[ih] Jon Postel's papers

[John Gilmore]

John Kristoff via Internet-history <internet-history at elists.isoc.org> wrote:
>                                                   I'd be curious what
> timelines, milestones, events, etc. you'd be particularly interested in
> for the most important Internet history insights.

I'd be interested in Jon's records around the replication of the root zone
files, and the transition of IANA functions to a non-governmental party.

Network Solutions (NSI) had been running the DNS zones for years under a
sole-source government contract, doing the minimal work required to
register domain names.  It got the job because it had a minority owner,
and that provided priority in government contracting.  NSI was bought in
1995 by beltway bandit SAIC for $4.7M.  SAIC immediately politicked NSF,
which then allowed NSI to charge every domain holder $50/year for their
formerly free domain names.  This monopoly and SAIC's effort to profit
from it on the backs of every Internet user did not sit well with many.
(SAIC later took NSI public for $54M of investor dollars, without
diluting its control over NSI, and immediately handed a large chunk of
those dollars back to itself scot-free.  Then it sold the whole thing
for billions, during the tech bubble.  SAIC is a company without outside
shareholders -- it is "employee-owned" -- so there are no outside
parties nor investors with authority nor influence over what corrupt
things the employees decide to do.)

President Clinton asked a friend who had no connection with the Internet
community, Ira Magaziner, to investigate the situation and make a
recommendation.  Ira went around and interviewed lots of people
involved, but was mum about what he might eventually recommend.
Magaziner had no actual authority, but he had the ear of the President,
so many people deferred to him.  (It wasn't clear whether the US
President had any authority over the Internet either, but this was still
at the stage when key parts of the infrastructure were being funded by
the US government -- including IANA.)

I was part of the CORE (Council of Registrars) effort to start up some
legitimate new competing TLD's.  This effort was catalyzed by the
nonprofit Internet Society, and Jon Postel and I were both board members
there.  Jon was also collaborating in his IANA role.  Jon had the
authority as the IANA, to add new TLD names to the root zone.  But he
had no backing against attacks by a billion-dollar beltway bandit with a
monopoly; he was just an academic with a small government contract.  So
ISOC and CORE agreed to fund legal assistance and indemnification for
IANA in return for IANA adding the new domain names that CORE needed.
There were some serious questions about whether NSI/SAIC would quietly
allow their monopoly to expire -- even though they would retain the
lucrative .COM.  We thought it more likely that they would file a bogus
lawsuit to drag out and muddle the process in the hope of permanently
disrupting it.  At the time, I was also on the EFF.org board and knew
lots of good lawyers.

CORE incorporated as a nonprofit trade association, signed up almost a
hundred registrars, and raised tens of thousands of dollars in initial
joining fees from each of them.  It used that money to subcontract with
Emergent Corp. to build the central registry hardware and software that
would operate the seven new TLDs if and when they were established.  It
defined protocols and wrote client software for registrars to interact
with the registry, and got it all working in a San Francisco data
center, manned 24 hours a day by trained operators.  CORE had a dozen
registrars successfully doing test transactions with the central
registry.  But we couldn't go into real operation without those new TLDs
getting into the domain name systems' root zone.

The root zone had been traditionally provided by IANA to NSI's "A root
server" periodically (by FTP?).  Each of the dozen-or-so other root
servers would then replicate it from the A root server using the
standard DNS zone transfer protocol.  These root servers were operated
by well connected volunteers all over the globe.  Jon was (reasonably)
concerned that if he added seven competing TLDs to the root zone, then a
corrupt NSI would refuse to accept the update at the "A" root server,
and the TLDs would remain unusable, despite his authority to define the
contents of the root zone, and despite NSI having no authority to define
its contents.

So Jon started asking root server operators to change their DNS
configuration so that they would replicate the root zone directly from
IANA's root server, rather than from NSI's root server.  This would
have, and did have, no effect on Internet domain queries, since IANA's
server was always serving up the same data as NSI's server.  Jon started
by asking the most likely candidates, and had successfully converted
more than half of the root servers to direct replication from IANA.
When he asked the next root server (I think it was the one run by the
Army), they told NSI about the request.  NSI escalated the issue to SAIC
and to Ira Magaziner.  On 1998-01-30 or so, there was a fractious phone
call from Ira Magaziner to Jon Postel and some USC-ISI lawyers.
Magaziner basically told Jon "Put those back or you'll never work on the
Internet again".  Despite the unlikely idea that newbie policy wonk
Magaziner could have anything to do with whether Internet co-inventor
Jon Postel could ever work on the Internet in the future, Jon
unfortunately agreed to do so, rather than asserting his authority as
the IANA to run the root zone as he determined best.  Someone leaked
this incident to the press, with a spin that Jon was "destabilizing the
Internet" rather than that Jon was cutting out the inadvertent control
of a company with an interest in monopolizing the Internet for its own
profit.

Two weeks later, on 1998-02-15, CORE's data center operator had departed
at 15:45 before their replacement operator had arrived (the new operator
was ill and only arrived at 19:00).  Meanwhile, the data center was
broken into by thieves, the chain-link fencing around the servers was
cut, and two entire Sun Enterprise 450 servers, worth about $70,000,
were stolen.  Nothing else in the whole multi-tenant data center was
stolen.  This was obviously a targeted theft, and who could have wanted
to target CORE except SAIC?  The theft was investigated by the police,
but was never resolved.  CORE's contractor had good offsite backups and
the equipment was insured.  They installed a second pair of Sun servers
overnight, and were back to full operation within 29 hours.  Even if
they had been running operational TLDs, the TLDs would have continued
functioning just fine.  But for one day, the people who owned those TLDs
would have been unable to make changes in them.

Ultimately, Magaziner's "Green Paper" and "White Paper" proposals backed
NSI's monopoly, which continues to this day over .com, by far the most
popular and lucrative top-level domain.  The CORE registrars became
resellers of NSI's service, and CORE dissolved as a back-end registry.
Jon Postel died of a leaking heart valve later that year, which left a
void that the corrupt, bloated and self-serving ICANN (which was created
based on Magaziner's model) was happy to fill.  A few of the smaller
TLDs were hived off to other orgs (one of which went to ISOC, where top
employees later tried to buy it for a borrowed billion dollars, making
the money back by vastly increasing the price of renewals for every
nonprofit on the Internet).  After many years, many new competing TLDs
were created, none of which has been particularly successful.  In short,
the fix was in, and the beltway bandits won.  NSI is still charging
premium prices (in the $15 range) for each year of back-end .com domain
registration that costs them less than a penny a year to provide.

There's a bit more background on this in an interview with me by Salon
from 2002:

  https://web.archive.org/web/20120109194541/http://www.salon.com/2002/07/02/gilmore_2/

I repeat, it would be interesting to see Jon's papers and records about
that time.  Most of them would probably be emails, and there would
be thousands or tens of thousands of them.

	John