|< Day Day Up >|| |
When the wave of growth hit, the ARPANET elite-Kahn, Cerf, Postel, Crocker, Clark, and a handful of other colleagues-had been working together on networking continuously for about 15 years. The group formed a tightly knit cadre with strong backing from U.S. government research agencies. As the infrastructure and user base expanded, however, their status as researchers began to blur into a new and very different role as the managers of a new international standards organization. Rising to meet this challenge, they succeeded in forming a robust and unusually open protocol development community. As the importance and size of the Internet increased, they constructed more formal organizational structures around themselves to maintain their position as stewards of the Net. This process culminated in the formation of the Internet Society in 1992. It can be interpreted as an attempt by the Internet cadre to institutionalize their community. The latent contradictions in the process-most notably the ambiguous relationship to the U.S. government-set the stage for the Internet governance struggles of the late 1990s.
The first step in the formation of a governance hierarchy was the creation of an Internet Activities Board (IAB) in late 1983, the precursor of today's Internet Architecture Board. The IAB replaced a standing advisory committee for DARPA's Internet program that had been around since 1979, back when Kahn directed the Information Processing Techniques Office and Cerf was working for him as the Internet program officer. [36 ]Cerf left DARPA in 1982 to join MCI Telecommunications. The initiative for a new arrangement came from Cerf's replacement, Barry Leiner, and MIT's David Clark, the chair of the earlier committee. The new board was set up as a ten-person panel, with each member supervising a task force devoted to different technical aspects of internetworking. Vint Cerf was designated the first IAB chair, and remained so for the next eight crucial years of Internet expansion. As chair, he decided who else could join the board. The members in turn elected a chair every two years.
The new IAB was just another DARPA committee, a self-selecting group of the original Internet people with no legal identity. According to one contemporary, Jon Postel was the 'defacto Internet standards process' and 'IAB served as his reviewing team.' [37 ]
Things changed in 1986-1987, when the National Science Foundation became involved in funding the Internet backbone. The stimulus to growth led to increased scale and more complex engineering problems; Postel and the IAB alone could not keep up with the growth. In response to the pressing need for near-term Internet technical standards development, one of the original task forces, known as 'Internet Architecture,' evolved into the Internet Engineering Task Force (IETF). Unlike the other task forces, which were limited to invited members, IETF began to hold public meetings four times a year starting in January 1986. At its fifth meeting at the beginning of 1987, 50 people showed up and spontaneously organized themselves into working groups. The channel into Internet activity created by IETF was entered by a growing number of computer scientists and engineers from the public and private sectors. Attendance exceeded 200 by July 1989; in 1992 more than 650 participants attended its summer meeting. The IETF had taken on a life of its own.
Where before engineers had spoken of a 'DDN community' [38 ]or 'ARPA community,' it was now 'the Internet community' or just 'the community.' IETF meetings and email lists were its social center. The IETF developed its own culture, a technical meritocracy where an informal dress code reigned and working groups could assemble or disband quickly and with minimal bureaucracy. In marked contrast to traditional standards organizations, participants were considered to be individuals and not representatives or delegates of organizations. The emerging community remained nonincorporated and mostly virtual. 'There is no membership in the IETF. Anyone may register for and attend any meeting. The closest thing there is to being an IETF member is being on the IETF mailing lists' (RFC 1391, January 1993).
Unlike standards communities grounded in coalitions of vendors or carriers, the early IETF considered interoperability and empowerment of the end user to be basic norms. The standards themselves were nonproprietary. All documentation was open, noncopyrighted, and freely available online. The community 'believes that the value of technical ideas should not be decided by vote but by empirical proof of feasibility or, in the language of the engineers, by running code' (Hofmann 1998). The community's political modus operandi was reflected in its famous credo, coined by David Clark in 1992: 'We reject presidents, kings and voting; we believe in rough consensus and running code.' IETF participants liked to draw unfavorable comparisons between themselves and traditional standards organizations, especially its global competitor, the OSI community.
'In the IETF world we produce running code that has documents that describe it. A lot of other standards organizations produce documents and that's the end of it.' [39 ]
Within this emergent community, the DARPA/ISI veterans stood at the top of the informal pecking order and assumed the role of village elders. There was, in fact, a latent tension between the new participants pouring into the IETF meetings, who thought of themselves as self-governing, and the residual DARPA hierarchy, which thought of the IETF as a 'subsidiary organization' under the control of the IAB. But this tension remained in the background until 1992. The relationship at this stage was symbiotic: the old guard provided a basic structure and process within which the others could work productively.
By the fourteenth IETF meeting in July 1989, the IAB/IETF reorganized itself to assume the basic structure that it still retains (although later some important changes were made in how the occupants of leadership positions are selected). The number of task forces was trimmed to two: Internet Engineering and Internet Research (IRTF). Simultaneous with this reorganization, a flurry of new RFCs issued forth from Cerf and Postel documenting the procedures and functional relationships among the elements of the Internet technical community. [40 ]The documents portray a hierarchical authority structure with the IAB on top, and below it the steering groups, area directors, and working groups of the IETF and IRTF. The notion of an 'official IAB protocol standard' was first promulgated at this time (RFC 1083, December 1988).
Robert Kahn left DARPA and, together with Cerf and ISI's Keith Uncapher, formed in 1986 a new nonprofit organization, the Corporation for National Research Initiatives (CNRI) to 'foster research and development for the National Information Infrastructure.' Both the IAB and the IETF received funding support from the U.S. government. The IETF was supported by means of a cooperative agreement between the National Science Foundation and CNRI. As more federal government agencies were drawn into the TCP/IP Internet, a Federal Research Internet Coordinating Committee (FRICC) was created, in fall 1987. The committee's founding agencies were DARPA, NSF, NASA, the Department of Energy, and the Department of Health and Human Services. FRICC was described by Cerf as the 'sponsor' of U.S. Internet research and the 'source of support for IAB and its subsidiary organizations' (RFC 1160; see also GAO 1989).
RFC 1083 (December 1988), which defined a standards-making process for the new, extended Internet community, was also the first public document to mention an Internet Assigned Numbers Authority (IANA). IANA was said to be located at ISI, and Postel was listed as the contact. Under a longstanding contract between DARPA and ISI, IP address and protocol parameter assignment functions were listed as work items along with several other functions. The contract had been renewed in 1988, which may help to explain the sudden appearance of the IANA label. The contract itself did not use the label, however.
At any rate, the new RFCs made no mention of the DARPA contract. They claimed that IANA's authority was derived from the Internet Activities Board, [41 ]which was said to have given the IANA 'policy-setting authority' over assignment functions. [42 ]The new documents further claimed that Postel, acting as 'the IANA,' had 'delegated' the administrative aspects of the assignment function to SRI's DDN-NIC. [43 ]A new world was being defined by the RFCs. In that world, the IAB and Postel's assignment function, both established by DARPA, took on an independent existence. Cerf himself described the IAB at this time as 'an unincorporated, volunteer organization, with support to participating individuals/organizations from the U.S. government or from the individual's own organization.' [44 ]
These descriptions reflected the technical community's growing conception of itself as an autonomous, self-governing social complex. Explicit claims on the right to manage name and address assignment were being made by an authority structure that existed solely in Internet RFCs and lacked any basis in formal law or state action. The authority claims nevertheless had significant legitimacy within the technical community. Not only was Postel known, respected, and trusted within the IETF and the supporting government agencies but the RFC series was, for both old and new participants in IETF, the way reality was defined on the Internet. One former NSF official described the situation as an 'enlightened monarchy in which the federal government funded the best brains. Their output was RFCs, which were approved through a collegial, though sometimes brutal process of someone advancing an idea and everyone beating on it until the group consensus was that it would work. These RFCs became the ‘law' of the Internet-‘law' in the sense of operational practice, not legal jurisdiction. The federal ‘managers' were personally involved and ‘tracked' the activities as participants and partners. This participatory management model was so uniform and effective that many who were involved in the activity assert (even today) that the Internet was a ‘working anarchy.'' [45 ]Mitchell and most other participants in this process emphasize the atmosphere of trust, collegiality, and cooperation that existed at the time. These halcyon days of the IETF were, of course, grounded in unique and irreproducible conditions. Once the original trust and collegiality were shattered, as they were in 1996, the whole model of a 'bottom-up, consensus-based' assignment authority became a travesty.
Beginning in 1991 the ARPA cadre did attempt to place an organizational and legal capstone around their efforts. They founded a private, nonprofit organization called the Internet Society, which some contemporaries say was inspired by the National Geographic Society (Comer 1995, 11). That step toward formalization immediately engendered conflicts over its authority and methods, however.
The initial impetus for the formation of the Internet Society was the need to protect IETF area directors and working group chairs against lawsuits. Noel Chiappa, whose design of the Proteon router had made him a millionaire, was one of the first members of the Internet Engineering Steering Group (IESG). Concerned about his potential liability, he asked his lawyer about the risk this position entailed and was told that IETF's unincorporated status made him personally liable for standards decisions. That uncomfortable fact was relayed to the IESG and to Cerf in spring 1990. The importance of IETF had grown to the point where its decisions could have economic consequences. Already, participants who felt they had been treated unfairly by the IAB hierarchy were threatening to take legal action. [47 ]An Internet Society could provide liability insurance to responsible parties in the IETF.
Funding seems to have been another consideration. As TCP/IP internetworking was now a well-developed technology, DARPA was winding down its support. NSF also had a well-defined policy of fostering selfsupporting projects, and could not be counted on to fund Internet administration indefinitely. As early as November 1990, Cerf wrote to a colleague about his idea for 'The Internet Society, which might be a way of funding the operation of the IAB/IETF/IRTF.' [46 ]
A corporate identity also met the community's need for legitimacy in the international environment. Data networking and the telecommunication industry were beginning to converge. That brought with it a need for more extensive liaisons between the Internet world and the established international telecommunication standards organizations. Anthony M. Rutkowski, an American adviser to ITU Secretary-General Pekka Tarjanne, wanted to bring the Internet activity into the international standards community; at the same time, he saw the more open and flexible standardization processes pioneered by the Internet community as a model that the older institutions should imitate. In the traditional telecom world, as he was learning in Geneva, the idea of putting standards documents online and making them freely accessible was still a very radical one (Malamud 1992). Rutkowski proposed to get the Internet community recognized by international telecommunication standards organizations as a 'major community of interest and a significant standards making forum with which liaison is required.' [48 ]Cerf invited Rutkowski to join the IAB as its 'international person' in October 1990.
The Internet Society (ISOC) was formed in January 1992. The initial board of trustees included Bob Kahn, Vint Cerf, Mike Roberts of Educom, Charles Brownstein of the National Science Foundation, Lawrence Landweber of the University of Wisconsin, Lyman Chapin, Geoff Huston, Frode Griesen, and Juergen Harms. All but three were Americans. Rutkowski was made executive director. Chapin was chair of the IAB at the time, and Cerf was still an IAB member. In June 1992, at a meeting in Kobe, Japan, the newly constituted Internet Society board prepared a draft charter for an Internet Architecture Board (IAB) that brought 'the activities of ISOC and the Internet Activity Board into a common organization' (Cerf 1995). It can be seen as an attempt to self-privatize Internet governance in a way that finessed the issue of whether approval or any other action from the U.S. government was needed.
As soon as it was formed, the Internet Society exposed the chasm between the mindset of the IAB hierarchy that had created it and the rank and file of the IETF. As Brownstein mused later, 'It proved to be difficult to convince the IETF itself that ISOC was its legal capstone.' [49 ]In June 1992, only six months after it was founded, the IAB precipitated an outright revolt among IETF participants by announcing that an OSI technology, Connectionless Network Protocol (CLNP), would form the new standard for the Internet's routing and addressing in the future.
The Kobe incident was directly related to the pressures created by the Internet's growth. Classless interdomain routing (CIDR) had not been implemented yet, and the addition of large numbers of new networks to the Net threatened to deplete the IP address space soon. More worrisome, the rise of private, commercial Internet service providers (ISPs) alongside the single NSF backbone made routing much more complicated. With multiple, competing backbones and many new ISPs, as well as continual growth of other networks, routing tables were becoming too large for existing routers to handle.
The IAB chose CLNP because it would have provided a quick fix to the addressing and routing problems. It did so, however, in direct violation of established IETF bottom-up decision-making conventions. The IAB had discarded a recommendation of the IETF's steering group to allow further study and experimentation of the problem for six months. 'The problems of too few IP addresses and too many Internet routes are real and immediate, and represent a clear and present danger to the future successful growth of the worldwide Internet,' the IAB chair asserted. 'The normal IETF process of let a thousands flowers bloom, in which ‘the right choice' emerges gradually and naturally from a dialectic of deployment and experimentation, would in this case expose the community to too great a risk that the Internet will drown in its own explosive success before the process had run its course' (Lyman Chapin, July 1, 1992, cited in Hofmann 1998, 15).
The decision sparked a firestorm of protest that forced a full retraction at the next IETF meeting. [50 ]The controversy forced the IETF to confront fundamental questions about who made decisions within the Internet community and how the decision makers were selected. A new working group led by Steve Crocker was formed, the Process for Organization of Internet Standards, which became known as the POISED working group (RFC 1396). POISED redefined the nomination procedures for appointments to the IAB and the IETF's steering groups. Self-selection by the old Internet elite was no longer acceptable: 'There was a strong feeling in the community that the IAB and IESG members should be selected with the consensus of the community' (RFC 1396, 3). The recommendations stopped short, however, of advocating formal elections of leaders. The most important reason the IETF didn't institute voting was that Jon Postel and several other senior figures vowed that they would refuse to run for office in any electoral system. The technical cadre's allergy to democratic methods and public accountability ran deep and would later play a significant role in the battles over the structure of ICANN. By many accounts, that fatal misstep of 1992 discredited the IAB for several years, tarnishing the Internet Society as well. It took until 1996 for ISOC and IAB to regain enough authority within the broader Internet technical community to be in a position to assert leadership.
ISOC also was hampered by incompatible notions about its mission and methods within the board. Rutkowski wanted an industry-based standardization organization that followed the telecommunication industry model in membership if not in procedure. He was particularly interested in forging stronger links to the regional address registries in Asia-Pacific and Europe. Cerf, Kahn, and Landweber, on the other hand, wanted a professional organization to promote the Internet with an emphasis on individual membership. Their model proposed to raise money through conferences, workshops, and fund-raising among 'industry and other institutional sources' (Cerf 1995). These conflicts had not been resolved by 1995, and sharp personal differences began to develop between Rutkowski and other board members. Rutkowski was forced to leave. Lacking a clear purpose and method, the Society was not very successful at raising money. So while the IETF continued to grow in size and prestige, its meetings attracting over 2,000 participants, the Internet Society did not yet succeed in becoming its corporate embodiment (Werle and Lieb 2000). Figures 5.1 and 5.2 summarize some of the organizational relationships around 1993.
Figure 5.1: Internet governance circa 1993, private sector perspective
Figure 5.2: Internet governance circa 1993, U.S. government perspective
[36 ]Interview with John Klensin, November 13, 2000.
[37 ]The committee was known as the Internet Configuration Control Board (RFC 1160).
[38 ]Craig Partridge, email communication with author, July 2001.
[39 ]Steve Crocker, cited in RFC 1000 (August 1987).
[40 ]Cited in Hofmann (1998, 14).
[41 ]RFC 1083 (December 1988); RFC 1111 (August 1989); RFC 1120 ( September 1989); RFC 1160 (May 1990); RFC 1200 (April 1991); RFC 1250 (August 1991).
[42 ]An 'Internet FAQ' written by ISI staff members claimed, 'The task of coordinating the assignment of values to the parameters of protocols is delegated by the Internet Activities Board (IAB) to the Internet Assigned Numbers Authority (IANA).' This document seems to have evolved into RFC 1207 (February 1991).
[43 ]Interestingly, though, Cerf's major RFC documenting the IAB does not mention name and address assignment as one of its key functions. RFC 1160 defines the IAB functions as '(1) Sets Internet Standards; (2) manages the RFC publication process; (3) reviews the operation of the IETF and IRTF; (4) performs strategic planning for the Internet, identifying long-range problems and opportunities; (5) acts as a technical policy liaison and representative for the Internet community; and (6) resolves technical issues which cannot be treated within the IETF or IRTF frameworks.'
[44 ]Jon Postel to msggroup, November 15, 1985. See also email, Cerf to Aiken, March 17, 1995; RFC 1174 (August 1990); RFC 1207 (February 1991). The actual RFCs (1032 and 1020) announcing the transfer in 1987 were not written by Postel and do not mention IANA or any delegation from Postel or the IAB.
[45 ]Email, Cerf to Rutkowski, November 6, 1990, <">http://www.wia.org/ISOC/901106.htm>.
[47 ]Interview with Don Mitchell, December 19, 2000.
[46 ]Email, Cerf to Rutkowski, November 6, 1990, <http://www.wia.org/ISOC/901106.htm>.
[48 ]The case of Daniel Bernstein, who contended that his RFC submission had not been handled according to the IETF's documented processes, was another key event in instilling the fear of lawsuits among the IAB/IETF hierarchy. See <http:// ittf.vlsm.org/ietf/16.txt>.
[49 ]Email, Rutkowski to Cerf, April 5, 1991, <http://www.wia.org/ISOC/910405.htm>.
[50 ]Interview with Charles Brownstein, May 18, 2000.
|< Day Day Up >|| |