This action might not be possible to undo. Are you sure you want to continue?
In September 1994, twenty five years after installing the machine that spawned the
Internet, the veterans of the ARPANET project – the Internet’s precursor – regrouped for an
anniversary celebration in Boston. Using a huge mural of the world as a backdrop, a
photographer set up to take a shot of the most distinguished engineers present. There was
evidently some jostling for position among the nineteen who made it in, but no question
about who would sit front and center in this prestigious class of middle-aged white males.
That spot went to the ubiquitous Vint Cerf, one the Internet’s most inventive and
celebrated personalities. He had directed the ARPANET project from 1976 through 1982 and
was a truly seminal figure in the long series of technical and political developments that led
to the modern Internet. His productive career brought him countless honors plus the
benefits of financial success. He would becoming a recurring figure in the DNS War, and
even more prominent in its aftermath.
Cerf was flanked on his right in this picture by Bob Taylor, former director of the
Information Processing Techniques Offices at the Advanced Research Projects Agency
(ARPA), under the U.S. Department of Defense. It was Taylor who had first envisioned the
ARPANET as such, and put the project in motion in 1966. No slacker either, Taylor went on
to create Xerox Corporation’s Palo Alto Research Center (Xerox PARC), the laboratory
where computing innovations from Ethernet to the graphical user interface underlying the
Apple MacIntosh and Microsoft Windows were nurtured and inspired.
On Cerf’s left was Frank Heart, head of the team that submitted and won the bid for
the ARPANET proposal. Heart’s contribution was undeniably important, but his chances to get
a seat in the front row weren’t hurt by the fact that his employer – Bolt Beranek and Newman
(BBN) – was paying for the party and the photographer.
Steve Crocker, “Initiating the ARPANET,” Matrix News , 10.3, March 2000, http://www.mids.org/-
Figure 6 Depiction of early ARPA Network. Hosts are shown as rectangles,
Interface Message Processors (IMPs) are shown as circles. Graphic is taken
from a presentation by Bob Braden.
ARPA’s initial award to BBN in 1968 was just over one million dollars. It was
arguably one of the smartest and most effective investments ever made by the U.S.
Government. Taylor had conceived the ARPANET as a project that would enable a few large
computers at universities and research institutions to interoperate with each other. First
two, then four, then more. Data signals were converted into an analog format so that they
could be carried across the public telephone network. The ARPANET’s proof of viability was
an encouragement to the development of numerous public and private networks over the
Many other networks sprouted and flourished in the ensuing years, including
MILNET, CSNET, NSFNET, Bitnet, FIDONET, and eventually Compuserve, Prodigy, and
America Online. But the suite of protocols that emerged from the ARPANET community made
it possible to link those various networks together, joining thousands and eventually millions
of nodes into a network of interconnected networks – an internet. Older systems either
merged in, or expired from lack of use. Even the ARPANET was shut down in 1990. High
speed digital lines interconnected the various networks that survived and prospered. The
network of networks came to be treated as a distinct entity – The Internet.
Presently, most people are familiar with the fact that Internet traffic is carried by
private firms called Internet Service Providers – ISPs. When the ARPANET cornerstone was
put in place, no reference was ever made to anything like an ASP – an ARPANET Service
Provider. But for all practical purposes BBN filled that role. The equipment which
constituted the network’s first node had been delivered and installed at UCLA by a BBN
team in September 1969. The company grew as innovation after innovation fostered
explosive use of networking technology and spawned the global Internet.
BBN’s generosity as host for the anniversary party was not driven entirely by
sentiment for days gone by. By the early 1990s the commercial environment for service
providers had become extremely competitive. Facing pressure in the marketplace, the
company was looking for new ways to advertise itself. It was in the process of launching its
own commercial ISP, called BBN Planet. (That venture that was later absorbed by GTE,
which was, in turn, swallowed by Verizon.)
From an advertiser’s perspective, a celebration that brought together some of the most
famous names in modern computing under BBN’s roof looked like a good idea. Reporters
were invited, of course. The company had also commissioned Katie Hafner and Matthew
Lyon, a husband and wife team, to write a book about ARPANET history that would showcase
BBN’s important contributions. This became the critically-acclaimed best seller, Where
Wizards Stay Up Late: The Origins of the Internet (1996). One of its plates included the
group portrait of the ARPANET’s pioneers, featuring Cerf, Taylor, and Heart on the front line,
reproduced here on page 56.
See his July 1961 Ph.D thesis proposal, “Information Flow in Large Communication Nets” at
http://www.lk.cs.ucla.edu/LK/Bib/REPORT/PhD/proposal.html, and the 1964 book, Communication Nets;
Stochastic Message Flow and Delay.
Email by David P. Reed, “The Internet and nuclear attack,” reprinted by Dave Farber, “IP: Must read
(especially the press) The Internet and nuclear attack,” IP December 28, 2001.
* * *
The second row of the group portrait included Larry Roberts, whom Taylor had hired
in late 1966 to execute the project on behalf of the Department of Defense. Roberts became
the government’s man in the trenches. Highly regarded for his prodigious talent and
discipline, Roberts was the true leader of the project. He had drawn up the ARPANET’s
specifications, supervising activity around the country as work came to fruition. Roberts left
government work in the mid 1970s, going on to create the first private data
telecommunications carrier, Telenet (also absorbed by GTE), and later serving as President
and CEO of DHL. In the picture, Roberts sat next to his longtime friend, Len Kleinrock. The
two had worked together in the early 1960s on advanced aerospace defense projects at MIT’s
Lincoln Laboratory. Kleinrock wrote a seminal paper on packet switching in 1961 and
published the first book on the subject in 1964. After completing his Ph.D. at MIT 1963,
he joined the faculty at UCLA, doing research in mathematical queuing theory; a subject of
great practical utility to anyone who wanted to measure the performance of a computer
Roberts awarded the contract that put Kleinrock and the students working under him
– including Cerf – in charge of configuring the ARPANET’s first Interface Message Processor
(IMP). Supplied by BBN, the refrigerator-sized machine was hooked up to a hulking Sigma
7 computer at UCLA. It was also connected, via a phone line leased by BBN, to another
IMP/computer combination at the Stanford Research Institute in Palo Alto. As these
IMP/computer pairings mushroomed to include dozens of nodes, the ARPANET became an
increasingly useful network. (The idea that the system was designed to survive a nuclear
attack is a persistent urban legend, but the IMPs were indeed built into “blast hardened”
cabinets to help demonstrate the concept of a “survivable network.”)
Also in the third row of the picture was Robert Kahn, who had been a key player at
BBN in the first years of the ARPANET contract. Kahn left the company to work for the US
Government in 1972, and set out to develop a way of connecting computers as peers, rather
than as components in a hierarchy. This was a formidable task, given the diversity of
machines and operating systems to be interconnected. Kahn knew he needed help solving the
peering problem. Cerf had finished his Ph.D. by then and was teaching at Stanford
University. Kahn picked him as his collaborator, and Cerf never returned to academia.
Over the course of 1973 and 1974, Cerf and Kahn devoted themselves to developing
a technology they called the Transmission Control Protocol (TCP). TCP was a huge leap
forward that vastly improved the ability to exchange messages between different types of
computer systems. The protocol afforded a software substrate upon which all the various nets
of the world could ultimately interconnect. The development of TCP made the Internet
possible, and thus raised up its developers as legends.
* * *
There were several more rows of ARPANET veterans in the photo. Smiling broadly,
way back in the next to the last row, were two engineers who might have won a ranking
closer to the front of if they had elbowed for it. But their temperaments were famously
unobtrusive. One, Steve Crocker, had been Cerf’s best friend at Van Nuys High School in
Los Angeles. Next to him was Jon Postel, who had graduated from the same school a few
years later. In addition to wearing a wide happy grin, Postel sported the longest beard of
anyone in the picture.
Cerf, Crocker, and Postel were literally present at the creation of internetworking.
When the first messages were sent between computers at the Stanford Research Institute and
UCLA, they were all working as graduate students under Kleinrock. The first host-to-host
message was sent by another student, Charley Kline, at 10:30 PM on October 29. 1969.
There was no profound message in the transmission. In fact, the initial attempt caused a
George Johnson, “From Two Small Nodes, a Mighty Web Has Grown,” New York Times, October
12, 1999, D1.
For a photo of the entry, see “The Day the Infant Internet Uttered its First Words,”
crash. Characteristically, it was Postel who had set up and maintained the IMP log book in
which that portentous first sign-on was recorded for history.55
Cerf, Crocker and Postel had been together in other commemorative pictures over the
years. One turned up in a special issue of Time Magazine, celebrating the legacies of the
1960s. They didn’t make the cover. That was a scene from the Woodstock music festival.
But inside, there they were, sitting around a table, mischievously talking into tin cans
connected by sausage links.
The actual topology of the ARPANET was considerably more sophisticated than a few
string and cup telephones. BBN made the connections possible, installing four IMP nodes
by December 1969 (the third and fourth were at UC Santa Barbara and the University of Utah
in Salt Lake City), and hooking them up via dedicated phone lines. In those early days Cerf,
Crocker, Postel, and many other graduate students around the country were working to make
those links practical by designing mechanisms that would allow the IMPs to communicate
with their local “host” computers.
With Kleinrock’s support, Crocker had taken the lead in coordinating this dispersed
group of students and volunteers. His goals were fairly simple... to articulate the basic
standards that would allow different kinds of computers to talk with the IMPs, and to
describe new features of the system as they emerged. Those meetings, which began in 1969,
were consecrated as the Network Working Group (NWG). Years later, in 1979, vestiges of
Crocker’s NWG were reunited as the Internetworking Conference Control Board (ICCB).
The ICCB was reconstituted in 1983 as the Internet Activities Board (IAB). The IAB, in turn,
spawned the venerable Internet Engineering Task Force (IETF), which remains the primary
venue for development of the Internet protocol suite.
While still a graduate student, Crocker launched a documentation process for the
NWG that eventually became the formal basis for modern Internet standards. That process
Several different series of documents, including Internet Engineering Notes, were used by that same
community. “See Internet Archaeology: Documents from Early History,” http://www.rfc-editor.org/history.html.
endures to this day. He titled the series “Request For Comments,” a humble name that
understates its impact. Over 3500 RFCs have been published since Crocker put out the first
one in April 1969.
Postel, up until his death in 1998, authored (or co-authored) more RFCs than any
other individual. In 1978 he and Cerf (along with Danny Cohen) presented a major
improvement to the Transmission Control Protocol that Cerf and Kahn had introduced four
years earlier. The revision, known as the Transmission Control Protocol/Internet Protocol
(TCP/IP) became the elegant platform from which the Internet blossomed beyond all
expectation. Postel helped manage the conversion of the ARPANET and other attached
networks from the predecessor Network Control Protocol (NCP) to TCP/IP between 1981
and 1983. He also authored the specifications for the Simple Mail Transfer Protocol (SMTP)
in 1982. It is the basis for relaying email across the Internet’s diverse networks.
These are all huge achievements, but Postel is most famous for something else. When
he was a student at UCLA, one of his dissertation advisors, Dave Farber, volunteered him
to begin recording the various numbers, addresses, and technical parameters that were being
assigned and reserved as the ARAPANET phenomenon grew. Alex McKenzie at BBN had
been updating host table numbers and noting the changes via the RFC series, but there were
many other types of assignments that needed to be tracked. Until then, the records were
dispersed, and were at risk of becoming haphazard. One repository was a set of index cards
that Bob Kahn carried in his shirt pocket. Postel took on the task of consolidating the
information and ensuring it was kept in order, up to date, and publicly available. It turned out
that he was very good at it.
Postel became the reference point for people who needed information on how those
parameters were being used, or who wanted new parameters allocated for specific uses they
had in mind. He made sure that numbers which needed to be well-known were indeed well-
known. His real passion all along, he told me, was high speed, high performance computing.
For most of his career, the tedious but important drudge work of recording and publishing
Diane Krieger, "Heavenly Father of the Net: An Interview with Jon Postel," The Networker, 7.5,
Summer 1997, http:/www.usc.edu/isd/publications/networker/96-97/Summer_97/innerview-postel.html.
data assignments initially occupied only a moderate portion of his responsibilities.
Nevertheless, those responsibilities kept accumulating. He had helped Crocker edit RFCs
since the beginning, and took charge in late 1971 when Crocker left UCLA and went to work
at ARPA. Postel ended up publishing not only technical standards in the RFC series, but
records of best current practices, informational statements, and even some April Fool’s Day
pranks. He became an institution.
When the Hafner and Lyon book, Where the Wizards Stay Up Late, was published
in 1996, they described Postel as an “unsung hero.” By remaining in a government-funded
think-tank environment rather than moving to the private sector, he had passed up the chance
to become as wealthy as his colleagues. Many had done quite well, making their fortunes at
the leading edge of the Internet boom. Some had gone off to create their own companies,
while others climbed high on the corporate ladder. The young wolves of the IMP era were
now, for the most part, passing through a prosperous middle age together. They were also
becoming celebrities. Cerf and Kahn were receiving numerous public awards, including
perhaps more honorary Ph.D.s between them than any other two humans. They awkwardly
shared the informal but respectful title, “Co-fathers of the Internet.” When Roberts also
claimed to be the Internet’s father (perhaps rightfully), Cerf gracefully declared himself
“midwife.” Kleinrock finessed the paternity question by touting himself as “Inventor.”
Postel, however, was the subject of honorifics that were meant for him alone. He was
best known as the Name and Numbers Czar. He was frequently called the God of the
Internet, and sometimes even Supreme Being. As the Internet became an operational
system, Postel became the eminence grise among the community’s greybeards. He was no
unsung hero to people in the know. And that was about to become a much larger group.
* * *
In mid 1997 Postel sat alone for a full page picture that ran inside the October issue
of Internet World. Looking aged beyond his years, posing dourly behind a desk piled over
by paperwork, his beard had grown long enough to be worthy of an unrepentant 60s radical.
But now the beard was very very gray. The magazine’s cover was all about him. It showed
a picture of a gold bar, stamped www.$$$$.com. The headline trumpeted “...GOLD RUSH....
EXCLUSIVE INTERVIEW With the Man in the Center of the Domain-Name Battle, Jon
Postel.” (Shown on page 56.)
The wizard behind the curtain of the Internet ultimately become the most famous
wizard of them all – not just a star, but the man of the hour. Postel was finally in the spotlight
at center stage, fully eclipsing the old hands of the ARPANET, even Cerf. Now the world had
found out who he was, raining down attention and demands ever more relentlessly with each
turn of the Domain Name crisis. The smile was gone.
To understand how Postel earned his stature in the Internet community it helps to
understand his diverse contributions to the creation of “internetting.” Most significantly was
that, as things first got underway, it was Postel who meticulously tracked and published the
assignment of the ARPANET’s socket numbers. These numbers were loosely analogous to the
local extensions that might be served by a single phone number. But sockets were ordered
the same way on each host computer. It was as if every company with an internal phone
system always used the same distinct set of extensions for every office, from the CEO to the
mail-room. Everyone relied on Postel for the up-to-date directory.
Constructed out of memory segments, sockets were virtual openings into a
computer’s processes. Since they were only simplex (one-way) connections, they often had
to be used in pairs. Odd numbers were typically used on the server side, and even numbers
were used by the caller. Different socket numbers had to be reserved for specific server-side
conversations, applications such as TELNET, FTP, FINGER, Date and Time, Short Text
Messages, and so on.
To play out the analogy, it was as if someone were obliged to dial out from a specific
extension on his or her own computer to ask the time, and the computer that was called
would send the answer from another specifically designated extension. Later on, the
Jon Postel, “RFC 349: Proposed Standard Socket Numbers,” May 30, 1972 http://ftp.isi.edu/in-
invention of TCP opened the way for full duplex, two way connections across a single
channel, called a port.
In RFC 349, published in May 1972, Postel moved to formalize his role as both guide
and gatekeeper for numeric assignments on the ARPANET:
I propose that there be a czar (me ?) who hands out official socket numbers
for use by standard protocols. This czar should also keep track of and publish
a list of those socket numbers where host specific services can be obtained.58
Postel was acknowledged, ultimately, as the system’s “numbers czar,” a moniker
which was, in those days at least, a term of endearment. As time passed and the Internet
evolved out of the ARPANET, he remained in charge of the allocation of all “unique parameter
values” used by the Internet engineering community, including IP numbers, port addresses,
and the top of the domain name hierarchy. His control of that last resource made him the
power behind the root. His knowledge of how the all the different protocols and assignments
were interwoven and interdependent made him one of the few people on earth who could
simultaneously envision both the big picture and the critical minutiae of how the Internet’s
protocols worked. Cerf, Kahn, and others set the course, but Postel’s hand, perhaps more
than any other, steadied the Internet’s symbolic rudder as it evolved from the ARPANET in the
1960s to an engine of global commerce at the turn of the millennium.
Years later, he told a Congressional committee how his role in the early ARPANET
experiments evolved into a job of such pivotal significance.
Communication of data between computers required the creation of
certain rules ("protocols") to interpret and to format the data. These protocols
had multiple fields. Certain conventions were developed which would define
the meaning of a particular symbol used in a particular field within a
Collectively the set of conventions are the "protocol parameters." In
a project like the ARPANET with the developers spread across the country, it
was necessary to have coordination in assigning meaning to these protocol
See the prepared statement of Jon Postel Before the U.S. House of Representative, Committee on
Science, Subcommittee on Basic Research, “Internet Domain Names , Part 1,” September 25, 1997. For hearing
transcripts, see http://commdocs.house.gov/committees/science/hsy268140.000/hsy268140_0f.htm.
parameters and keeping track of what they meant. I took on the task of doing
In other words, Postel’s job was to keep people from stepping on each others’ toes.
Internet growth depended upon technical interoperability across potentially enormous arrays
of hardware and software. The rapidly expanding community required an explicit system for
setting and referencing new codes and protocols. Fortunately, most of the numbers and
symbolic values the engineers needed could be denoted in series, and allocated one after the
other. To avoid chaos and conflict, the programming community needed a reliable point of
contact where such numbers could be distributed and immediately recorded as taken. To
sustain such a high pace of invention, that contact had to be responsive and accessible.
Semantics had to be nailed down and published. The Internet needed a memory of what had
already been said and a system for describing what had been meant by it. Only with such
discipline could the curse of Babel be avoided.
To simply assign a meaning makes one a guide, but as the designated enforcer of that
discipline on behalf of a community – whether he was said to be coordinating, regulating or
controlling the assignments – Postel held an office equivalent to lord high gatekeeper.
Anyone intending to identify a new value knew that its official realization depended on
Postel making the proper pronouncements.
The only penalty for refusing to play the game with the czar’s numbers was the
inability to play with the people already using the czar’s numbers. This was the paradoxical
ambiguity of Postel’s power. Rules constrain, and rules enable. “Coordination in assigning
meaning” was necessary if people intended to play nicely together. Within the game, the
practical day to day act of coordinating meaning was essentially equivalent to regulation;
Postel generally exercised the final word. But playing the game at all was considered a
voluntary act. Outside of the game there was no formal effort by the players to compel
compliance. They counted on markets to do that for them. “The phrase ‘let the market
Dave Crocker, “Re: Stopping independent publications (Re: Comments on IESG charter and
guidelines drafts),” POISED March 4, 2003.
Scott Bradner, “RFC 2119: Key words for use in RFCs to Indicate Requirement Levels,” March
Harald Alvestrand, “RE: Impending publication: draft-iab-considerations-02.txt” IETF September
Joe Touch prepared a Curriculum Vitae shortly after Postel’s death. See
decide’ used to be the watchword in the IETF,” wrote one well–known insider. “Our job is
to do engineering, not make market decisions.”60
One’s relationship with the engineering community depended on proof of fealty. In
later years, RFC authors started using words like “MUST” and “SHOULD” to express what
it took to conform with an IETF standard. MUST indicated an “absolute requirement.” It
was to be obeyed instantly and without question, not unlike a soldier hearing an order from
a superior officer. The approach took firm hold within the IETF culture, though such
directives had far lower standing beyond it. According to Harald Alvestrand, IETF chair in
the early 2000s, “a MUST in an RFC has no enforcement mechanism whatsoever.”
The most that can be said is something akin to “if you do the opposite of what
these words say, and claim to be conformant to this RFC, we will laugh at
Within the parameters of the Internet standards game, there was no effective
difference between the meaning of coordination and regulation. For people who loved the
game of using and inventing Internet standards, there was no need to worry about the
distinction. The players at the table needed a dealer, an enabler who would guarantee a fair
turn. Postel made a career of performing that function. His standing was bound to increase
as the Internet became the most important game in town.
Postel moved around a bit after completing his Ph.D. at UCLA in 1974, working for
various defense contractors on ARPANET-related projects. The longest stint was at SRI,
Hafner and Lyon (1996: 232-3).
Postel’s Curriculum Vitae, http://www.postel.org/postel-cv-1997.txt.
where he worked with Doug Engelbart (standing in the third row of BBN’s commemorative
picture) at the Augmentation Research Center (ARC). If the Internet had a spiritual
birthplace, it was there. The ARC had been founded by another ARPA scientist and BBN
engineer, the late J.C.R. Licklider. “Lick,” as he was known, was perhaps the most far
thinking of them all, writing about intergalactic networks and human computer symbiosis
when transistors were barely out of the laboratory. Licklider and Engelbart shared a vision
of the computer as a device that would extend human communication and augment the
human intellect. ARC is where the mouse and the concept of windows were invented. It was
also the place where numerous researchers, Postel included, learned to think aggressively
about how the use of computers could revolutionize human civilization.
Since ARPA’s mission was to focus on leading-edge experimental projects, the
ARPANET had to be turned over to a new home once it was up and running. Taylor tried to
convince AT&T that taking it would be a worthwhile commercial investment, but he was
unsuccessful. Instead, responsibility was transferred to the Defense Communications Agency
(DCA), a military operations group which was developing its own IMP network with BBN’s
help. Postel was still working at SRI at that time. With the shift, work on the ARPANET
system took on the sensibilities of a routine mission. The new, more practical orientation
undercut the earlier trail-blazing feel.
By this time ARPA had been renamed as the Defense Advanced Research Projects
Agency (DARPA). The addition of the word “defense” didn’t undermine the agency’s
commitment to cutting-edge, avant-garde research, nor its ability to tolerate an intellectually
open environment that had room for relatively nontraditional and nonconformist
personalities. In 1977 Postel began working at a DARPA-funded “think-tank” – the
University of Southern California’s Information Sciences Institute (ISI). He eventually
become ISI’s Associate Director for Networking. ISI was far removed from USC’s campus
and academic life, leaving Postel largely free of the teaching requirements that can preoccupy
Personal interview with Vint Cerf, November 15, 2001.
Rony and Rony. (1998: 122-3.)
Cerf interview, November 15, 2001.
“Contract Between ICANN and the United States Government for Performance of the IANA
Function, US Dept. of Commerce Order Number 40SBNT067020,” http://www.icann.org/general/iana-
contract-09feb00.htm. See also Brian Carpenter, Fred Baker, Mike Roberts, “RFC 2860: Memorandum of
Understanding Concerning the Technical Work of the Internet Assigned Numbers Authority,” June 2000.
faculty members. People who needed number or parameter assignments knew to contact him
there. Over time the contact point for those resources came to be known as the IANA – the
Internet Assigned Numbers Authority.
DARPA contracts tended to be rather generalized, conglomerating the work that
ultimately constituted the RFC Editor, IANA and other tasks. That precedent was set by Cerf
during the years he was in charge at DARPA.“In all the time that I was writing and managing
the ISI R&D activity,” Cerf recalled, “I don’t believe we ever used the term IANA. But what
we did was to write very broad language about the areas in which we expected research to
The convention was followed by Cerf’s successors. The term IANA did not even
appear in DARPA budget documentation until 1993, and then only buried within the contract
for the Teranode High Performance Computing Project.
IANA’s emergence as a named institution occurred gradually. Because of this, and
because of its close identification with Postel as an individual, its rise has been the subject
of extended controversy. For insiders IANA was Jon. For outsiders it was an official agency
with real standing. Even if the confusion wasn’t intentional, little was done to correct it.
“Jon’s position as IANA was always a little ambiguous,” Cerf admitted, “and we always kind
of left that way.” IANA was finally made the subject of a contract between the Department
of Commerce and ICANN in February 2000.69
IANA’s enduring authority stemmed from a web of self-referential, and similarly
quasi-official relationships that had operated within the Internet engineering community
since the earliest ARPANET days. Even though US Government funding sustained Postel’s
activities at ISI for many years, his long-run potency as a community leader attested to a
broader social foundation.
Members of the Internet engineering community established standards of merit by
conferring recognizable distinctions on each other. Any anthropologist or sociologist would
recognize the process. In this case, the engineers were building up multifaceted stocks of
culture – a legacy of titles, stories of achievement, citations in RFCs, the binding ties that
result from participation in contracts and agreements, and so on. There was even a self-
referential style of etiquette called “netiquette.” These social inventories served to bootstrap
status relations as the community expanded. Over time, certain patterns of respect took hold
and deepened. Knowledge of those patterns, such as who had the highest status, was essential
for anyone seeking membership in the community. The ability to identify high status insiders
– and, even better, get close to them – was especially useful for those who desired to climb
There is an expression, “Let’s run it up the flagpole and see who salutes.” That was
how authority worked on the Internet. People had, with good reason, gotten into the habit of
saluting flags raised by Cerf, Postel and any others who could regularly demonstrate
expertise in specific domains and convey that expertise with strong communication skills.
That was a pragmatic basis for getting real work done. Newcomers were generally quite
willing to adopt those deferential habits and even promote them. Anyone who made a valid
goal-oriented demonstration of expertise and competency could be elevated by grants of
doctrinal respect. Those displays took on a life of their own. To get along, one had to adopt
the prevailing manner. There was little patience for puffery and lack of acuity. The heralded
openness of the Internet Engineering community implicitly contained an invitation to
Bob Braden, a long-time colleague of Postel’s at ISI, fondly recounted an incident
that makes the case. Braden had challenged the style of endnote references that RFC authors
were required to follow. Postel’s response, recalled Braden, “was equivalent to ‘Get used to
Bob Braden, “Re: rfc-ed reference style,” IETF March 21, 2003.
Jon Postel, “RFC 204: Sockets in use,” August, 1971.
it!’” Braden did, and for years thereafter, urged others to do the same. He had decided that
habits that worked well enough over a long time deserved a standing beyond a fashion of
manners – they were imbued with legitimacy.
This entrenchment of particular styles, manners, and habits of obedience is a
mechanism by which particular societies come to recognize themselves as such. Not all
people are fully aware of how their own individual deeds underlie and reproduce social
conceptions of legitimacy. Background conceptions, after all, are often mistakenly taken as
pre-existing and historically absolute. Nevertheless, Braden’s recollection reflects the way
in which many people do acquire some hazy self-consciousness of their social agency. They
are enlightened enough to affirm their actions as constitutive, insisting that certain behaviors
be upheld as virtuous by their very constitutiveness.
As in any society, the dogma of reputation took on a life of its own. The IANA was
as much a cultural phenomenon as it was a formalistic vestige of the Internet’s routinization.
* * *
In fact, the IANA function performed at ISI was a steady reification of the work
Postel had been doing since he started organizing the entries in IMP logbook. Postel was
always around, policing records and keeping things tidy. An early sign of what was to come
institutionally, even before he volunteered to be Czar, is found in RFC 204, “Sockets in use”
which he published in August 1971.
I would like to collect information on the use of socket numbers for
"standard" service programs. For example Loggers (telnet servers) Listen on
socket 1. What sockets at your host are Listened to by what programs?
Recently Dick Watson suggested assigning socket 5 for use by a
mail-box protocol (RFC196). Does any one object ? Are there any
suggestions for a method of assigning sockets to standard programs? Should
a subset of the socket numbers be reserved for use by future standard
She described their 15 ½ year long working relationship this way:
My fondest "story" about how the world looked at Postel & Reynolds as IANA and RFC Editor came from one
of our Internet friends/colleagues. This person sent an email message to Jon and I one day stating, "Please don't
take this as an insult, but you two work so seamlessly together I can't tell who is the IANA and who is the RFC
Editor? So, who does what? Which one of you administers IANA? Who works on the RFCs?" Jon and I were
sitting side by side as usual, reading this email together. Jon turned and looked at me with a big grin on his face,
turned back to the keyboard and started typing a reply. It was one word, "Yes." To this day, I took his response
as a wonderful compliment of how he felt about our work together.
Internet Architecture Board, “RFC 1083, IAB Official Protocol Standards 1, Internet Engineering
Task Force,” December 1, 1988.
Postel asked that comments be sent to him via “The SPADE Group” at UCLA’s
Boelter Hall. The name hinted at the groundbreaking work the engineering students there
believed they had undertaken.
The socket table was updated periodically through the seventies. Other lists such as
“Link Numbers” were added in along the way. The title changed over time, but stabilized as
“Assigned Numbers” in RFC 739, which Postel published in November 1977.
Each RFC in the Assigned Numbers series included a politely phrased directive, “If
you are developing a protocol or application that will require the use of a link, socket, etc.
please contact Jon to receive a number assignment.” That state of affairs persisted until RFC
870 was published in October 1983, the same year TCP/IP became the official protocol of
the ARPANET. The RFC Editor and parameter assignment tasks had grown demanding
enough to justify assistance from another ISI staff member, Joyce Reynolds, who had been
working there since 1979. Starting with RFC 870, Reynolds was named as the primary
author and contact. Postel remained on as coauthor and continued to drive the task of
organizing parameter lists and making assignments. Though Postel was the better known
personality in the assigned numbers business, the importance of Reynolds’ contribution and
the depth of their collaboration was not well understood outside the immediate community.72
Five years later, with the publication of RFC 1083 in December 1988, the “Internet
Assigned Numbers Authority” was finally referenced in print. The term also appears in a
lower numbered RFC, 1060, but that RFC shows a publication date out of sequence, in
March 1990. In any case, by the end of the decade the use of the term IANA was coming in
Phone interview with Joyce Reynolds.
Mueller( 2002: 93). A slide from Rutkowski, “History of Supporting Names and Numbers,” includes
the text “ISI denominates Postel DARPA role as IANA (1988).
http://wia.org/pub/identifiers/identifier_management.gif. See also Rutkowski’s “US DOD [Internet] Assigned
Numbers [Authority]*, Network Information Centers (NICs), Contractors, and Activities: known detailed
to vogue at ISI and across the Internet. The “assignment of meaning” had acquired an
Reynolds couldn’t recall precisely when she first heard the term IANA used as an
acronym. “One day Jon just said, ‘Instead of telling people, ‘Call Jon, or call Joyce for a
number,’ they’ll be told to call the IANA.’” Perhaps the renegotiation of the DARPA
contract in 1988 prompted a decision to adopt a more distinctly institutional look for this
aspect of their work at ISI. In any case, the acronym turned out to be an apt if inadvertent
play on the suffix which means “little bits of things,” as in Floridiana, or Americana. If there
is no pivotal moment of IANA’s inception to point out, it is because the long running
continuity of the assignment activity was so much more important to the members of the
burgeoning engineering community than commemorating a date of denomination.
This action might not be possible to undo. Are you sure you want to continue?