RE $50 Million NSF windfall??

[Dennis Ferguson]

> The Canadian situation has been hard to track. Canada appeared
> to be on the road to building Internet infrastructure but then
> turned back several months ago when it closed its IP Adddress
> Allocation facility. It is too bad that this set back occurred.

I supervised the group where this function was done in its initial
stages and I think the experience was actually a good lesson about
the problems of trying to distribute the task of IP address
allocation too thinly.  IP address allocation, like similar functions
(the North American Numbering Plan, and perhaps radio spectrum allocation,
come to mind) is better done centrally since by nature it is a
relatively low volume activity but requires a fairly high level of
technical effort to coordinate.  Canada, being only 1/10th the size
of the US and even a smaller fraction of the size of Europe and Asia,
was perhaps the best experience we've had in understanding how IP
address allocation benefits from the economies of scale of
centralization.

I am sorry you have been unable to track this, I know the benefit
of experience is hard to acquire from mailing lists and web pages.
Since you lack IP address allocation experience yourself it is perhaps
less clear to you why centralization of this particular function serves to
optimize the service.

I do notice that you've confused `building Internet infrastructure'
with IP address allocation, however, so maybe I can help you clear
this up.  Building Internet infrastructure is something that many
companies and organizations do, even in Canada, and is indeed an activity
of fundamental economic importance, often requiring large investments, which
must be encouraged, stimulated and provided the support necessary to
maximize successes.  Organizations which allocate IP addresses, on the
other hand, do not build Internet infrastructure at all (all the integers
we can use have already been built), their only function is to provide
a service (only one of many required) in support of those organizations
which do build the infrastructure.

This may be hard to understand, but the distinction is important.  You
need to look at the big picture. The activity which is important here
is building Internet infrastructure.  The cost of IP address allocation
is, or should be, just minor overhead in terms of the total cost of
constructing a really large Internet, both locally and world wide.  But
while IP address allocation is not a large task in the scheme of things
it is extremely important that the task be performed very well, since the
ultimate size to which we can grow the Internet, and reap the benefits
from having a very large Internet, are highly dependent on the skill with
which this small service is performed.

In other words, the economic benefit to the U.S. of having just one very
skilled IP address allocation registry located on a beach in Tobago will
far exceed that of having 100 address registries in the 100 largest U.S.
cities, none of which are very good since they can't get enough practice.
I think that many government people who understand the bigger picture, as
well as infrastructure builders who need this service performed, know this
very well, which is why many see central allocation under the control of
the infrastructure builders who depend on the function as being optimally
good for the industry.

> Please do not point to some RFC. Some U.S. Government
> people would not know an RFC from the IRS if it walked
> up and said hello.

This is true.  But this is why many U.S. Government people turn to those
who do have experience, and who do understand, to help them find the best
path to follow.  I think this has happened in this case.

Dennis Ferguson

P.S. I never did see your answer to this question:

>Message-Id: <v02140b06af3eaff4490f@[192.52.71.233]>
>Date: Sat, 1 Mar 1997 23:27:08 -0500
>To: Jim Fleming <JimFleming at unety.net>
>From: John Curran <jcurran at BBNPLANET.COM>
>Subject: RE: ARIN Comments
>Cc: "'naipr at arin.net'" <naipr at arin.net>
>
>At 19:32 3/1/97, Jim Fleming wrote:
>
>>The main point of the allocation is to spread the
>>economic benefits of registry operations around.
>>If the IPv4 IP address leasing market is allowed to
>>develop and rates are deregulated, then a multi-billion
>>dollar per year leasing industry can blossom.
>
>Now this makes sense...   your goal is to create a 
>multi-billion dollar registry industry, and mine is
>to ensure operational stability and transition of a 
>key function from a for-profit to an industry-led
>non-profit association that recognizes stewardship
>responsibilities for address space and consequential
>routing resources.
>
>Are you certain that there is not a sufficiently large
>marketplace being formed for DNS registries already such 
>that your goal can be attained without also restructuring
>for multiple commercial IP registries for the americas?
>
>/John