US CODE: Title 15, Chapter 1, Section 2.

[Stephen Satchell]

At 7:07 PM 1/31/97, Karl Auerbach wrote:
>If an ISP needs to get more address to satisify a customer request and
>if its higher ISP (if any) is unwilling or unable to satisfy that
>request, then that ISP has exactly one place to go: ARIN.
>
>And if a business wants its own block, then there is one place to go:
>ARIN.
>
>So I count one place: ARIN.
>
>There could have been others, but they have all agreed among
>themselves to carve the world up into exclusive zones in which each
>will have the sole and exclusive right to allocate address blocks.
>
>Perhaps we just ought to drop the geographic limitations and let the
>three registries allocate anywhere in the world.

If there weren't significant technical problems with your suggestion, I
might actually go for it -- having multiple registries so fits the "flavor"
of the Internet.

Unfortunately, I still have many of the articles on routing from my 1972
ARPAnet days, and from what I've been able to glean from more recent
publications there is damn little science and all too much art to designing
routing schemes that work.  When you consider that SIGCOMM Computer
Communication Review has article after article after article about the
current "hot flash" in routing that doesn't pan out, I think that your cry
of "conspiricy" is very hollow indeed.

Aggregation is a fact of life.  Just as many people work very hard to keep
Malthus from having the last laugh, there are a large number of very brainy
people that are trying to stave off Metcalf's Prediction -- that the
Internet will collapse by its sheer weight in the next year or so.  Until
we have routers which can handle 16,777,214 endpoint addresses in parallel
without introducing unacceptable processing delays -- or unacceptable
equipment prices -- we are stuck with trying to use what we've got.

IPng isn't going to help this.  If anything, a 128-bit address space is
going to make it *worse*, not better.

Blocks do need to be allocated by geography, either physical or at least
topologically.  A central registry makes sense for that.  Now if you want
to talk about how to reduce the cost of running such a registry, and
therefore reduce the fees required to make such a registry work, that's
great.  But to talk about a free-for-all is just inviting a chaotic
collapse of the 'Net.

Is that what you want>?

---
Stephen Satchell, Satchell Evaluations
<http://www.accutek.com/~satchell> for contact and other info
Opinions stated here are my PERSONAL opinions.