[ih] Jon Postel's papers

[Karl Auerbach]

(This is a re-send from my business email address.  The list manager 
software doesn't seem to like when I post from my personal email.)

Thanks for the detailed write-up.

I was also involved in much of this.  Much of that was head butting with 
a lawyer from Jones Day who drew up and proselytized (rather forcefully) 
much of the legal text of what was to become the ICANN/RIR system. I did 
not get the sense that Ira Magaziner was more than a figurehead and that 
this non-government lawyer was really the person calling the shots.  But 
that's a story for a different day.

Back to Jon Postel and his papers...

Jon sometimes worked via ephemeral media - such as phone calls.

My final phone chat with Jon involved the IP address registry system 
(the RIRs).  We quickly agreed that the goal of such a system was to 
promote good CIDR address block aggregation in ways that helped reduce 
the number of address block prefixes or ASNs that had to be advertised 
into (and carried by) the routing systems.  This mean that CIDR 
allocations ought to be done in conformity with the general shape of 
Internet connectivity.  (We did not delve into this long enough to face 
the question of whether there ought to be CIDR address block 
re-assignments if the routing clumpiness of the net were to 
significantly change.)

For instance, we recognized that much of Africa was effectively routed 
through New York (a fact made quite clear a few years later on Sept 11, 
2001.)  As such an address registry in Africa would probably be better 
cast as part of the formative ARIN that covered North America.  (Of 
course, that notion would have ruffled many feathers of those who 
consider having a regional address registry as a kind of status symbol 
without regard to whether that is an optimal way to manage Internet 
technical resources.)

Jon and I agreed that from a technical point of view IP address 
registries ought to be considered fluid rather than permanent, i.e. that 
they ought to be created, split, or coalesce based on how the large 
clumps of Internet connectivity formed and changed. That, obviously, is 
a road that was not taken during the formation of ICANN and the regional 
IP address registries.

I mention this in the context of the ongoing troubles with the African 
RIR, Afrinic.  (It would be, to some, a very annoying rock tossed into a 
calm pond were the ongoing legal dispute over that RIR had to recognize 
that RIRs can be, and perhaps ought to be, impermanent, transient 
entities.)

     --karl--