H. It works...

The Internet Protocol is sought, and found! — detail about the distinction between paper standards and implementation standards is given — standards processes of ISO and ITU compared to those of ISOC — structure of ISOC and IETF explored — emergence of W3C wondered about — the politics of standards making is discussed and critiqued — TCP/IP is compared to OSI.

1. But wait, I have a stupid question.
   Who, exactly, has promised us this new dawn of simplicity? And how will it happen? How can we establish that this difference between the standards of the past— proprietary and bureacratic— and the standards of the future— open and stupid— how do we establish that this difference is not simply one difference among others? How do we establish that it is not simply a choice that references some stability that reigns unseen over it, whether represented by culture or by science? Telecommunications, in its history, is essential to this change. Telecommunications as the apotheosis of what Bernard Siegert calls "the epoch of the postal system," i.e. the epoch of communication as control. We have already seen the light of the stupid network shine progress on the darkness of the telecommunications industry. Recall that for stupid to happen, three things were necessary: abundant infrastructure, underspecification, and IP. The first has an obvious history entwined in national development projects— from trans-atlantic cabling to the wholesale the wiring of the planet to the grid of telecom satellites circling the earth, it is emblem of modernity and control at a distance. The second is what allows software to migrate from the center to the edges of the network, leaving an empty, homogenous space, imagined only as a realm of flashes of light and waves of voltage, smart bits in a stupid river. The third gives the lie to the second: IP the Internet Protocol is a specific protocol that describes the manner in which such bits should discipline themselves, package and label themselves for their journey on any given medium. It is not the lowest level protocol, but it is a kind of de facto lingua franca. But my stupid question still stands: Where did this "universal standard" come from?

2. Lets answer by starting with standards as conceived of by Schmidt and Werle in Coordinating Technology [Schmidt98]:

"Standards have only recently begun to..."(3)

3. Schmidt and Werle's book looks at attempts at explicit international stanardization. Of its three case studies, only one—the facsimile (fax) standard— is familiar and successful. The other two can be read as counterparts to internet tecehnologies that have since come to dominate international information networks. Videotex was successful only in France as Minitel, and X.400 message handling remains a relatively unused standard for e-mail. The book is a valuable guide to the international regime of standardization, even as its serves as an example of how the internet has bypassed any such official international regimes in favor of its own standards processes, explained below. Schmidt and Werle begin by making analytic distinctions:

"Analytically, three modes can be distinguished. Governments may impose mandatory standards hierarchically as binding solutions to coordination problems. In markets, standards can emerge as de facto or industry standards. Their diffusion is based on market leadership or on frequency-dependent bandwagon and imitation processes, in which the number of actors attracted by a standard increases with the number of those who have already adopted the standard. A growing number of standards are agreed on in committees, which, whether explicitly institutionalized or less formal in nature, are dedicated to the joint elaboration of standards. Typically, these standards are voluntary consensus standards. (43)"

4. The italicized words are powerfully self-evident and necessary discriminations for any attempt to understand the standardization of the world. We have seen this distinction implicit in many of our cases so far. We can read 'network effects' in this reference to markets, and we can sense the problem of healthcare between governments and committees in any given standardization effort. On the one hand, they are necessary to understand the political nature of standards and the relationship to governance, on the other they obscure by reference to a supposedly clean tripartite distinction of politics, economics, and technology. It is not the role of this analytic distinction to question governence per se, but rather to assume on good faith, that governence is conducted by governments.

5. Schmidt and Werle go on to explain that in their case-study, telecommunications has always been a national endeavor, rarely coordinated at a supra-national level, and highly controlled by national governments. Several projects of nation-building also meet this criteria: railroads and national highway systems, for example and there can be little denying that such efforts, including telecommunications, are tightly coupled with the history of nationalism and colonialism, and have historically served as one of many measures of 'development' and national power. In addition, international telecommunications development has been part of the project of the creation of an international order of secrecy, diplomacy, war, and economic liberalisation. This much must simply be assumed[1].

6. Smith and Werle's work is an invaluable reference for the complicated ecology that is the world of standards organizations that have developed in the international order. They begin by identifying the epi-center of standardization: ISO/IEC and ITU.

7. ISO (pronounced eye-so, and not an acronym. According to ISO, whereas the acronym for the International Organization of Standardization would be different in each language, ISO, as in 'isonomic', derived from the greek, reads the same in 'any' language. A small clue to the fact that international still means Euro-American, in particular, Hellenic Euro-American) began in 1947, as an international organization "the object of which would be to facilitate the international coordination and unification of industrial standards." Today the group's membership is made up of about 90 countries, each represented through its own national standards body (such as ANSI in the US). The IEC (International Electrotechnical Commision), founded 1906, complements the functions of ISO. ISO's successes are familiar to everyone, a low-level recognizablility that drapes the globe: international paper sizes (A6, A4, etc.), international film speed codes, freight containers, screw threads, telephone and bank cards, and now the ISO 9000 management and quality control standards that are required of any firm doing business in the EU (aside: see transcript with Jan Moe for some details on the difference between ISO and FDA). ISO's stated goals are unashamedly in support of the global neo-liberal consensus:

Worldwide progress in trade liberalization
"An industry-wide standard, internationally recognized, developed by consensus among trading partners, serves as the language of trade. "

Interpenetration of sectors
"No industry in today's world can truly claim to be completely independent of components, products, rules of application, etc., that have been developed in other sectors."

Worldwide communications systems
"Full compatibility among open systems fosters healthy competition innovation, improved productivity and cost-cutting. "

Global standards for emerging technologies
"In the very early stages of new technology development, applications can be imagined but functional prototypes do not exist. Here, the need for standardization is in defining terminology and accumulating databases of quantitative information. "

Developing countries
"Development agencies are increasingly recognizing that a standardization infrastructure is a basic condition for the success of economic policies aimed at achieving sustainable development. Creating such an infrastructure in developing co untries is essential for improving productivity, market competitiveness, and export capability. "


8. Across the street from ISO in Geneva sits the ITU (International Telecommunications Union), one of the oldest international organizations, which started as an international telegraph convention in 1865. In 1947, the ITU became a part of the United Nations, and in 1956, the CCITT (Comité Consultatif International Télégraphique et Téléphonique) was formed out of two earlier groups that focused on standardizing telecommunications. In 1992, after apparent simplification and reorganization, the CCITT was folded into the ITU-T, one of three sections that now include "radiocommunications" (a spectrum is haunting Europe...), telecomunications development (working groups of the world...), and telecommunications standardization (first draft as tragedy...). According to Schmidt and Werle, "The historic division of labor among the CCITT, the ISO, and the IEC has been eroded because a clear separation of technical domains has proved to be unfeasible as information processing and telecommunications rely to a considerable extent on the same basic technology."(50) Thus the recommendations of the ITU-T are often published as ISO/IEC standards.

9. Already, we can see that the analytic distinction above is tenuous at best, since ISO is not a government (national or international) agency, nor simply a committee, given that its standards are so widely obeyed, and in some cases mandated by the EU, such as the ISO 9000 standards (for that matter, ITU's legitimacy is uncertain as well, to the extent that all international governing bodies' legitimacy is tenuous). But the problem becomes more difficult as Schmidt and Werle continue. In addition to these three types of standards (government, market, committee) they identify standards created by "para-standardization" bodies in an attempt to influence the decision making of ISO and the ITU. Their examples include the X/Open group, ECMA (the European Computer Manufacturers Association) and the various scientific and technical committees such as those of the IEEE (Institute of Eletrical and Electronics Engineers). Para-standardization committees are formed with particular interests (in the case of ECMA and to some extent X/Open, they were formed as a reaction to the dominance in Europe of IBM), whose goals may not necessarily coincide with those of ISO and ITU. These groups are analogues of the american administrative state's experts, fact-finders, working groups, white-paper writers, and recommendation-makers. In most cases they are autocthonous groups, formed out of industry cooperation for mutual gains, posing as expertise for the purposes of steering technical decisions past politicians but not politics. In some cases, they may be advocacy, or consumer groups, but perhaps with the recent exception of an ICANN Watchdog group, few such groups have concerned themselves with this level of detail [2].

10. Schmidt and Werle include under this "para-standardization" rubric— as one network among others— is ISOC (the Internet Society) and its standardization body, the IETF (Internet Engineering Task Force). The Internet Society explains the hierarchy:

"At the technical and developmental level, the Internet is made possible through creation, testing and implementation of Internet Standards. These standards are developed by the Internet Engineering Task Force. The standards are then considered by the Internet Engineering Steering Group, with appeal to the Internet Architecture Board, and promulgated by the Internet Society as international standards. The RFC Editor is responsible for preparing and organizing the standards in their final form. The standards may be found at numerous sites distributed throughout the world, such as the InterNIC."

11. It might appear that ISOC's structure of approval is complicated, however, a comparison between the organization of ISOC and ITU/ISO reads like a parody of bureacratic centralization. The IETF meets three times a year, membership is open to anyone, working groups post RFCs (Requests for Comments) and then dissolve. In contrast the description of the ITU-T/ISO is hopelessly elaborate, with a set of bureaucratic procedures, rules for membership and variety of variously formal connections to ISO/IEC, national standards organizations, and industry consortia (the diagram in Schmidt and Werle's book, for example, is maddeningly crisscrossed by boxes and lines of various shades). The IETF publishes all of its standards freely, and available at no charge, uncopyrighted as RFCs. Until recently, the ITU's standards were expensive and difficult for individuals to obtain. According to Schmidt and Werle, the definition of the standards must be kept strictly separate from the implementation of the standard, precisely the opposite of the IETF model, "rough consensus and runnning code" in which a vote must be a clear majority, and must be based on at least two working implementations of the standard.

12. Officially, the ITU/ISO and ISOC are not competing organizations. Many aspects of the internet technologies are governed by ITU/ISO standards, e.g. the various character sets that currently allow for the internationalization of documents on the internet (recall my example of the character for Apple, which is not included in either ithe ISO-charsets or in the Unicode format). But the fact that the ITU did not identify the IETF or ISOC as committees that should be collaborated with or consulted— until 1995— is a telling oversight. In fact, it is the ITU and ISO's particular relationship to standards that provokes Carl Malamud's ire in his Exploring the Internet, in which he contrasts the openess of the internet standards process with the relentlessly mocked bureacracy of the ITU: "In June of 1991, after endless messages back and forth to Tony [of the ITU], I had received a fax from the Secretary-General asking me to post all ITU standards on the Internet on an "experimental" basis. I promptly booked a plane ticket to Geneva for August to pry the data out of the cold clammy hands of the bureaucracy."(5) Malamud's goal of posting the ITU standards on the internet was stifled by both ISO and ITU, and it serves him as a repeatable lesson on the openness of the internet over against the crypt of bureaucratic standards organizations, inevitably associated with death and decay.

13. The internet standards and the procedure of creating and validating them that the IETF and ISOC use date back to almost the beginning of the internet. The first RFC posted in 1969 (RFC1, "Host Software"), began a tradition of semi-formal standardization that would develop within universities (the first four nodes were UCLA, SRI, UCSB, and University of Utah) connected to the ARPAnet. What it is important to recognize is that these RFCs often describe an implementation of an existing protocol, something that should be tested, is being tested, or has been proven to work. They are not theoretical documents that attempt to standardize a protocol in the absense of any implementation, nor are they documents that mediate between competing standards. Theoretically, competing, redundant standards could exist, and both be 'standard' by this procedure— if there were working implementations— both would be open, and both would be usable. Over the 30 years since 1969, over 2600 RFC's have been posted. Not all RFC's are standards, most, in fact, are exactly what they say they are: a request for comments. Some, however, are published as the final word on a particular internet protocol. Periodically, the existing protocols and standards are themselves published as an RFC (the most recent is RFC 2500). With 20 years of development as a relatively small research and education network, the standards had plenty of time to improve and to sediment, sometimes as organic academic projects, sometimes as responses to unexpected uses of the internet. These standards became the de facto standards for the internet not because the IETF or the RFC process itself possessed sufficient legitimacy to enforce them, but rather because legitimacy came in that particular form unavailable only to internetworked software, summed up in an oft-cited phrase from Dave Clark "No kings, no priests, just a rough consensus and running code."
    In short: "It works."

The following introductory text is quoted directly from RFC 2026: "The Internet, a loosely-organized international collaboration of autonomous, interconnected networks, supports host-to-host communication through voluntary adherence to open protocols and procedures defined by Internet Standards. There are also many isolated interconnected networks, which are not connected to the global Internet but use the Internet Standards. The Internet Standards Process described in this document is concerned with all protocols, procedures, and conventions that are used in or by the Internet, whether or not they are part of the TCP/IP protocol suite. In the case of protocols developed and/or standardized by non-Internet organizations, however, the Internet Standards Process normally applies to the application of the protocol or procedure in the Internet context, not to the specification of the protocol itself. In general, an Internet Standard is a specification that is stable and well-understood, is technically competent, has multiple, independent, and interoperable implementations with substantial operational experience, enjoys significant public support, and is recognizably useful in some or all parts of the Internet. (emphasis mine)

14. As of RFC 2500, there were about 50 Internet standards, 60 draft standards, nearly 400 proposed standards, and over a 120 experimental protocols, another 100 informational protocols and 50 historical protocols. Internet standards are final and proven (they 'work'), draft standards are, according to the IETF (RFC 2026): "normally considered to be a final specification, and changes are likely to be made only to solve specific problems encountered. In most circumstances, it is reasonable for vendors to deploy implementations of Draft Standards into a disruption sensitive environment." Proposed standards do not receive this recommendation, and are considered to be in the testing phase. Informational Protocols are those that are proposed for some service not intiated by the working groups in the IETF or initially proposed to the IETF. They may eventually be included in an IETF standard, but in the meantime they are published only "for the benefit of the internet community." The description (RFC 2026) of "historic protocols" register the playful, occasionally recursive pragmatic IETF RFC process: "A specification that has been superseded by a more recent specification or is for any other reason considered to be obsolete is assigned to the "Historic" level. (Purists have suggested that the word should be "Historical"; however, at this point the use of "Historic" is historical.)" [3]

15. The completed internet standards include all of the low-level protocols for the most commonly used services on the internet, including definitions of IP for use on over 15 different kinds of physical networks (e.g. ethernet, ARPAnet, wideband, ATM, token-ring LANs, carrier pigeon etc). Among these services are the basic protocols such as TCP (Transmission Control Protocol), telnet, ftp (File Transfer Protocol), SMTP (Simple Mail Transfer Protocol), DNS (Domain Name System), SNMP (Simple Network Management Protocol), PPP (Point-to-Point Protocol), POP3 (Post Office Protocol 3).

16. It is interesting to note that as of June 1999, the standards for familiar world wide web services such as HTTP (both 1.0 and 1.1), URI (Universal Resourse Identifiers), URL (Universal resource Locaters), and HTML 2.0 were all either proposed standards (not yet approved for 'disruption sensitive environments') or they were informational protocols, published but not approved. One reason for this is that a second internet-related organization has come into existence— The W3C (the World Wide Web Consortium)— which has managed to assume much of the legitimacy granted the IETF, but with a very different structure of participation and openness. The W3C allows only organizations to join and has a much less explicit relationship to the "recommendations" it produces than the IETF or ISO/ITU. The W3C's recommendations for HTML (HyperText Markup Langauge), CSS (Cascading Style Sheets), HTTP (Hypertext Transfer Protocol) and XML (eXtensible Markup Language) are widely followed, but the process for participating is much more complex than that for the IETF. In addition, these recommendations are generally not as focussed on the implementation, as on a normative standard which corporations such as Netscape and Microsoft are trusted to follow— more in line with the model of ISO and ITU. Simson Garfinkel explores some of the W3C's recommendations, including their proposals for various privacy measures (such as the PICS system) in a recent Technology Review article [4].

17. The emergence of the W3C came with the first wave of hype centered on the growing popularity of the internet and the web, from about 1993 on. The techno-libertarian hand-waving of Wired and the promotion of the MIT Media Lab were two important forces in this wave of hype. During this time, Tim Berners-Lee was crowned the inventor of the Web, and was invited to come to MIT's LCS. Here, the W3C began in earnest. By virtue of being the only recognizably legitimate organization dealing with web standards (and how this recognition occurs is mysterious and significant) and by convincing several large software companies like Microsoft and Sun to join, the W3C has emerged as the standards organization for the World Wide Web. Although I do not follow out the implications of this, it is clear that the W3C represents a departure from the mode of governance of the IETF, towards that of ISO, and away from standards and towards recommendatiosn.

18. To return to the question of standardization posed by Schmidt and Werle, where the IETF is consigned to a "para-standardization" role, it should be clear that what is at stake is not the standard-making process per se, so much as its legitimacy. For Schmidt and Werle ISO/IEC and the ITU-T are the legitimate international organizations of standardization, just as ANSI is in the US or BSI in Britain, etc. by virtue of their history and some unarticulated mechanism through which national and international legitimacy is produced and retained. In fact, it may be the case that these two organizations, in their various forms, have controlled a great deal of the standardization of electronics and telecommunications over the last 50 years, but this does not necessarily explain their legitimacy. Organizations that form outside of this legitimate hierarchy are therefore presumed to be in relationships of support, influence, or subversion, but rarely as direct competition. The IETF, therefore, is not simply para-standard, but parasitic.

19. Schmidt and Werle are well aware of the political stakes of standardization: "Negotiating standards among representatives of nations opens the door to political considerations which are not directed at technical questions of compatibility but which tend to regard a national standard as a strategic element in the global competition among nations."(267) The political problem then becomes a problem of strategy amongst representatives of various nations, Schmidt and Werle discuss the problems of "relative-gains" as the political problem of this function of standardization as international coordination. The technical content of the standard may or not become an explicit part of this discussion (Schmidt and Werle insist that it doesn't), but its implementation surely does. Schmidt and Werle misread this problem: "Politics remains blind to the technicality of standards. It is activated when standardization (most likely of a comprehensive technical system, such as videotex, not of isolated components), has gained considerable national political significance" (271). This blindness is an artifact of the assumption that the ITU and ISO are not themselves subject the problem of standardization, that is, it is assumed that there is widespread acceptance of the ITU/ISO standardization process as a neutral technical game, out of which politics emerges when nations compete for comparative advantage.

20. This is clearly not the case. First, Schmidt and Werle recognize that the process is actually inter-organizational, not inter-national, and they go on to suggest how standardization often reflects national interests only through national support of a given organization (corporations in America, for instance, where almost all public utilities have been privatized). As such, the technical content of standards is the most important concern for competing organizations (though Schmidt and Werle insist that politics only emerges when national concerns are aired), and whether this is couched as a technical issues, or an economic one (and the line is fiber-optic thin, since efficiency is a chimeric techno-economic measure), this is still a political issue in both constitution and effect. If the actors do not represent it as politics, this is disingenuousness, as we will see again in the case of the open source movemnet. Second, the nature of the standards process is fundamentally different in the ISO/ITU world from that of the IETF, and this fact is both political and technical at the same time. Legitimacy and bureaucracy, and all of the attendant histories they invoke (from socialism to communism, from welfare states and social democracies to regulatory and administrative thickets), are the issue, and IETF's counter-proposal is openness based on that simple insistence: "it works." ISO/ITU stress control over a process when they standardize. Everything must be specified and the specifications must be rigourously separate from any implementation. It is a model of careful, deliberate, consensus-based standardization. The standard is expected to be obeyed because it has been approved by these organizations, and owing to their legitimacy, can be called complete. The IETF on the other hand accepts a proposal only if it has a working model associated with it. Complexity is mistrusted, but only to the extent that a standard might try to bureaucratically (read pointlessly) standardize something which might never occur in practice. As such, the model is one of lowest-common denominator standardization based on the existence of implementations. Pragmatic, fulidly programmable standardization.

21. The point here is not to insist that the way that the IETF standardizes is somehow objectively better than the model of international governing bodies, it is simply to illustrate how these two systems function as examples of differing political assumptions. The former points to an unarticulated notion of openness that is a familiar (American) reaction to centralization or bureacratization, resting on a kind of libertarianism that is disingenuous about its role in the constitution and shaping of the political sphere. The latter is successful example of an international coordinating institution in the service of the (European and American) neo-liberal consensus on economic liberalisation. It should be clear from this which political models do not exist as background to standardization, and which countries do not get representation except by submitting to American openness or European bureaucracy.

22. Now we can answer the question about stupidity— who gave us IP— and press further the question of yoking politics and openness to standards. The answer: the ARPA/DoD-funded, international university network of computer scientists and 'hackers,' through the IETF, gave us IP, and hence, this thing we call the internet. All of this may seem dreadfully obvious, but the fact remains that the internet has become the information infrastructure of the world, and that its specific technical constitution and history is essential for understanding what that means.

23. ISO and the ITU have not sat calmly by while this happened. In an article called "The Internet Religious War," [Drake93] William Drake explains how the 1970s and 1980s have seen a major split between TCP/IP (the IETF standard) and the ISO model for internetworking: The OSI (Open Systems Interconnection) reference model. OSI was supposed to be an alternative to "DoD/IP" and the perpetual threat of IBM (in the form of their Systems Network Architecture). His explanation echoes stupidity: "The debate is not merely the comparative efficacy of two sets of standards, but is rather between two competing visions of how international standardization processes and network development should be organized and controlled. Like ISDN, OSI is a remnant of what some observers call 'old paradigm' standardization, a model dating back to the early years of technical coordination at the ITU. It assumed that 'networks' meant national public switched networks configured as closed hierarchies and centrally controlled by monopoly PTOs [Public Telephone Operators]" (634). The article engages in much more new paradigm/old paradigm talk, proceeding by way of a review of Carl Malamud's book [Malamud92]. We are familiar by now with this engineering trope of the obviousness and simplicity of progress and the dense retrogressive nature of anything opposed to it, its power as explanans effaces almost any approach to the explanandum itself. But Drake at least opens the question as to what constitutes the important difference between the open internet and the undead bureaucracy of the ITU and ISO:

"The problem is not simply with "the bureaucrats" at the ITU, some of whom are quite progressive in their personal views about creating a more open body, but more importantly with the national administrations who actually control most aspects of ITU policy. No international organization secretariat can push member governments further than they want to go and unilaterally sustain controversial initiatives without expending major political capital, even if the opposition is wrong-headed and bad for the long-term prospects of the body." (648)

24. The representation Malamud gives, which is a typically self-righteous hacker-libertarian obnoxiousness, is countered here by a concern for what Drake calls "consensus-oriented incrementalists." The question then remains: " Is there a viable rationale for European governments to remain wedded to a future OSI and adopt restrictive policies on that basis when, for all practical purposes, TCP/IP now appears to have won the religious war through global user adoption? Can standards organizations legitimately claim copyright on their products? Would not broader dissemination of their standards be in the public interest?" (649)

25. This article was published in 1993, which is five years before, as an example, France officially conceded defeat in this religious war by organizing a National Fete de L'internet in March of 1998, and the question registers an honesty about the uncertainty of the future that has since been paved over by the techno-yuppie libertarian mantra of inevitablility. These questions were never officially or unofficially answered, or even publically addressed, partially because the growth of the internet has made people assume them answered, partially because five years actually isn't that long a time relative to the twenty five that have elapsed since the debate first began.

26. The introduction of the world-wide web, and the massive growth in connected institutions and users that followed also introduced new questions and new political issues that have yet to be addressed. The process of standardizing the internet has become much more formal, and the openness has served to increase the legitimacy of the IETF, ISOC, and the W3C. Given several generations of internet hackers, computer scientists, and amateurs trained in this standardization by implementation, the limited formalization that has taken place has been left largely unexamined. Unexamined because no one was really complaining about it, because this very specific, particular constitution of the internet has been maddeningly and ceaslessly obscured by a techno-yuppie, third-wave hand-waving media claiming it as an unregulated, ungovernable, libertarian paradise of free information and frictionless commerce (see only John Perry Barlow and the early EFF for starters... But the list is endless and very familiar by now), but most of all, because "it works."

Last Modified 11-Sep-99 9:01 PM ckelty@mit.edu

Go Back to the Start