‘Rough Consensus and Running Code’

and the Internet-OSI Standards War

Andrew L. Russell
The Johns Hopkins University

Internet historians recognize the technical achievements but often

overlook the bureaucratic innovations of Internet pioneers. The
phrase, “We reject: kings, presidents, and voting. We believe in:
rough consensus and running code,” was coined by David Clark in
1992. This article explains how the phrase captured the technical and
political values of Internet engineers during a crucial phase in the
Internet’s growth.

Historians interested in conflict and consensus overlooked, however, is the extent to which the
in technological systems stand much to gain success of the Internet depended on organiza-
from examining standardization—the process- tional innovations directed by Cerf, Kahn, and
es of creating and implementing technical stan- other Internet pioneers such as Jon Postel and
dards. Standardization requires sophisticated David Clark. Starting in the 1970s, researchers
technical work as well as organizational and in groups such as the Internet Configuration
strategic coordination. In the late 1970s, it was Control Board (ICCB), the Internet Activities
reasonable to assume that formal standards-set- Board (IAB), and the Internet Engineering Task
ting institutions, such as the International Force (IETF) coordinated Internet standards
Telecommunication Union (ITU) and the and architectural development. These institu-
International Organization for Standardization tions deserve special attention not only for
(ISO), would lead the coordination and stan- their technical achievements, but also because
dardization of information services, much as they fostered a voluntary international con-
they did for international telecommunications. sensus during a period of intense technical
Beginning in 1977, ISO oversaw a large, ambi- and institutional change in computing and
tious attempt to define a network architecture telecommunications.
called Open Systems Interconnection (OSI). These organizational innovations emerged
Between the late 1970s and the early 1990s, the as responses to external competition from ISO
OSI seven-layer model became enshrined in as well as to internal strains in the processes of
computer science curricula and endorsed by Internet standardization. As the Internet grew
governments around the world. Competing rapidly in the late 1980s and 1990s, groups such
networks—including experimental TCP/IP net- as the IAB and IETF started to become victims
works—were expected to fade away once OSI of their own success and struggled to preserve
protocols were standardized and implemented their founding principles. If we understand
by users and manufacturers. “politics” to mean relations of control between
By the mid-1990s, however, ISO’s slow stan- individuals and groups, the political values of
dardization process had failed to keep up with Internet architects and engineers were especial-
alternative, informal mechanisms that were ly evident as these people built political struc-
more effective at coordinating rapid techno- tures—standards bodies—from scratch.3
logical change.1 These informal mechanisms— Standardization is technically and organiza-
the focus of this article—provided vital tionally complex as well as deeply value-laden.4
institutional support for the eventual success Leaders and participants in the IAB and IETF
and global deployment of the Internet archi- articulated institutional rules, cultural tradi-
tecture and Internet standards such as TCP/IP. tions, and versions of their own history as they
In most existing histories of the Internet, responded to challenges from within and with-
leaders such as Vinton Cerf and Robert Kahn out. A memorable phrase encapsulates the pre-
receive plaudits for their technical work, espe- vailing technical and organizational values of
cially for their roles in the creation of the core those who were involved in the Internet stan-
Internet standards TCP and IP.2 What is often dards process from the mid-1970s to the pres-

48 IEEE Annals of the History of Computing Published by the IEEE Computer Society 1058-6180/06/$20.00 © 2006 IEEE
ent: “We reject: kings, presidents, and voting. Arpanet (PRnet was developed in Hawaii), but
We believe in: rough consensus and running faced a problem of trying to connect what he
code.”5 This article explains how this phrase— later called “two radically different networks”
which became a motto for Internet standardi- with different network capacities, protocols,
zation—articulated a common set of beliefs and routing systems.10 ARPA’s concurrent
about the work culture and engineering style of development of satellite packet switching in
Internet standardization. SATnet further compounded this problem,
In 1999, law professor Lawrence Lessig leading Kahn to conclude that network inter-
declared that “rough consensus and running connection could not be achieved on an ad
code” had broad significance as “a manifesto hoc, network-to-network level.
that will define our generation.”6 An examina- To overcome this problem, Kahn in 1973
tion of the origins of this credo—coined by rekindled an effective partnership with Vint
David Clark in 1992—illustrates some techni- Cerf—the two had worked together on the first
cal, rhetorical, and philosophical differences nodes of the Arpanet in 1969—and proposed a
between the Internet standards process and the new way to transport data packets. This mech-
competing ISO process. Most participants in the anism was a simple technical protocol (trans-
Internet standards process today consider mission control protocol, or TCP) and system
“rough consensus and running code” to be a of gateways (now known as routers) to transfer
succinct and accurate description of the data packets between the dissimilar networks.11
Internet standards process.7 For historians, Kahn’s ultimate goal was to make the network
“rough consensus and running code” stands as transparent, “a seamless whole,” invisible to
a revealing depiction of the international poli- the user who would be more interested in
tics of computer internetworking in the last accessing information over the network instead
quarter of the 20th century, as well as a point of of the operation of the network itself.12 This
entry for investigating why contemporaries basic principle—interconnection via standard
described the competition between the Internet protocols—is the Internet’s fundamental struc-
and OSI as “the Internet religious war.”8 ture and defining feature.13 In 1978, Cerf and
two other DARPA-funded network researchers,
Internet architecture: TCP/IP and the Danny Cohen and Jon Postel, split the func-
end-to-end argument tions of TCP into two protocols, TCP and the
In 1996, the Internet Architecture Board Internet Protocol (IP), that worked together in
(IAB) published RFC 1958, “Architectural the now-familiar combination of TCP/IP.14
Principles of the Internet,” to record aspects of Network transparency and application
architectural approach practiced in the Internet autonomy are the two key elements of a design
community over the previous 25 years. They philosophy first articulated in 1981 by David
wrote: “in very general terms, the community Clark, Jerome Saltzer, and David Reed, three
believes that the goal is connectivity, the tool veterans of ARPA-sponsored networking exper-
is the Internet Protocol, and the intelligence is iments at the Massachusetts Institute of
end to end rather than hidden in the net- Technology’s (MIT’s) Laboratory for Computer
work.”9 This section provides a brief overview Science. Their paper, “End-To-End Arguments
of the distinctive aspects of Internet architec- in System Design,” was written to explore a
ture by exploring these three principles: inter- design principle that they claimed had “been
connection, the Internet Protocol, and the used for many years with[out] explicit recog-
“end-to-end” argument. nition.”15 An outgrowth of substantial experi-
After initial networking experiments with ence with TCP/IP networking, the end-to-end
the Arpanet beginning in 1969, the Advanced principle held that the Internet’s complex
Projects Research Agency (ARPA) continued to functions should be performed at the end-
explore networking concepts for satellite and points, leaving only the (relatively) simple
radio. The problem of how to enable commu- tasks of interconnection and data transport to
nication between technically disparate systems the network. The authors commended the
also had implications for mobile military com- technical merits and simplicity of the end-to-
munications. In the early 1970s, ARPA devel- end model by concluding, “end-to-end argu-
oped a packet radio network (known as PRnet) ments are a kind of ‘Occam’s razor’ when it
based on the same packet-switching technolo- comes to choosing the functions to be provid-
gy being tested in the Arpanet. Robert Kahn, as ed in a communications subsystem.”16
program manager for this project, identified The end-to-end design principle thus calls
the need to connect packet radio networks to for a simple standard for data transfer that
large computers in the continental US via the allows new innovations to be added at the

July–September 2006 49
 ‘Rough Consensus and Running Code’

edges and on top of the “stupid network.”17 sion formally to expand the involvement of the
From this distributed design follows decentral- research community in making decisions about
ized control, both over the functionality of the the network:
network and over the content of the network
traffic. Marjory Blumenthal and David Clark [W]hen we started the internet program in the
argued in 2001 that end-to-end principles form mid 1970s, originally it was just me in the office
the heart of the running the program. And after Vint was hired,
then it was just Vint running the program with
conventional understanding of the “Internet me to kibitz. And he was so good at what he did
philosophy”: freedom of action, user empower- that he basically had everything in his head.
ment, end-user responsibility for actions under- What I worried about was what would happen if
taken, and lack of controls “in” the Net that limit he got hit by a truck? Number two, what would
or regulate what users can do.18 happen if he would ever have to leave? And
number three, how was anybody else in the com-
By building in minimum control and allow- munity ever going to be part of the thinking
ing the “broadest range of development” to process. So he set up, after some discussions, a
end users, the technical standards of the kind of kitchen cabinet, if you will, of knowl-
Internet have, as Lessig wrote, “erected a free edgeable people that he would convene periodi-
space of innovation.”19 cally. These were mostly the workers in the field,
On 1 January 1983, TCP/IP replaced NCP (an the key people who were implementing proto-
older networking protocol) as the standard host cols . . . When he left, that group stayed intact.24
protocol for the Arpanet.20 “With the use of
TCP/IP now mandatory on the Arpanet,” Kahn This group, Cerf’s “kitchen cabinet,” was the
recalled in 1994, “the rationale for connection ICCB, created in 1979 and “chaired by David
of local area networks was evident and the Clark from MIT to assist ARPA in the planning
growth of users and networks accelerated. It also and execution of the evolution of the TCP/IP
led to a rethinking of the process that ARPA was suite.”25 The ICCB expanded control over
using to manage the evolution of the net- Internet development by bringing more of the
work.”21 The next section examines the evolu- network users—technical experts distributed in
tion of this management process as the Internet universities, firms, and government agencies—
community faced the problems and challenges into Cerf’s inner circle in a more formal way.
that accompanied the growth of the network. For Kahn, the ICCB was important because it
“brought a wider segment of the research com-
Governance of Internet protocols, munity more directly into the decision-making
1979–1992 aspects of the Internet project which, until
Internet researchers created and presided then, had been almost solely undertaken by
over the governing bodies of Internet proto- ARPA.”26
cols—the ICCB, IAB, and IETF—to manage the Cerf left his position as head of the Internet
growth of end-to-end networks based on TCP/IP. research program at ARPA in 1982, and was
The history of these groups reveals the creation replaced in 1983 by Barry Leiner. Leiner and
of increasingly formal structures by a core of Clark, in response to the continued growth of
Internet researchers who strove to preserve the the Internet community, disbanded the ICCB
design principles and work culture that had fos- in September 1984 and created the Internet
tered the Internet’s successful growth. Advisory Board (IAB).27 Clark continued his
Cerf’s involvement with ARPA started at close involvement as the first chair of the IAB—
Stanford University, where he worked from a position that confers the title of “Internet
1974 to 1976 to implement TCP/IP. Kahn, who Architect.”28 Leiner created task forces within
joined ARPA’s Information Processing the IAB to focus on specific aspects of Internet
Techniques Office as a program manager in technology (such as gateway algorithms, end-
1972, hired Cerf to come to ARPA in 1976 to to-end protocols, and security) in an attempt to
run a program to pursue and coordinate keep discussions focused while accommodat-
Internet research. Cerf later recalled that, upon ing growing numbers of participants.29 The
moving to ARPA, “The single-minded goal was Board itself consisted of the chairs of the task
to get the Internet system up and running.”22 forces. There were no elections in the IAB.
When he began at ARPA, Cerf worked with Instead, as Vint Cerf explained in 1989, new
“only a few researchers” to “develop and test members were appointed by the chairman of
versions of the internet protocols.”23 Kahn, the IAB, with the advice and consent of the
Cerf’s manager at the time, recalled the deci- remaining members. Cerf wrote,

50 IEEE Annals of the History of Computing

 [m]embership changes with time to adjust to the
current realities of the research interests of the List of Acronyms
participants, the needs of the Internet system ARPA Advanced Research Projects Agency
and the concerns of the constituent members of CLNP ConnectionLess Network Protocol
the Internet.23 DARPA Defense Advanced Research Projects Agency
DoD Department of Defense
Given this description, it is not difficult to GOSIP Government Open Systems Interconnection Profile
see why the author and computer scientist Ed IAB Internet Advisory Board (1984–1986); Internet Activities
Krol described the IAB as a “council of elders.”30 Board (1986–1992); Internet Architecture Board (July
The IAB cannot be characterized as a democra- 1992–present)
cy, since nobody voted and the Board only let ICCB Internet Configuration Control Board
in the people they wanted. The very premise of IESG Internet Engineering Steering Group
the Internet—especially its protocol and net- IETF Internet Engineering Task Force
work design—required that the Board make its IPv4 Internet Protocol version 4
decisions by consensus. Democracy, with its ISO International Organization for Standardization
competing factions and its political compro- ISOC Internet Society
mises, was not an appropriate political model ITU International Telecommunication Union
for the IAB or the Internet. Instead, the IAB NCP Network Control Program
operated with a leadership of experienced tech- NIST National Institute of Standards and Technology
nicians and a rank and file organized by area of OSI Open Systems Interconnection
technical interest, more technocratic than PRnet Packet Radio Network
democratic. The IAB increasingly served as the RFC Request for Comments
steward of TCP/IP and the Internet, but had no TCP/IP Transmission Control Protocol/Internet Protocol
legal mandate or enforcement mechanisms. In TP-4 Transport Protocol Class 4
other words, IAB-backed protocols were de
facto standards, whose status as standards
depended on broad consensus and widespread
implementation. increasing size, value, and internationalization
One of the IAB’s task forces, the Internet of the IETF and Internet standards process
Engineering Task Force, first met in 1986. Due brought significant legal and practical prob-
to the tremendous growth in the Internet’s lems to the fore, such as antitrust liability,
practical and engineering side, there was soon copyright protection, and the needs to detach
“an explosion in the attendance at IETF meet- the process from the US government and to
ings” that compelled IETF chair Phill Gross to accommodate international participation.33 In
create a substructure for the group. In 1987, the 1990, Cerf began to formulate the idea of a pri-
IETF formed separate working groups to oversee vate board of overseers that would act as a legal
specific topics; in 1989 the IETF organized the and organizational umbrella for the IAB and
working groups into areas and designated area IETF and, at the same time, serve as a global
directors, who formed the Internet Engineering coordinating mechanism for promoting the
Steering Group (IESG). The IETF’s effectiveness Internet. This umbrella, the Internet Society,
was underscored that same year, when the IAB was launched as a private, professional mem-
“recognized the importance of the IETF, and bership organization in January 1992, with
restructured the standards process explicitly to Cerf as its president. In July 1992, the IAB
recognize the IESG as the major review body for changed its name from the Internet Activities
standards.”31 Having assigned responsibility to Board to the Internet Architecture Board and
the IETF for the short-term engineering of the became part of the Internet Society.34
Internet, the IAB streamlined the rest of its task A “Brief History of the Internet”—coau-
forces into the Internet Research Task Force with thored by a number of Internet pioneers
small research groups, such as the End-to-End (including Cerf, Clark, Kahn, and Leiner)—
Research Group and Internet Architecture Task describes this formation of community gover-
Force, dedicated to long-term issues in the evo- nance as a “steady evolution of organizational
lution of the Internet.32 structures designed to support and facilitate an
By the early 1990s, participation at IETF ever-increasing community working collabora-
meetings—which were held three times a year tively on Internet issues.”35 These structures
and open to anyone interested—continued to combined two models of governance. The first
increase at an explosive pace, reflecting grow- model, the structure led by Cerf that coordi-
ing interest from the research community as nated the development of the Internet at ARPA,
well as from the commercial community. The was a self-selected, experienced group—a

July–September 2006 51
 ‘Rough Consensus and Running Code’

“council of elders.” Historians of the Internet the TCP/IP Internet and the open systems
unequivocally praise this group as a source of interconnection (OSI) model were designed as
its astounding growth. Frequently described as nonproprietary networks that would allow
a meritocracy, this close-knit network of peo- users more flexibility. Although they shared
ple worked together since the early Arpanet some common architectural features, these two
days (many as graduate students at MIT or the models—the Internet and OSI—emerged with-
University of California, Los Angeles, or as in different organizational contexts under
engineers at Bolt Beranek and Newman, the somewhat different motivations. This section
consulting firm that had designed aspects of describes the ideas behind the development of
the Arpanet) and provided the bulk of the tech- OSI, and its adoption by the US Department of
nical and bureaucratic leadership necessary to Defense (DoD).39
keep the Internet up and running.36 (For more OSI is an example of a broader movement
on BBN, see the special issues of Annals: vol. 27, toward “open systems” that “encouraged com-
no. 2 and vol. 28, no. 1.) “Relatively few, com- patibility between systems from different man-
petent, highly motivated people were ufacturers.”40 The emergence of many types of
involved,” recalled Larry Press, “and they had open systems in the 1970s and 1980s (such as
considerable autonomy.”37 for stereo components and microcomputers)
Kahn’s observation about the ICCB’s role as helped smaller producers and consumers who
a sort of kitchen cabinet provides an insight did not want to be locked in to proprietary
into the second model of governance, which products from a single manufacturer.41 For
has received less attention from historians: the computer networks, open systems such as OSI
function of the ICCB, IAB, and IETF as mecha- and the TCP/IP Internet emerged as alternatives
nisms for engaging and directing the efforts of that could challenge the dominance of IBM
a distributed group of Internet researchers. The and its System Network Architecture. These
Internet user community was small enough in various standards coexisted throughout the late
1979 that the ICCB functioned simultaneous- 1970s and 1980s, as networking executives and
ly as both a “council of elders” and a “grass- engineers developed strategies and organiza-
roots mechanism” for Internet standards.38 By tional structures to facilitate network intercon-
1992, the IAB and Internet Society maintained nection.42 In short, open systems articulated a
the character of the council of elders, responsi- network interconnection strategy that at the
ble for architectural and bureaucratic oversight, same time facilitated a more decentralized
while the IETF, as its name implies, assumed industry structure.
responsibility for the distributed, hands-on OSI dates from 1977, when, according to
tasks involved in the engineering and imple- one of its protagonists, the ISO “recognized the
mentation of protocols and provided a forum special and urgent need for standards for het-
for interested newcomers. erogeneous informatic networks” and formed
The transfer of responsibility for technical a subcommittee to create a Reference Model of
standards from the ICCB to the IAB and then Open Systems Interconnection.43 The goal of
to the IETF demonstrated a strong desire on the this seven-layer OSI Reference Model was not
part of the council of elders to engage and to define the internal operations of networks,
empower the broader community that wanted but only to standardize the external interfaces
to contribute to the further development of the between networks: in other words, to set the
Internet. Under this system, the Internet’s ground rules for network interconnection.44
architectural oversight remained with the During the 1980s and early 1990s, OSI
reconstituted IAB, while the efforts of partici- enjoyed widespread support from national gov-
pants in the IETF were channeled toward the ernments, particularly in Western Europe,
technical aspects of protocol development and North America, and the Far East.45 OSI enjoyed
implementation. this level of support due in part to the strategic
position of its sponsor, ISO. ISO was an “offi-
OSI in Europe and GOSIP in the US cial” international standards body, meaning
This development of TCP/IP networks with- that it was populated by representatives from
in the ICCB, IAB, and IETF occurred while national governments who, in most cases,
other groups developed competing protocol acted on behalf of the interests of their nation-
architectures for computer internetworking. al telecommunications and computer firms.46
Vendors such as IBM and DEC (Digital ISO’s organizational culture—concerned with
Equipment Corporation) offered proprietary defining and controlling the future of infor-
networking solutions that were difficult if not mation and telecommunication services on
impossible to connect to other networks. Both behalf of its representatives from national gov-

52 IEEE Annals of the History of Computing

ernments—resembled contemporary demo-
cratic bodies insofar as it featured voting, par-
tisan compromises, and rule-making behavior OSI continued to gain
designed to protect financial interests. Such
processes stand in stark contrast to the research momentum in August 1988,
and military orientation of the people and
institutions that developed Internet protocols. when the National Bureau of
The international support for OSI influ-
enced the officials in the DoD who were Standards published a
responsible for procuring equipment for their
own computer networks.47 Even though they version of OSI for US federal
had sponsored the development of Internet
standards, DoD managers believed in the early agencies called GOSIP.
1980s that OSI networks were likely to emerge
as de facto and de jure global standards. To bet-
ter understand the competing standards, the Tensions between the OSI and Internet com-
DoD asked the National Research Council in munities became apparent in the early 1980s,
1983 to evaluate TCP and TP-4, its functional well before GOSIP came into being. For exam-
counterpart in the OSI Reference Model. The ple, in their 1983 paper describing the similar-
final 1985 report presented three options: keep ities between the ARPA and ISO protocol
the two as costandards; adopt TP-4 as soon as architectures, Danny Cohen and Jon Postel
it was shown to be ready for military networks; painted the ISO model as an abstraction, far too
or keep TCP and defer indefinitely a decision rigid in its reliance on seven interrelated levels,
on TP-4.48 The DoD supported the second and inappropriate to be used “as a model for all
option and planned to “move ultimately seasons.”54 In an unusually colorful paper,
toward exclusive use of TP-4.”49 Cohen and Postel—both of whom were instru-
OSI continued to gain momentum in mental in the early history of TCP/IP—mock-
August 1988, when the National Bureau of ingly speculated that “mystical” traditions such
Standards published a version of OSI for US fed- as Early Zoroastrianism, New Testament celes-
eral agencies called GOSIP (for “Government tial beings, and the Christian seven deadly sins
Open Systems Interconnection Profile”) version might have “shaped the choice of Seven.”55
1, which was built around TP-4 and many OSI Another Internet advocate, in his 1991 “tech-
specifications. By August 1990, federal agencies nical travelogue” of networking in 21 countries
were required to procure GOSIP-compliant across the world, suggested that trying to
products.50 Through this procurement require- implement OSI over slow, low-quality lines was
ment, the government intended to stimulate “akin to looking for a hippopotamus capable of
the market for OSI products.51 However, many doing the limbo.”56
network administrators resisted the GOSIP pro- The resentment of Cohen, Postel, and their
curement policy and continued to operate Internet colleagues stemmed from their frustra-
TCP/IP networks, noting that the federal man- tion with the technical aspects of OSI as well as
date, by specifying only procurement, did not with ISO as a bureaucratic entity. Where TCP/IP
prohibit the use of products built around the was developed through continual experimen-
more familiar and more readily available tation in a fluid organizational setting, Internet
TCP/IP.52 engineers viewed OSI committees as overly
bureaucratic and out of touch with existing net-
‘OSI Bigots’ and ‘IP Bigots’: Cultural works and computers. OSI’s political and formal
dimensions of a standards war process did not endear the TCP/IP Internet
A spirited rivalry between respective advo- community—who were accustomed to a decen-
cates of OSI and TCP/IP networks emerged as tralized division of labor throughout the stan-
they fought for jurisdiction over standards for dards process—to the ISO Reference Model. In
computer internetworks. Richard des Jardins— a scathing 1985 critique of OSI and its advo-
an early contributor to the ISO Reference cates, one veteran of the Arpanet and Internet
Model and president of the GOSIP Institute— community, Mike A. Padlipsky, characterized
captured the intensity of this rivalry when, in the ARPA Internet Reference Model as
a 1992 article, he compared the “OSI Bigots” “Descriptive” and ISO Reference Model as
and the “IP Bigots” to people who objected to “Prescriptive.” Another networking pioneer,
“the convergence of cultures and races in the David Mills, agreed in a 2004 interview:
world at large.”53 “Internet standards tended to be those written

July–September 2006 53
 ‘Rough Consensus and Running Code’

of the Internet supported the community’s

broader interests and the IAB’s mandate—
Apart from the external OSI keeping the network open for anyone who
wanted to connect.63
threat, the Internet
Internet’s constitutional crisis
community faced many During its June 1992 meeting in Kobe,
Japan, the IAB developed a draft discussion pro-
internal problems posal to use CLNP as the basis for a larger
address space. To the IAB, this seemed a respon-
throughout the Internet’s sible path to take, given the limits of the IPv4
address space and the desire for the Internet to
rapid growth during the accommodate as many users as possible. As IAB
member Christian Huitema recalled in 1996,
1980s and early 1990s.
The IAB discussed [the draft proposal to incor-
porate CLNP] extensively. In less than two weeks,
for implementers. International standards were it went through eight successive revisions. We
written as documents to be obeyed.”57 “Another thought that our wording was very careful, and
way of putting it,” Padlipsky wrote, “is that we were prepared to discuss it and try to con-
whereas the Descriptive approach is suitable for vince the Internet community. Then, everything
technology, the Prescriptive approach is suitable accelerated. Some journalists got the news, an
for theology.”58 announcement was hastily written, and many
Apart from the external OSI threat, the members of the community felt betrayed. They
Internet community faced many internal prob- perceived that we were selling the Internet to the
lems throughout the Internet’s rapid growth ISO and that headquarters was simply giving the
during the 1980s and early 1990s. As Internet field to an enemy that they had fought for many
advocates battled against OSI, they also contin- years and eventually vanquished. The IAB had
ued to struggle with the organizational problems no right to make such a decision alone.64
of their own standardization process. Some crit-
ics felt that the IAB “failed at times to provide a To the “general membership of the IETF,” Carl
solid agenda and timetables of engineering Cargill commented, “this was rank heresy.”65
problems” for the IETF to address.59 The infor- Scott Bradner preferred a political metaphor to
mal character of the IAB’s oversight of the IETF the military and religious metaphors used by
had created problems in the past, especially Huitema and Cargill: he later referred to the
when IETF engineers perceived that IAB deci- events that followed as a “constitutional cri-
sions favored the commercial interests of ven- sis.”66 The technical discussion proposed by the
dors over the technical consensus of the IETF.60 IAB inspired a movement within the IETF that
Tensions between the IAB and IETF exacer- challenged the organizational hierarchies of
bated a disagreement about a major technical Internet standards. The fact that a mere propos-
obstacle to Internet growth. One of the chief al—it is important to note that the IAB’s CLNP
concerns of Internet architects in the late 1980s discussion was in no way a final decision—
revolved around addressing and routing prob- would provoke such outrage from hundreds of
lems built into the current version of TCP/IP engineers and computer scientists reflects the
(IPv4): the finite amount of address space in passion and commitment of engineers in the
IPv4 was projected to be running out quickly. pitched battle of a standards war.
If the exponential growth of Internet users con- Many IETF participants, while aware of the
tinued, the bottleneck would prevent new con- limitations of IPv4, assumed that TCP/IP and
nections to the network.61 The IAB perceived OSI “will coexist for a long time,” and certain-
that a solution might be reached through the ly did not anticipate that another group would
OSI functional counterpart to IP called attempt to change the Internet’s fundamental
ConnectionLess Network Protocol, or CLNP.62 protocol without first consulting the IETF.67
Although it was aware of strong opposition Tradition, supported by IAB and IETF docu-
to OSI within the Internet community, the IAB mentation, gave the IETF the right to stan-
felt that working with CLNP could help the dardize protocols. Additionally, the IETF,
Internet overcome the address space problem. consisting mostly of academic and government
From this perspective, the technical rationale researchers, resented that OSI was a complex
for incorporating CLNP into the architecture and costly system, driven by the political con-

54 IEEE Annals of the History of Computing

cerns of ISO—the “standards elephant.”68 This inception, the IETF never had members, only
cultural conflict—which was, by 1992, over a participants, and hence it could not have a for-
decade old—made CLNP especially unpalatable mal voting structure. In the tradition of Cerf’s
to the IETF as a replacement for their favored ICCB discussions, IETF leaders encouraged new-
Internet Protocol. comers to contribute their expertise, and
At the July 1992 IETF meeting in approved proposals that enjoyed broad support
Cambridge, Massachusetts, irate IETF partici- within the group. IETF veterans place an accept-
pants protested to the Internet Society about able level of agreement at around 80 to 90 per-
what they perceived as a unilateral decision by cent: a level high enough to demonstrate strong
the IAB. Approximately 700 IETF participants support, but flexible enough to work in the
demanded that the newborn Internet Society absence of unanimity. In short, rough consensus
intervene and ensure that the IETF would was an apt description of this informal process
remain in control of the standards process.69 In in which a proposal must answer to criticisms,
the face of a massive “palace revolt,” the IAB, but need not be held up if supported by a vast
with Vint Cerf prominent among them, relent- majority of the group.72 To IETF participants,
ed. As Cerf addressed the IETF, he slowly this process was vastly superior to the bureau-
removed the layers of his signature three-piece cratic and political approach of ISO.73
suit, performing a striptease that revealed a T- As a complement to rough consensus, run-
shirt: “IP on Everything.” The T-shirt, accord- ning code means “multiple actual and interop-
ing to Cerf, was to reiterate a goal of the IAB: to erable implementations of a proposed standard
run IP on every underlying transmission medi- must exist and be demonstrated before the pro-
um.70 Like Cerf, David Clark turned his plenary posal can be advanced along the standards
presentation into a memorable occasion, one track.”74 Since most standards begin with a pro-
that would “rally the troops” and reaffirm the posal from an individual or group within a
values of the Internet community.71 working group—and not from the IAB or IETF
leadership—the party behind the proposal
‘We reject: kings, presidents, and must provide multiple working versions of the
voting’ proposal. This burden of proof on the proposed
Clark framed his talk in terms of architectur- standard facilitates the adoption of new IETF
al choices the IETF would have to make. After standards across the Internet’s diverse comput-
spending several minutes urging the audience ing platforms. Running code also evokes a major
to focus on network security and the basic difference between the IETF and ISO approach-
assumptions of the protocol architecture, Clark es: where the IETF protocols represented “the
considered how the IETF should “manage the result of intense implementation discussion
process of change and growth.”5 As he remind- and testing,” ISO committees developed a the-
ed the IETF audience of the vital importance of oretical model that was difficult to alter or
the values of the process by which they made implement fully.75 According to Lyman
standards, Clark punctuated his discussion with Chapin, a participant in both ISO and Internet
his summary of the IETF approach: “We reject: standardization,
kings, presidents, and voting. We believe in:
rough consensus and running code.”5 The IETF it didn’t take long to recognize the basic irony of
community responded with overwhelming OSI standards development: there we were,
approval. “Rough consensus and running code” solemnly anointing international standards for
was so popular that Marshall Rose, a vocal par- networking, and every time we needed to send
ticipant in the IETF “palace revolt,” created the electronic mail or exchange files, we were using
ultimate form of computer geek approval: T- the TCP/IP-based Internet!76
shirts with the phrase emblazoned across the
front. (On the occasion of the IETF’s 20th Or, as Internet pioneer Einar Stefferud was
anniversary in 2006, Clark delivered an encore fond of saying, “OSI is a beautiful dream, and
of this presentation. The video of his talk is TCP/IP is living it.”77
available from [Link] Beyond serving as a concise description of
“Rough consensus and running code” gen- the IETF’s organizational and technical
erated and sustained this level of enthusiasm approach, “rough consensus and running code”
because of the way it framed the successful also served as a means of self-identification and
aspects of the IETF process in opposition to the a positive summary of the IETF’s model for
ISO process. The rough-consensus component standards development. The internal divisions
of this motto refers to the decision-making exacerbated by the controversy over CLNP
process within IETF working groups. Since its prompted a good deal of reflection among

July–September 2006 55
 ‘Rough Consensus and Running Code’

those who were committed to defending the Internet standards community refer to the
traditions of the IETF. Although most of his “rough consensus and running code” ideal,
presentation was devoted to the pressing tech- Clark’s rejection of alternative forms of deci-
nical and organizational problems within the sion making—kings, presidents, and voting—
Internet standards community, Clark’s memo- reminds us of the close links between network
rable phrase was an attempt to unite the frac- standards and international politics.
tured community by contrasting it to their OSI Throughout the 1970s, 1980s, and 1990s, the
rivals. After their successful campaigns against open and decentralized technical architecture of
the IAB and the CLNP proposal, one can imag- the Internet embodied the technical and orga-
ine IETF engineers leaving the July 1992 meet- nizational values of its design group. Internet
ing with a certain sense of optimism about the architecture and organizations were created at
future of the Internet. the same time, by the same people, as part of an
effort “to unite the community behind a single
Conclusion objective—to focus the effort and guarantee the
By the time National Institute of Standards continued growth of the Internet.”80 As the evi-
and Technology (NIST) abandoned GOSIP in dence in this article demonstrates, defining a
favor of TCP/IP in 1994, the grand future single goal and then uniting the community
planned for OSI was on the rapid decline.78 The behind it was no small feat.
market for network protocols had tipped in These organizational and procedural prob-
favor of TCP/IP, epitomized by the popularity of lems intensified throughout the 1990s, and con-
a new application—the World Wide Web—that tinue to haunt the IETF and IAB—as well as the
was designed to take advantage of the Internet’s broader reaches of information technology stan-
end-to-end architecture.79 The veterans of the dards—today.81 The structure and process of
Internet-OSI standards war were no doubt wiser Internet standards set precedents and influenced
from the experience, but the millions of users the development of subsequent efforts to set
who got on the Internet in the mid-1990s were standards for digital networks in institutional
oblivious to the fact that their new toy was the imitators such as the World Wide Web
product of a protracted international struggle. Consortium, the Global Grid Forum, and, most
For those who read the last issue of Tele- recently, the Voting Systems Performance
communications Policy in 1993, William Drake Rating.82 The Internet standards community not
provided an insightful summary: only introduced technological innovations; it
also pioneered organizational innovations such
The debate is not merely about the comparative as the use of electronic mailing lists to build con-
efficacy of two sets of standards, but it is rather sensus around technical work, a process open to
between two competing visions of how interna- all interested stakeholders, and the free and
tional standardization processes and network unrestricted distribution of standards.
development should be organized and controlled.8 As it embodied a new style of standardiza-
tion, the Internet standards community con-
The religious, political, and military stantly dealt with problems that stemmed from
metaphors that participants used to describe the tension between centralized authority and
the competition between the Internet and OSI grassroots initiatives, as well as the rising influ-
confirm that this was no mere technical dis- ence of commercial values. Since 1992, IETF
pute. At the height of the “religious war” participants in multiple working groups and
between TCP/IP and OSI, tensions within the mailing lists have spent tremendous amounts
IETF and IAB over architecture and organiza- of energy defining formal procedures for the
tional power created a “constitutional crisis.” IETF, to the point that many in the IETF feel
This crisis, a divisive force in a community that that their technical work is suffering from a lack
had always prided itself on its attention to due of attention.83 In light of its history, it seems cer-
process and consensus, forced engineers in the tain that the IETF’s ongoing efforts to refine and
IETF and IAB to examine their core procedural reform its structure and process will dictate the
beliefs. In other words, strains in the technical future success of its standards. Organizational
architecture—the address space—prompted and political conflicts within standards bodies
strains in the organizational architecture. define the terrain within which effective col-
Forged in the face of this crisis, the credo laborations can take place. Standards wars such
“rough consensus and running code” articulat- as the Internet-OSI conflict provide ample
ed a political philosophy, a style of network research opportunities for historians who want
architecture and of engineering. While it is to understand computers, networks, and the
now common to see participants in the people who designed and used them.

56 IEEE Annals of the History of Computing

Acknowledgments ent. They were sort of the same genre. Just like,
I gratefully acknowledge the close reading and say, Chinese and Americans are of the same
thoughtful comments of Scott Bradner, Vinton genre except one speaks Chinese and one speaks
G. Cerf, Stuart W. Leslie, Mark Pittenger, and English, one lives on one side of the world, they
Philip J. Weiser. go to sleep during your daytime, etc.”
11. V.G. Cerf and R.E. Kahn, “A Protocol for Packet
References and notes Network Intercommunication,” IEEE Trans.
1. For descriptions of the features (and problems) of Comm., vol. COM-22, no. 5, 1974, pp. 637-648.
formal standards setting, see P.A. David and M. For an account of the decisive international con-
Shurmer, “Formal Standards-Setting for Global tributions to TCP/IP, especially from French com-
Telecommunications and Information Services,” puter scientists such as Louis Pouzin who had
Telecommunications Policy, Oct. 1996, pp. 789- developed the Cyclades network, see J. Abbate,
815; and W.J. Drake, “The Transformation of Inventing the Internet, pp. 123-133.
International Telecommunications Standardiza- 12. R. Kahn, “Testimony before the Subcommittee
tion: European and Global Dimensions,” Telecom- on Basic Research of the Committee on Science
munications in Transition: Policies, Services, and on the Subject of Internet Domain Names,” 31
Technologies in the European Community, C. Stein- Mar. 1998; available from [Link]
field, J.M. Bauer and L. Caby, eds., Sage Publica- [Link]/[Link].
tions, 1994, pp. 71-96. 13. D.D. Clark, “The Design Philosophy of the DARPA
2. The best single account of Internet history is J. Internet Protocols,” Proc. SIGCOMM 88, Comput-
Abbate, Inventing the Internet, MIT Press, 1999. er Communication Rev., vol. 18, no. 4, 1988, pp.
Cerf and Kahn were awarded the 2004 Turing 106-107.
Award for their pioneering efforts. 14. Cerf, Cohen, and Postel split the protocol into
3. According to one history, the Internet’s architec- TCP and IP to reduce the requirements on
tural principles “embody some value judgments network gateways and leave complex tasks such
and reflect the fundamental political and ethical as tracking reliable packet delivery to the
beliefs of the scientists and engineers who computers at the ends of the network. See J.
designed the Internet”; National Research Coun- Abbate, Inventing the Internet, pp. 129-130; V.G.
cil, The Internet’s Coming of Age, National Acade- Cerf, “Protocols for Interconnected Packet Net-
my Press, 2000, p. 35. works,” Computer Communication Rev., vol. 18,
4. See H. Nissenbaum, “How Computer Systems no. 4, 1980, pp. 10-11; and J. Postel, ed., “DOD
Embody Values,” Computer, Mar. 2001, pp. 118- Standard Internet Protocol,” IETF RFC 760, Jan.
120. 1980; [Link]
5. D.D. Clark, “A Cloudy Crystal Ball: Visions of the 15. J.H. Saltzer, D.P. Reed, and D.D. Clark, “End-To-
Future,” plenary presentation at 24th meeting of End Arguments in System Design,” ACM Trans.
the Internet Engineering Task Force, Cambridge, Computer Systems, vol. 2, no. 4, 1984, p. 277.
Mass., 13-17 July 1992. Slides from this presenta- 16. J.H. Saltzer, D.P. Reed, and D.D. Clark, “End-to-
tion are available from [Link] End Arguments,” p. 287.
videos/future_ietf_92.pdf. 17. D.S. Isenberg, “The Rise of the Stupid Network,”
6. L. Lessig, Code and Other Laws of Cyberspace, Computer Telephony, Aug. 1997, pp. 16-26.
Basic Books, 1999, p. 4. 18. M.S. Blumenthal and D.D. Clark, “Rethinking the
7. On “rough consensus and running code” as IETF Design of the Internet: The End to End
credo, see E. Huizer, IETF-ISOC Relationship, IETF Arguments vs. the Brave New World,” ACM
RFC 2031, Oct. 1996; [Link] Trans. Internet Technology, vol. 1, no. 1, 2001, pp.
[Link]; and S. Bradner, “The Internet Engi- 70-109.
neering Task Force,” Open Sources: Voices from the 19. Lessig notes that end-to-end concepts are not
Open Source Revolution, C. DiBona, S. Ockman, unique to the Internet, and are distinguished for
and M. Stone, eds., O’Reilly, 1999, p. 50. their ability to provide a broad range of services.
8. W. Drake, “The Internet Religious War,” Telecom- Two examples are the electricity grid and the
munications Policy, Dec. 1993, p. 643. highway systems: they are simple architectures
9. B. Carpenter, ed., Architectural Principles of the with minimal requirements for participation
Internet, IETF RFC 1958, June 1996; [Link] (standard plugs or motor vehicles) that do not
[Link]/rfc/[Link]. constrain the behavior of the participants. L.
10. R.E. Kahn, oral history interview by J.E. O’Neill, Lessig, The Future of Ideas: The Fate of the
24 Apr. 1990, OH 192, Charles Babbage Inst. Commons in a Connected World, Random House,
(CBI), Univ. of Minnesota, Minneapolis. Kahn 2001, pp. 26, 39.
continued, noting that “all the details were differ- 20. For a discussion of the strategy behind this transi-
ent. I don’t mean conceptually they were differ- tion, see J. Postel, “NCP/TCP Transition Plan,”

July–September 2006 57
 ‘Rough Consensus and Running Code’

IETF RFC 801, Nov. 1981; [Link] 1462, May 1993; [Link]
rfc/[Link]. [Link].
21. R.E. Kahn, “The Role of the Government in the 31. See B. Leiner et al., “A Brief History of the Inter-
Evolution of the Internet,” Comm. ACM, vol. 37, net”; [Link]
no. 8, 1994, p. 16. Kahn was director of ARPA’s shtml. Since 1989, the number of IETF areas has
Information Processing Techniques Office from shifted between seven and ten.
August 1979 until September 1985. Because Cerf 32. B. Leiner et al., “A Brief History of the Internet”;
left ARPA in October 1982 and Barry Leiner did and R. Kahn, “The Role of the Government,” p.
not replace him until August 1983, Kahn person- 17. For Internet Research Task Force mission and
ally managed the transition to TCP/IP. R.E. Kahn, activities, see [Link]
oral history interview, OH 192, CBI. 33. See S. Bradner, “The Internet Engineering Task
22. V.G. Cerf, oral history interview by J.E. O’Neill, 24 Force,” OnTheInternet, vol. 7, no. 1, 2001, p. 24;
Apr. 1990, OH 191, CBI. and J. Abbate, Inventing the Internet, pp. 207-
23. V. Cerf, “The Internet Activities Board,” IETF RFC 208.
1160, May 1990; [Link] 34. V. Cerf, “IETF and ISOC,” 18 July 1995;
[Link]. [Link]
24. R.E. Kahn oral history interview, OH 192, CBI. l. Kahn also worked to provide institutional sup-
The term “kitchen cabinet” dates from the port for the growth of the Internet when he
administration of President Andrew Jackson, who founded the Corporation for National Research
preferred to consult with an informal group of Initiatives in 1986. See [Link]
advisors—allegedly in the White House kitchen— about_cnri.html.
instead of his formal “Parlor” cabinet. See R.V. 35. B. Leiner et al., “A Brief History of the Internet.”
Remini, Andrew Jackson: The Course of American In a 1993 article, Dave Crocker concurred: “In
Freedom, 1822−1832, Johns Hopkins Univ. Press, general, the IETF is applying its own technical
1998, pp. 315-330. design philosophy to its own operation.” D.
25. V. Cerf, as told to B. Aboba, “How the Internet Crocker, “Making Standards the IETF Way,” Stan-
Came to Be,” The Online User’s Encyclopedia: Bul- dardView, vol. 1, no. 1, 1993, p. 54. The 1968
letin Boards and Beyond, B. Aboba, Addison-Wes- musings of Melvin Conway are also strikingly rel-
ley, 1993. Kahn and Cerf note that the ICCB, evant: “There is a very close relationship between
unlike the IAB and IETF, functioned in a time the structure of a system and the structure of the
where there was very little commercial interest in organization which designed it.” See M.E. Con-
the Internet, no personal computers, and a nas- way, “How Do Committees Invent?” Datamation,
cent networking industry. R.E. Kahn and V.G. vol. 14, no. 4, 1968, p. 30; and “Conway’s Law,”
Cerf, “What Is The Internet (And What Makes It at [Link]
Work),” Dec. 1999; [Link] [Link].
us/what_is_internet.html. 36. A.L. Norberg and J.E. O’Neill, Transforming Com-
26. R.E. Kahn, “The Role of the Government,” p. 16. puter Technology, Johns Hopkins Univ. Press,
27. The acronym “IAB” remained consistent since 1996; and T.P. Hughes, Rescuing Prometheus, Pan-
1984, but the “A”—and the meanings behind theon Books, 1998, pp. 15-68.
it—have changed. From 1984 to 1986, the IAB 37. L. Press, “Seeding Networks: The Federal Role,”
was the Internet Advisory Board; in 1986 its Comm. ACM, vol. 39, no. 10, 1996, pp. 11-18.
name changed to the Internet Activities Board; in 38. The description of the ICCB as a “grass-roots
1992 it changed once again, this time to the mechanism” comes from Kahn, “The Role of the
Internet Architecture Board. See “A Brief History Government,” p. 18.
of the Internet Advisory/Activities/Architecture 39. Various aspects of OSI adoption—too numerous
Board”; [Link] and complex to describe in full in this article—
28. Clark was Internet Architect from 1983 to 1989; permeate Internet history. Although we lack a
Cerf served from 1989 to 1992, and was followed comprehensive historical analysis of the Internet,
by Lyman Chapin (through March 1993). OSI, and other computer internetworking archi-
29. See B. Leiner et al., “A Brief History of the Inter- tectures of the period, three excellent
net”; [Link] comparisons may be found in J. Abbate, Inventing
shtml, and R.E. Kahn, oral history interview, OH the Internet, pp. 147-169; National Research
192, CBI. Task forces or working groups have Council, Global Networks and Local Values: A Com-
been a consistent feature in the history of techni- parative Look at Germany and the United States,
cal standards bodies. See C. Cargill, Open Systems National Academy Press, 2001, pp. 23-45; and T.
Standardization: A Business Approach, Paladin Egyedi, “‘Tension between Standardisation and
Consulting, 1997. Flexibility’ Revisited: A Critique,” Standardisation
30. [Link], “FYI on ‘What is the Internet?’” IETF RFC and Innovation in Information Technology: Conf.

58 IEEE Annals of the History of Computing

 Proc. 1st IEEE Conf. Standardisation and Innovation 46. For an analysis of ISO’s position within the system
in Information Technology, K. Jakobs and R. of European standardization bodies, see S.K.
Williams, eds., IEEE Press, 1999, pp. 65-74. Schmidt and R. Werle, Coordinating Technology:
40. J. Abbate, “Government, Business, and the Mak- Studies in the International Standardization of
ing of the Internet,” Business History Rev., vol. 75, Telecommunications, MIT Press, 1998, pp. 39-57.
no. 1, Spring 2001, p. 167. See also M. Libicki, 47. M. Witt, “Moving from DoD to OSI Protocols: A
Information Technology Standards: Quest for the First Step,” Computer Communication Rev., vol.
Common Byte, Digital Press, 1995, pp. 75-129; 16, no. 2, 1986, pp. 2-7.
J.S. Quarterman and S. Wilhelm, UNIX, POSIX, 48. National Research Council, Transport Protocols for
and Open Systems: The Open Standards Puzzle, Department of Defense Data Networks: Report to
Addison-Wesley, 1993; C. Cargill, “Evolution and the Department of Defense and the National
Revolution in Open Systems,” StandardView, vol. Bureau of Standards Committee on Computer-Com-
2, no. 1, 1994, pp. 3-13; and S. Schindler, “Open puter Communication Protocols, Board on
Systems, Today and Tomorrow—A Personal Per- Telecommunications and Computer Applications
spective,” Computer Networks, vol. 5, no. 3, Commission on Eng. and Technical Systems,
1981, pp. 167-176. National Academy Press, 1985.
41. R.N. Langlois, “Networks and Innovation in a 49. J. Postel, “A DoD Statement on the NRC Report,”
Modular System: Lessons from the Microcomput- IETF RFC 945, May 1985; [Link]
er and Stereo Component Industries,” Research rfc/[Link].
Policy, vol. 21, no. 4, 1992, pp. 297-313. See also 50. “U.S. Government Open Systems
R. Sanchez and J.T. Mahoney, “Modularity, Flexi- Interconnection Profile,” US Federal Information
bility, and Knowledge Management in Product Processing Standards Publication 146, version 1,
and Organization Design,” Strategic Management Aug. 1988, cited in V. Cerf and K. Mills, “Explain-
J., vol. 17, Winter special issue, 1996, pp. 63-76. ing the Role of GOSIP,” IETF RFC 1169, Aug.
42. See P.E. Green Jr., ed., Network Interconnection 1990; [Link] See
and Protocol Conversion, IEEE Press, 1988; and R.J. also P. Janson et al., “Architectural Directions for
Cypser, Communicating for Cooperating Systems: Opening IBM Networks,” p. 314 (“Many govern-
OSI, SNA, and TCP/IP, Addison-Wesley, 1991. IBM ment agencies around the world, including the
executives and engineers were keenly aware of U.S. Department of Defense, require OSI on all
the importance of network migration and systems they purchase”).
integration. See, for example, L.M. Branscomb, 51. Throughout the latter half of the 20th century,
“Computer Communications in the Eighties— DoD procurement policies were effective means
Time to Put It All Together,” Computer Networks, for driving market growth in semiconductors,
vol. 5, no. 1, 1981, pp. 3-8; P. Janson, R. Molva, computers, and software. See D.M. Hart, “Cor-
and S. Zatti, “Architectural Directions for Open- porate Technological Capabilities and the State:
ing IBM Networks: The Case of OSI,” IBM Systems A Dynamic Historical Interaction,” Constructing
J., vol. 31, no. 2, 1992, pp. 313-335; and T.J. Corporate America: Historical Perspectives on Big
Routt, “Integration Strategies for APPN and Business, Society, and Politics, K. Lipartito and D.B.
TCP/IP,” Business Communications Rev., Mar. Sicilia, eds., Oxford Univ. Press, 2004, pp. 168-
1995, pp. 43-49. 187.
43. H. Zimmerman, “OSI Reference Model—The ISO 52. TCP/IP and OSI, despite the rhetoric of many of
Model of Open Architecture for Open Systems their proponents, could be engineered (with sub-
Interconnection,” IEEE Trans. Comm., vol. COM- stantial effort) to work together. Since 1987,
28, no. 4, 1980, p. 425. ISO, created in 1947, is a according to Cerf and NIST’s Kevin Mills, there
worldwide federation of national standards bodies. had been efforts “within the Internet community
44. R. des Jardins, “Overview and Status of the ISO to enable integration of OSI, coexistence of OSI
Reference Model of Open Systems Interconnec- with TCP/IP, and interoperability between OSI and
tion,” Computer Networks, vol. 5, no. 2, 1981, TCP/IP”; V. Cerf and K. Mills, IETF RFC 1169. See
pp. 77-80. also E. Huizer, “The IETF Integrates OSI Related
45. See, for example, T. Whitty, “OSI: the UK Work,” ConneXions, vol. 7, no. 6, 1993, pp. 26-28.
approach,” Comm., vol. 7, no. 2, 1990, pp. 20- 53. R. des Jardins, “OSI is (Still) a Good Idea,”
24; L. Caffrey, “EPHOS: Towards a European ConneXions, vol. 6, no. 6, 1992, p. 33. Des
GOSIP,” Computer Networks and ISDN Systems, Jardins added, “Let’s continue to get the people
vol. 19, no. 3-5, 1990, pp. 265-269; and R. of good will from both communities to work
Cowan, ed., Information Technology Standards: together to find the best solutions, whether they
The Economic Dimension, Organisation for are two-letter words or three-letter words, and
Economic Co-operation and Development, 1991, let’s just line up the bigots against a wall and
pp. 23-30. shoot them.”

July–September 2006 59
 ‘Rough Consensus and Running Code’

54. D. Cohen and J. Postel, “The ISO Reference [Link]

Model and Other Protocol Architectures,” Infor- 65. C. Cargill, Open Systems Standardization, p. 257;
mation Processing 83: Proc. IFIP 9th World Comput- S. Bradner, “The Internet Engineering Task
er Congress, R.E.A. Mason, ed., North-Holland, Force,” OnTheInternet, p. 24; V. Cerf, “IETF and
1983, p. 34. ISOC”; [Link]
55. D. Cohen and J. Postel, “The ISO Reference [Link].
Model,” p. 30. 66. S. Bradner, “IETF,” OnTheInternet, p. 24.
56. C. Malamud, Exploring the Internet: A Technical 67. Chapin remarked in May 1992, “The most com-
Travelogue, Prentice Hall PTR, 1992, p. 191. prehensive solution [to the shortage of Internet
57. D.A. Mills, personal interview with author, 26 addresses] is to replace the Internet Protocol in
Feb. 2004. the Internet with the Open Systems Interconnec-
58. M.A. Padlipsky, The Elements of Networking Style, tion Connectionless Network Protocol. That idea
and Other Animadversions on the Art of Intercom- is already almost universally accepted.” Clearly
puter Networking, iUniverse, 2000, 1st ed., Pren- Chapin was mistaken in his assessment of com-
tice Hall, 1985, p. 11. In his characteristically munity support for CLNP. E. Messmer, “Internet
witty prose, Padlipsky recommended that the Architect,” p. 46.
ISO Reference Model, or ISORM, be pronounced 68. ISO’s designation as a “standards elephant”
“Eyesore-mmm.” comes from David Clark’s presentation to the July
59. C. Malamud, Exploring the Internet: A Technical 1992 IETF plenary, “A Cloudy Crystal Ball: Visions
Travelogue, p. 196. of the Future” (see ref. 5).
60. C. Malamud, Exploring the Internet: A Technical 69. C. Cargill, Open Systems Standardization, p. 257.
Travelogue, p. 151. 70. V.G. Cerf, personal communication, 27 Jan.
61. Published in December 1991, IETF RFC 1287 2002. A photograph of Cerf wearing his shirt is
noted “increasing strains on the fundamental available at [Link]
architecture, mostly stemming from continued [Link]. Beginning in 1994, the IETF
Internet growth.” D. Clark et al., “Towards the developed a new version of the Internet Protocol,
Future Internet Architecture,” IETF RFC 1287, known as IPv6, to succeed IPv4; but for a variety
Dec. 1991; [Link] of reasons beyond the scope of this article, a uni-
Lyman Chapin, IAB chairman in 1992, said in versal transition to IPv6 has not yet occurred.
May 1992 that the shortage of Internet addresses Interested readers might visit [Link]
was “definitely the most significant engineering [Link]/pub/ipng/html/.
problem on the Internet now.” E. Messmer, 71. S. Bradner, “IETF,” On The Internet, p. 24.
“Internet Architect Gives Long-Term View,” Net- 72. S. Bradner, “IETF,” Open Sources, pp. 47-53. See
work World, vol. 9, 18 May 1992, p. 37. also D. Crocker, “Making Standards the IETF
62. For one proposal to incorporate CLNP, see R. Cal- Way,” pp. 48-54; and S. Harris, “The Tao of the
lon, “TCP and UDP with Bigger Addresses IETF—A Novice’s Guide to the Internet Engineer-
(TUBA), A Simple Proposal for Internet Address- ing Task Force,” IETF RFC 3160, Aug. 2001;
ing and Routing,” IETF RFC 1347, June 1992; [Link]
[Link] 73. The informal name of the document series in
63. In addition to this technical rationale, Clark sug- which Internet Standards are published—
gested that ISO and the American National Stan- “Requests for Comments”—underlines the inten-
dards Institute (ANSI) were pressuring the IAB to tion for Internet standards to be descriptive
implement ISO as the Internet host protocol. documents, not final and unchangeable prescrip-
D.D. Clark, personal communication, 27 Oct. tions. IETF standards are published as RFCs, but
2001. not all RFCs are standards. See RFC Editor et al.,
64. C. Huitema, IPv6: The New Internet Protocol, Pren- “30 Years of RFCs,” IETF RFC 2555, 7 Apr. 1999;
tice Hall PTR, 1998, p. 2. Minutes of the IAB dis- [Link] (“Hiding in
cussion are available from the history of the RFCs is the history of human
[Link] institutions for achieving cooperative work”).
.1992–06–[Link]. According to “The Internet Among the humorous and offbeat RFCs, two
Standards Process,” written by L. Chapin and classics are R. Callon, ed., “The Twelve Network-
published as IETF RFC 1310 in March 1992, the ing Truths,” IETF RFC 1925, 1 Apr. 1996;
IAB (still at this time the Internet Activities Board) [Link] and D. Wait-
delegated to the IETF “the primary responsibility zman, “A Standard for the Transmission of IP
for the development and review of potential Datagrams on Avian Carriers,” IETF RFC 1149, 1
Internet Standards from all sources.” Internet Apr. 1990; [Link]
Activities Board, L. Chapin, chair, “The Internet 74. S. Bradner, “IETF,” OnTheInternet, p. 26.
Standards Process,” IETF RFC 1310, Mar. 1992; Bradner’s summary is telling: “In brief, the IETF

60 IEEE Annals of the History of Computing

 operates in a bottom-up task creation mode and solutions/2004-February/[Link]. For a
believes in ‘fly before you buy.’” Bradner, “IETF,” summary of the IETF’s recent problems, see E.
Open Sources, p. 51. Bradner is a longtime partici- Davies, ed., “IETF Problem Statement,” IETF RFC
pant of the IETF who served as an area director 3774, May 2004; [Link]
and IESG member between 1993 and 2003. [Link].
75. M.A. Padlipsky, Elements of Networking Style, p.
104. Andrew L. Russell is a PhD can-
76. Lyman Chapin, quoted in G. Malkin, “Who’s didate in the Department of the
Who in the Internet: Biographies of IAB, IESG, History of Science and Tech-
and IRSG Members,” IETF RFC 1336, May 1992; nology at Johns Hopkins Univer-
[Link] sity. He holds a BA in history
77. Einar Stefferud, quoted in M.T. Rose, “Comments from Vassar College and an MA
on ‘Opinion: OSI Is (Still) a Good Idea,’” ConneX- in history from the University of
ions, vol. 6, no. 8, 1992, pp. 20-21. Colorado at Boulder. He also
78. See M. Libicki, Information Technology Standards, worked at the Harvard Information Infrastructure
pp. 108-119; D.C. Wood, “Federal Networking Project (1997−1999).
Standards: Policy Issues,” StandardView, vol. 2,
no. 2, 1994, pp. 218-223; J.S. Quarterman, “The Readers may contact Andrew Russell about this
Demise of GOSIP,” Matrix News, vol. 4, no. 10, article at Dept. History of Science and Technology,
1994, p. 6; and R. Hunt, “The Future of TCP/IP The Johns Hopkins University, 3505 N. Charles St.,
and OSI/GOSIP—Migration or Coexistence?” Baltimore, MD 21218; arussell@[Link].
Proc. Networks 93 Conference: Held in Birmingham,
June-July 1993, Blenheim Online, 1993, pp. 423- For further information on this or any other com-
437. puting topic, please visit our Digital Library at
79. T. Berners-Lee, Weaving the Web: The Original [Link]
Design and Ultimate Destiny of the World Wide
Web by its Inventor, HarperSanFrancisco, 1999, p.
16.
80. C. Huitema, IPv6: The New Internet Protocol, p. 1.
Huitema was on the IAB from 1991 to 1996 and
was Internet Architect from March 1993 to July
1995.
81. For a lengthy and enthusiastic appraisal of the
IETF as “an international phenomenon that con-
forms well to the requirements of [Jürgen] Haber-
mas’s discourse ethics,” see A.M. Froomkin,
“Habermas@[Link]: Toward a Critical
Theory of Cyberspace,” Harvard Law Rev., vol. 16,
no. 3, 2003, pp. 749-873. For a more critical
appraisal of the problems of informal standardi-
zation bodies (including the IETF), see P.J. Weiser,
“The Internet, Innovation, and Intellectual Prop-
erty Policy,” Columbia Law Rev., vol. 103, no. 3,
2003, pp. 573-576.
82. “Rough consensus and running code” provided a
model for standards development for the World
Wide Web. See T. Berners-Lee, Weaving the Web,
p. 98. See also “Global Grid Forum Overview:
Structure and Process”; [Link]
org/L_About/Struc_Proc.htm; and “Voting Sys-
tem Performance Rating Charter”; [Link]
org/charter (“The underlying approach and
structure of VSPR is modeled after the IETF”).
83. As one participant stated recently, the “process-
process” is “sucking resources out of IETF’s tech-
nical/standards work.” S. Dawkins, “Two New
Proposals/Drafts—Details,” 10 Feb. 2004; avail-
able from [Link]

July–September 2006 61