registry maintainers will have no trouble
correcting similar problems in the future, because that magic will
compel everyone to voluntarily keep their entries complete, accurate,
and up to date, or perhaps magic will simply cause missing and
erroneous entries to spontaneously correct themselves.
>> I can't make the case any more clearly. I would also like the simple
>> way to work, but wishing it will not make it so.
> Then I guess I give up because I still don't understand your
> examples or
> why what I put forth won't work.
That is definitely the right response; I am going to follow your lead