NOTE ALSO: At the time of this writing, October 1998,  the last sentence of the first paragraph seems overly strong. The initial draft of this paper was written from the perspective that the Green Paper of January 1998 had blocked the MoUvement. The paper in large part analyzed the defeat of the MoUvement. However, the release of the White Paper gave new life to the MoUvement's prospects, and the context of that event motivated the last sentence of the first paragraph. At this time, things are very much undecided.
 

by Craig Simon.

A case study for Bandwidth Rules

This is a rough draft, not running code.
 
 

The Mouvement (pronounced like movement) discussed in this article refers to supporters of a document titled "Memorandum of Understanding on the Generic Top Level Domain Name Space of the Internet Domain Name System." The document's name is frequently abbreviated to gTLD-MoU, or just MoU, hence Mouvement. It is a clumsy word, but one of many awkward terms in a story that contains enough acronyms to fill a whole can of alphabet soup. The Mouvement allied a diverse group of Internet engineers, trademark attorneys, and officers of Geneva-based intergovernmental organizations in a bid to establish formal power over a crucial technical feature of the Internet called the domain name system (DNS). A key institution of the proposed structure was called the Council of Registrars (CORE). The Mouvement's effort to have CORE's computers recognized as the authoritative working center of the DNS was opposed by a conglomeration of competing business interests, and, for several months, by the U.S. government. An alternative institutional framework is now being established and is likely to incorporate several of the Mouvement's central principles.

With annual revenues under $200 million, the domain name registration business accounts for a relatively small share of the global Information Technology (IT) industry. Nevertheless, the DNS is a fulcrum of Internet activity. The battle to reorganize it, which erupted in September 1995, is central to understanding how the people involved in designing the Internet as a global structure are challenging the rules, and therefore the rule, of the existing system of sovereign states.
 

The following discussion is organized into four sections. The first will serve as an introduction that briefly presents the empirical background of the discussion and raises a conceptual framework. The second is meant to provide a lengthier, straightforward accounting of the salient technical and historical details. The third will focus on conceptual and analytical questions, opening the way for an argument (to be added in the next draft) where facts and framework converge. The fourth section, an attached document, includes my submission in response to the U.S. government's February 20 1998 Notice of Proposed Rulemaking titled "Improvement of Technical Management of Internet Names and Addresses."
 

INTRODUCTION

DNS management is currently vested in the Internet Assigned Numbers Authority (IANA), a group of half a dozen or so people who are actively involved in monitoring Internet standards and providing key technical services to individuals associated with the Internet Engineering Task Force (IETF). Both these organizations evolved out of the ad hoc Network Working Group (NWG) that began activities in the late 1960s and morphed into the Internetworking Conference Control Board (ICCB). The ARPANET, then the NSFNET, and finally what we know now as the Internet emerged due to the work of these groups (Hafner and Lyon 1996, Braden 1998). The IANA has operated since 1984 under the direction of Dr. Jon Postel, Associate Director for Networking, at the University of Southern California's Information Sciences Institute (ISI) through a research grant funded by the Defense Advanced Research Projects Agency (DARPA). That grant reached approximately $600,000 in its last year, fiscal 1998. The search for alternate funding sources for IANA was therefore a pressing concern for both the Mouvement and interested parties in the U.S. government (Rutkowski 1998).
 

The accelerating growth of the Internet in the early 1990s underscored the increasing importance of the institutions which formulate and regulate its standards. Thus, to provide IANA with a clearer status, it was formally chartered in 1992 by two groups: 1) the Federal Networking Council (FNC), a US government agency constituted by individuals who had been enduring proponents of the ongoing internetworking research, and; 2) the Internet Society (ISOC), a non-profit Internet support and advocacy organization. ISOC's declared mission is, "To assure the beneficial, open evolution of the global Internet and its related internetworking technologies through leadership in standards, issues, and education."⁽¹⁾ ISOC's board of trustees include a number of distinguished individuals from the information technology industry who share a strongly outspoken idealism about the potential virtues of Internet-based communication. They share a proclaimed desire to provide stewardship for the Internet through that organization, which is committed to provide an umbrella of insurance, financial support, and non-intrusive oversight for the IETF and its governing committees, granting assistance when called.
 

The IETF is not incorporated, and is renowned as a notoriously informal, clubbish, and steadily growing group of technical specialists drawn from large and mid-sized Internet-related firms, as well as from government and academia. Its processes are remarkably open.⁽²⁾ Entry to the tri-yearly IETF technical meetings is granted to anyone who pays the relatively low price of admission--about $300. Nearly 2,000 people now attend its week long meetings, where participants thrash out the details of new Internet standards in open sessions and from a generously endowed computing and connectivity facility called the "terminal room," all the while feasting on free food and coffee. In addition, numerous IETF online mailing lists are open to all comers, and all contributions are voluntary except for the work of a small secretariat. Members nevertheless abide according to a highly disciplined, self-described meritocracy within an organizational structure that includes several clearly delineated and carefully supervised technical areas, each containing a number of Working Groups. Two committees provide oversight--primarily the Internet Engineering Steering Group (IESG), constituted by the Area Directors, and also the Internet Architecture Board (IAB), a panel of technical luminaries.⁽³⁾ Internet standards--whether finished, experimental, or informational--are humbly referred to as Requests for Comments (RFCs). The IANA's position as publisher of the RFC series, and the assignment of the names, numbers, and protocols that enable those is so critical to the IETF's work that Jon Postel is properly regarded as a focal point of authority for the entire Internet.
 

Although responsibility for the distribution of assignments rests with the IANA, over the years a significant portion of those day-to-day operations have been delegated elsewhere. Through the 1980s into the early 1990s, the DNS operations center, institutionally known as the InterNIC (Internet Network Information Center), was run by the Stanford Research Institute (SRI) under contract with the Defense Information Systems Agency (DISA). In January 1993 the InterNIC was moved to Network Solutions, Inc. (NSI) of Herndon, Virginia, through an instrument called the Cooperative Agreement. This was authorized and funded at an annual rate of approximately $1,000,000 by the U.S. National Science Foundation (NSF).⁽⁴⁾ The award and redelegation took place in concert with the IANA. The agreement was scheduled to expire on March 31, 1998, but its language allowed for extension. Two other companies, AT&T and General Atomics, were named as contractors for specialized tasks in the Cooperative Agreement, though General Atomics dropped out in 1995.
 

In mid 1997, the NSF announced that it would not renew the Cooperative Agreement, but retained the option to activate a provision for a six month "ramp down" extending the NSI's status through the end of September 1998. Due to the general push for privatization of such activities, US funding for the IETF (under $1,000,000) had already been spun off by the beginning of 1998. Thus, the various contracts with the IANA and NSI were among the last formal relations with the United States government that still chained directly from the Internet's origins as a Cold War-era, Pentagon-funded research project.
 

Despite the legacy of these sustaining ties to the US government, members of the IETF and other Internet standards making bodies have developed institutional practices which are highly independent of US control. The IETF motto stresses "rough consensus," traditionally implying that authority for the group's actions arises from within the group, rather than being imposed from outside agencies. CORE supporters have articulated sophisticated normative precepts that are fundamentally at odds with traditional conceptions of nation-state sovereignty, and they have openly derided the competence of many US officials involved in making Internet oversight policy. CORE ostensibly arose from a joint effort by ISOC and the IANA to design a successor structure prepared to take over DNS operations after the expiration of the Cooperative Agreement with NSI. But it also reflects deliberate attempt by elite members of Internet community to act on an inchoate set of geopolitical precepts and wrest Internet governance from the hands of U.S. sponsorship. It was one of many attempts arising from within the Internet community to accelerate the construction of an expressly global medium for human interactivity.
 

All sides to the CORE controversy--including U.S. government officials and legislators--claimed to favor US divestiture from the Internet's management functions, leading to full privatization of DNS oversight and IANA's other activities. With few exceptions they spoke glowingly of the ultimate "internationalization" of the Internet. Despite these expressions of shared principles, however, the controversy was waged fiercely, focusing on the efforts to charter CORE as a non-competitive, non-profit "public trust" based outside the United States. CORE sputtered after the Clinton Administration intervened in late 1997 and early 1998 to reassert the power of the U.S. government over Internet policy making. The leading agent on behalf of the administration was Ira Magaziner, Senior Advisor to the President, a political ally who had worked closely with Hillary Clinton years before on the administration's health care proposal. Magaziner headed an interagency working group that included: J. Beckwith "Becky" Burr, Senior Internet Policy Advisor of the National Telecommunications and Information Agency (NTIA); Brian Kahin and Karen Rose. Magaziner's group initially deterred Postel from taking the steps necessary to implement CORE's plan, but after a long and notably open review of the issue, the interagency group produced a White Paper that conceded many critical points, and "punted" the problem to the "private sector," sending it effectively back into the hands of the IANA and the Internet Engineering community. This long episode deserves close attention for the issues it raised with regard to burgeoning interest in what is often called "global Internet governance."
 

The Mouvement's campaign to establish novel forms of operational and regulatory control over an expressly global infrastructure should be seen (at least in part) as presenting a challenge to forms of rule which stem from agreements between sovereign states. A review of this controversy can demonstrate how people might eventually acquire a sense that they are participants in a global society rather than subjects of an international one. It also provides insight into the ways contemporary technocratic elites are exploiting their standing as trusted professionals, engaging in social engineering under the rubric of machine making. And it reveals how the presuppositions of Western liberalism are being re-articulated in the so-called Information Age.
 

Although the Mouvement's leaders often spoke of "internationalization," they were in fact promoting an ambitious form of globalism, one that may eventually constitute the most sophisticated technical mechanism of social organization yet seen in human history. Their ambition can be made clearer by applying a constructivist analytical framework (Onuf 1989; Kubalkova, Onuf, and Kowert 1998), showing how the production of a related set of technical, managerial and commercial rules could have culminated in a coherent form of social rule--a set of socially conditioned influences through which people coordinate their everyday routines (Giddens 1984). All proponents of Internet expansion favor the growth of markets and other institutional practices which constructivists categorize as heteronomy. Leaders of the Mouvement had also developed an astute vision of how to establish forms of hierarchy appropriate to facilitating that end. The Mouvement's attempt to actualize that vision led to a battle which raised questions of hegemony over the Internet. The constructivist vocabulary and other pertinent conceptual issues will be addressed shortly. For now it is appropriate to continue introducing the technical and historical background of the MoU dispute.
 
 

Evolution of the DNS
 

The DNS is the part of the Internet's architecture that links unique expressions--domain names--like miami.edu with underlying instantiated physical addresses--IP numbers--that read like 129.171.32.100. The current version of the Internet Protocol, IPv4, uses the formatting style nnn.nnn.nnn.nnn, also called the quadruple octect, or dot quad, allowing the numbers in each of its four sections to range between 0 and 255. This masks four octets of 2⁸ for a maximum of 2³², or nearly 4.3 billion possible addresses on the system.
 

Whereas IP numbers are absolutely essential to Internet operations, domain names are not. Still, the DNS provides a navigation mechanism which is so convenient and attractive, domain names have become nearly ubiquitous in global culture, and many people therefore consider the DNS to be an essential facility of the Internet. The DNS is employed for several reasons. Its method of identifying locations provides a reliable way for people familiar with the English alphabet to "surf" the Internet's popular World Wide Web service using names they can often easily guess or recall. The next planned upgrade of IP--version 6-- will contain 2¹²⁸ (3.4*10³⁸) addresses, making the expression of IP numbers even more cumbersome. The DNS also greatly enhances the portability of the Internet's stored resources among its physical locations. As a consequence, the numeric address of a resource's "host" can change without altering the name of the destination that people rely upon to access it. This flexibility is highly useful for maintenance and other purposes. It permits a dynamism in the Internet's physical structure that can be hidden from account holders, and thereby greatly stabilizes the outward behavior of the system.
 

Host names were initially single words, like ibm or usc, resulting in e-mail addresses like myaddress@thisplace. By 1983, after about two thousand hosts had been connected to the Internet, it was clear that this flat addressing scheme was becoming unwieldy. Based on work initiated by Postel and Paul Mockapetris, it was slowly replaced by a hierarchical scheme requiring the use of Top Level Domain (TLD) suffixes like .edu, .us, .uk and .gov in conjunction with Second Level Domain (SLD) and optional Lesser Domain (LD) or 3^rd Level Domain (3LD) identifiers listed as composites from right to left.⁽⁵⁾ This extensible domain hierarchy enables coexistence of location designations like sis.miami.edu, law.miami.edu, and even foo.blah.miami.edu. E-mail addresses therefore had to be reassigned to look like myaddress@thisplace.net. The goal was scalability--a design that would facilitate an easy and efficient distribution of names across the system as more names were added. The new scheme, fully instituted by 1986, reserved two letter endings for country codes like .us and .ca., and designated seven three letter endings for prescribed purposes.
 

A partial restatement was published in March 1994 as RFC 1591, "Domain Name System Structure and Delegation" which declares, "The IANA is not in the business of deciding what is and what is not a country." For this, the IANA defers to a list known as ISO 3166, "Codes for the Representation of Names of Countries and their Subdivisions," published by the International Organization for Standards based in Geneva. Subdivisions like the Paris metropolitan area. .fx are allocated at the request of the state having authority over the area. The 10 member committee which performs this work is the Berlin-based ISO 3166 Maintenance Agency (MA), includes 5 regular members from the UN Statistical Division (New York), UNCTAD/UNECE (shared vote/Geneva), IEAE (Vienna), UPU (Bern), and the ITU. The other 5 come from national standards bodies, currently the US (ANSI), UK (BSI), Germany (DIN), SIS (Sweden) and AFNOR (FRANCE). The MA re-evaluates the list as needed to comply with occasional changes in the UN Terminology Bulletin published by the General Assembly.⁽⁶⁾
 

RFC 1591 refers to the other top level domains as generic TLDs. Of these, .mil and .gov are exclusively controlled by agencies of the United States government--DARPA and the FNC respectively. The specialized zones .edu and .int are for educational institutions and international organizations. The remaining three zones, .com, .net, and .org were specified in turn for commercial areas, non-profit organizations, and network infrastructure providers, though NSI no longer takes steps to ensure that registrants fit into these categories. Mockapetris, Postel and others evidently hoped to see the rise of usages like us.mycompany.com and uk.mycompany.com, although this style of naming did not take hold in the commercial area. In fact, the burgeoning popularity of SLDs under .com resulted in a much flatter address space there than the DNS designers initially anticipated. Unsuspecting of the coming boom in demand for new generic TLDs, Postel wrote in RFC 1591, "It is extremely unlikely that any other TLDS will be created."
 

The earliest versions of the DNS relied on the distribution of a file called HOSTS.TXT, maintained by SRI and others, with revisions passed around the Internet almost daily. By the late 1980s, however, the increasing size of the file was a source of difficulties. Updating the source grew cumbersome, and questions arose as to who maintained the most accurate list of named hosts. Also, since the carrying capacity of the data "pipes" connecting the Internet's hosts was limited, the sheer size of the transmitted file made relatively heavy demands on the system's resources. An innovation by Mockapetris introduced in 1989 laid the foundation for an arrangement enabling the interoperation of an array of separate registries supported by a highly efficient and scalable distribution system.⁽⁷⁾ Consequently, the DNS has been called the largest distributed database in the world.(Cricket and Liu, Rony and Rony, Shaw).
 

NSI performs three important activities relevant to the DNS. First, it hosts the primary root of the Internet, located in a machine often referred to as Server A, but properly called a.root-servers.net. Server A is the authoritative guide to all of the DNS registries recognized by the IANA, as well as the SLDs hosted by NSI. It is the Internet's de facto arbiter of all requests through the DNS for IP addresses, but the root does not provide true directory services like a global "411" or "yellow pages." Server A is better described as providing indexed lookup table which serves to redirect domain name queries across the Internet as needed. A chief technical virtue of the contemporary DNS is that its index of names and numbers is distributed so that Server A will not have to field all requests directly. Several times a week NSI propagates an updated zone file from Server A to the secondary root name servers B through M in the US, England, Japan and Sweden. (J is also operated by NSI. B and L, by the IANA.) Tens of thousands of local name servers run by Internet Service Providers (ISPs) around the world point to those various root servers for domain name requests. Those machines store each successfully resolved domain name requests in a local memory cache, so that the more popular IP addresses can be distributed immediately by the nearest server the next time an identical query arrives.
 

There is no legal requirement that any ISP or other machine connected to the Internet must use the existing root system to resolve its DNS queries. However, the most popular software used by ISPs for DNS management--the Berkeley Internet Name Daemon (BIND)--is configured to work this way⁽⁸⁾. BIND is reconfigurable, but nearly all ISPs accept the defaults which point to the IP addresses of the known root servers. BIND is distributed free of charge by Paul Vixie, who is the predominant author of the most popular version. Vixie's company, the Internet Software Consortium (ISC), also operates server F.
 

Second, NSI maintains the registry database for the best known commercial TLDs--.com, .net, and .org--which together account for about 2 million names, nearly two thirds of the total number of names resolvable on the Internet as of April 1998. Other TLD registry databases are dispersed throughout the world, visible through the DNS only because they are listed in Server A. The IANA, at USC, officially provides registry services for .us, and many other national registries such as Canada's .ca are also maintained by academic institutions. This is changing with the Internet's rapid commercialization, however. For example, the German registry was transferred from the University at Karlsruhe to the Bundespost in the early 1990s. Many small or relatively poor countries have begun using technical contacts from the industrialized world to help them manage their national registries. British-based Netnames provides such services for Afghanistan (.af), .American Samoa (.as) Bhutan (.bt), Palau (.pw), and Turkmenistan (.tm) and others. A Paris-based company does the same for Burundi (.bi) Rwanda (.rw), the Republic of Congo (.cg), the Democratic Republic of Congo (formerly Zaire, .cd), and the Channel Islands Geurnsy (.gg )and Jersey (.je). Several sub-Saharan domains are managed by Randy Bush, the IESG Area Director for DNS issues. Bush was directly involved in building much of the Internet connectivity on the continent. Several national TLDs are considered to have been leased, implying that their technical contacts have wide latitude to exploit the global commercial opportunities of the domain name business. This is true of most of the TLDs managed by Netnames, as well as the Cocos Islands (.cc), the British Indian Ocean Territory (.io) and Niue (.nu). This occurred as the .com space became relatively saturated. "Good" names like smith.com were already taken, so the demand for service in new TLDs grew commensurately.
 

Third, NSI serves as a registrar, dealing directly with individuals who seek to acquire or transfer domain names. While a registry database--the zone file--typically only contains fields listing domain names and corresponding IP addresses, a registrar's database also contains contact information for the domain holder, the domain's technical and administrative contacts, plus other fields related to billing. Registry and registrar functions do not have to be housed under one roof as they now are at NSI. In the United Kingdom, for example, Nominet serves as a registry shared among several commercial registrars. The diverse array of registry and registrar services provided by Netnames can also be separated.
 

In the late 1980s and early 1990s, when there was little contention for domain names, SRI, and then NSI had provided .com, .net, .org, and .edu registrations free on demand, upholding the principle "first-come, first-served" (FCFS), and members of the then relatively small and elite Internet insiders club who knew where and what to ask could expect to see their registration requests completed in three days. In early 1994 the journalist Joshua Quittner took advantage of this policy and the general lack of knowledge about the domain names by registering mcdonalds.com and then announcing in a New York Times Op-Ed piece that the McDonald's Restaurant Corporation had missed its chance to secure the address. He later transferred the registration to the corporation in exchange for a $10,000 charitable donation to an elementary school and continued to popularize the issue (and himself) by writing articles with names like "The Great Domain Name Goldrush."⁽⁹⁾ This prompted a scramble for names as various entities became concerned about protecting their names and marks on the Internet. The surge in requests overtaxed NSI's facilities, leading to registration delays as long as eleven weeks. In September 1995, responding to the surge in requests for names, the NSF and NSI amended part 4 of the Cooperative Agreement, dropping the "cost plus fixed fee" provision of the contract, and instead permitting charges of $35 per year for registration of SLDs ending in .com, .net, and .org. (names for educational institutions remained free). A $15 surcharge was also levied on behalf of the US Information Infrastructure Fund. Since initial registrations were required to cover a two year period, the "price" of a name effectively jumped to $100. Domain name registration quickly became a lucrative business for NSI, which invested heavily in advanced equipment and new employees. NSI now averages well over 4,000 new registrations every day, earning about $8 million monthly
 

The institution of charges by NSI was a threshold event in the DNS crisis. Online discussions of domain policy and other Internet governance issues had been occurring in a variety of places such as the com-priv news list hosted by PSInet, a large provider of Internet backbone services. That list was not focused on DNS issues, but on the general question of Internet privatization. NSI also hosted several discussion lists where DNS governance issues frequently came up: rs-talk and rs-info, regarding the root servers, domain-policy, which was intended for discussion of NSI's dispute resolution procedures, and namedroppers which was more oriented toward technical concerns. Related discussions on the Usenet centered on the comp.protocols.tcp-ip.domains newsgroup. Critics of the revision charged that NSI's move was so significant, the entire contract should have been reopened to bidding. Personnel from NSI and NSF responded that the original agreement allowed for such changes.
 

NSI's critics also pointed out that its commercial policies had been growing considerably more aggressive after it was purchased in early 1995 by Science Applications International Corporation (SAIC), a prominent defense contractor with strong connections to retired officers of the Central Intelligence Agency and the National Security Council. (SAIC has also purchased other companies involved in number assignments such as telephone area codes and commercial bar codes.) On June 22, 1995 NSI it had unilaterally changed its policy regarding contested names, announcing that disputed names would be removed from Server A's zone file until the contesting parties resolved the matter through negotiation or litigation. This policy was soon revised so that the contested name could remain visible on the Internet if the name holder agreed to indemnify NSI. The policy was roundly criticized as unfriendly to domain holders. This issue, added to the complaints about registration delays, had simmered over the summer (Oppedahl, Rony and Rony).
 

The response to NSI's September announcement was immediate, including the formation of several new automated mailing list devoted to the issue. The initial conversation soon consolidated within newdom, a list sponsored by Mathew Marnell, a small business operator, under the auspices of the "International Internet Industrial Association" at iiia.org His inaugural message, titled, "So what shall we do?" on September 15 1995 read.
 
 

The way I see it is that NSI and NSF did a naughty thing by not including the general community in their decision. There are quite a few things that could be done I suppose. We could all try to hijack the root domains and point everything at each other instead of at the root servers. Not a good idea. We could all drop our .com, .net, .org domains and moving to geograpical [sic] name space. Might force the NIC to rethink it's new policy. We might create some NewNIC for the express purpose of registering domains for domains that we as Internet users want. This would have to be funded somehow, and would have to be supported, but it would have to be a commercial entity. It would also have to compete with other people that get top level domains and register people that want them. Before posting to the list you may want to think about which domains you'd like to possibly have root level control over. Maybe someday you will, and maybe not, or you could think toward the a new NIC that lots of people can use.  As the ancient curse goes, "May you live in interesting times." We do.(10)

 
 

Marnell reiterated the Internet community's displeasure with NSI in a follow-up message that same day.
 
 

Jon Postel's reaction included the following comment, sent by e-mail to ISOC's board:
 

Most commenters advocated creating alternate TLDs and registries to compete against NSI, and a few, like Crystal Palace Networking even announced their plans to do so. Yet no consensus emerged on fair and proper procedures for awarding new TLDs to companies wishing to provide such service. Alternatively, Scott Bradner, head of the IAB and an ISOC board member, called for a major rethinking that would allow for a more scalable architecture based on shared registries. This would allow for an eventual introduction competition within the .com address space.
 

Discussion continued at a high pace, with the first draft proposal by Postel, Bradner, and Bush appearing on January 22 1996 under the name "Delegation of International Top Level Domains."⁽¹³⁾ It was filed with the other draft RFCs as "draft-ymkb-itld-admin-00.txt." This reflected the normal style for naming such documents, except for the letters "ymkb," which stood for "you must be kidding." An alternate proposal, "Top Level Domain Delegation Draft," by Karl Denninger was released days later, Jan 25.⁽¹⁴⁾ The primary difference between the approaches involved the question of assessing fees on new registries to provide funding for IANA and IETF activities. Denninger strongly opposed this. When the next major Postel draft emerged in March contemplating even higher fees, Denninger and his allies were outraged.
 

(This section will be reworked to follow the chronology and the draft language more closely.)
 

Before long, many more businesses were drawn to the scene, hoping to "cash in" by offering registrations with potentially popular suffixes like .web, .arts, .xxx, and others. NSI, protecting its monopoly over commercial registrations, refused to take the technical steps which would have made those new domains immediately visible throughout the entire Internet. The procedures would have been rather simple, amending the root as if another national registry had just gone online. The alternate registry operators cried foul, claiming they had been censored and subjected to unfair, anti-competitive practices. Critics accused NSI of restraint of trade, but no legal proceedings or official investigations were undertaken at that time.
 

In any case, the final authority, Postel, did not direct NSI to add the alternate TLDs to the root. As the IANA's first and only director, he had traditionally been responsible for accepting the "credentials" of any new national registry. That power was also vested in the IANA by the Cooperative Agreement, which stated that NSI's services were to be provided in accordance with the IETF standard stipulating the IANA's "discretionary authority to delegate [responsibility] with respect to numeric network and autonomous system identifiers."⁽¹⁵⁾
 

Like many Internet veterans, Postel was wary of the way NSI had engineered a financial windfall for itself through its monopoly of commercial TLDs. Yet he hesitated to subject the DNS to a potentially stressful infusion of new commercial TLDs. No one really knew how many the Internet could support beyond the 180 or so then in the system (ISO 3166 country codes still take the lion's share). It was technically feasible to add millions of TLDs in Server A's index, but practical human issues of how to manage the entries kept estimates in the low thousands. The reluctance to open up the TLD space on a first come, first served basis was fortified when it was learned that some of the aspiring commercial registry operators had also engaged in a disreputable, predatory practice known as "cybersquatting"--registering a number of desirable SLDs like nike.com, and then reselling them at premium prices.⁽¹⁶⁾ Some observers estimate that over 15 percent of the SLD names currently registered under .com are held by speculators. One prospective TLD registry operator, ophthalmologist Stephen Page, even claimed the entire alphabet of single characters--.a though .z--indicating that a "land rush" was now imminent for TLDs as well. (Mouvement critics have pointed out that Paul Vixie's ISC was partly funded through the lucrative sale of names like tv.com and radio.com).
 

It fell on Postel's shoulders to maintain order and establish a clear policy. He is not only the primary investigator under DARPA's grant, making him the official keeper of all "unique parameter values" for the Internet engineering community, he is an eminence grise among the community's "greybeards," present at the creation of the very first ARPANET connection between UCLA and the Stanford Research Institute in 1969, out of which the NSFnet and later the Internet evolved. During the September 1997 Congressional hearings on the DNS controversy, Postel described how his role in the early ARPANET experiments evolved into a job of pivotal significance.
 
 

Postel was more interested in work on high speed, high performance computing, and this volunteer drudge work initially occupied only a small portion of his responsibilities. Over the years he nevertheless settled into the position as the Internet's "numbers czar" (at first a term of endearment), and the IETF's esteemed RFC Editor, publishing technical standards, records of best current practices, informational statements, and even some April Fool's Day pranks. He has written or co-authored a substantial number of important RFCs, and is also a member of ISOC's board of trustees. His long tenure in such a technically-focused management position kept him in firmly within the Internet community's leadership circles, but without requiring him to play a partisan role in the debates and deal making that escalated as the Internet grew. According to another Internet veteran, MCI's John Klensin, "Postel doesn't like controversies, and that has turned out to be a big asset to the Internet." Postel's gifts for succinct, accurate communication and a generally low key demeanor also served him well for many years. Yet these strengths were insufficient to the complicated challenge of creating a DNS structure that was both technically viable and commercially competitive. Settling on a technical standard is quite different from creating markets in which behaviors and outcomes are by nature unsettling.
 

Ending NSI's monopoly was central to the task. He underscored that approach in the following way: "What are the priorities here? My list is: 1. Introduce competition in the domain name registry business. 2. Everything else."⁽¹⁷⁾ He began with tried and true Internet engineering techniques, participating in dialogues on public lists like newdom, and inviting comments by uploading drafts of his proposals to the Internet. Such drafts are normally treated as works-in-progress, designed to promote response and revision, with no official status. But consensus was elusive, the arguments grew unusually heated, and polarization resulted. A strong cohort believed that it had been a mistake to create commercial TLDs in the first place. Within this group, one faction wanted all commercial names to revert to country codes, in the form mycompany.com.us. Another faction favored slotting all commercial registrations into a new category of special TLD (sTLD) equivalent to the 37 business categories recognized by the International Trademark Association (INTA). Some argued that it was necessary to ensure that domain names would only be provided to entities which could prove they were legally constituted and accountable under a sovereign jurisdiction.
 

There were calls for imposing waiting periods between application and final authorization of names of up sixty days, during which trademark searches and challenges could be conducted. These and other restrictions were proposed expressly to ensure the rights of trademark holders. Conversely, others believed that domain names should be treated as no more than a manifestation of free speech, and that nothing should be done to impede someone from creating names like i-hate-thatcompany.com. According to this view, there was no need to structure DNS policy to provide relief in cyberspace for trademark holders; the laws of meatspace were already sufficient. There were discussions of new technologies like globally shared registries (which Postel initially doubted was feasible), and futuristic directory services that would someday make the DNS obsolete. There was also running commentary that also included incessant criticism of NSI, the normal share of philosophical bantering, an excess of noisy chatter submitted by class clowns or people of dubious mental health, and the inevitable Internet "flame wars."
 

Contentious new entrants were beginning to create very difficult problems, and not only on the e-mail lists. Some of them had already set up shop as unauthorized TLD registries and were accepting payments for domain names. The best known of these, AlterNIC, run by Eugene Kashpureff, began service on April 1, 1996 with the TLDs .xxx, .nic, .med, .ltd, .lnx, and, .exp.⁽¹⁸⁾ A small portion of the name server operators on the Internet, perhaps around 2 percent at the height of their influence,⁽¹⁹⁾ had reconfigured their software, improvising their way around NSI's root, and making the alternate TLDs visible among themselves. The potential for this type of behavior had always been known, and a vocal minority of Internet veterans with a strong predisposition toward free markets applauded the improvisations. Others hissed. No laws were being broken, but most members of the Internet's "old guard" believed this presented a clear peril to the integrity of the DNS. An ability to sidestep the legacy root violated the rule of uniqueness that underpinned a unified and reliable addressing system. This turn of events threatened to pollute the name space with incoherence, and fragment the community, raising the nightmarish prospect that people would soon have to ask "which Internet?" when using various e-mail and web page addresses.
 

Postel's options under these circumstances were highly restricted, especially by the risk of committing an anti-trust violation. He well understood that it would have been illegal under US law to "bless" any particular company without undertaking a formal process of open contract bidding. Moreover, IETF procedures insist on market neutrality, forbidding the adoption of standards which are proprietary or encumbered by patent royalty obligations. This supposedly increases the value of the IETF's imprimatur and reduces the legal risk of embroiling its members in anti-competitive practices.
 

In June 1996 Postel gave ISOC's board a new draft document named "New Registries and the Delegation of International Top Level Domains,"⁽²⁰⁾ proposing a framework that would be used to add 150 new iTLDs (international TLDs) into the root. 50 new registrars would each be assigned up to 3 TLDs, presumably to promote competition through economies of scale equivalent to the three commercial TLDs held by NSI. The proposal was particularly careful to stipulate that new registries must indemnify ISOC and IANA against any trademark infringement proceedings undertaken as result of action by the new registries or their clients. The motivation for this was that NSI's own terms of service regarding trademark policies were being attacked vociferously for undermining the first-come first-served principle and free speech. (Oppedahl 1996, Mueller 1996). The trademark problem was vexing, raising a host of complex technical and jurisdictional issues, and contributing to the induction of a whole new field of legal scholarship and case law often called cyberlaw (Kahin and Nesson 1997). NSI's dispute policy, initiated in June 1995, displayed great fear of liability in trademark dilution or infringement suits, and resulted in many names in .com, .net, and, .org being put "on hold"--removed from the root zone. NSI does not report these domain name disputes, leaving many unanswered questions about how widely and how consistently its policies have been applied (However, see Mueller forthcoming). It is known that NSI nearly withdrew the registration of juno.com because of a trademark complaint by a company in a different business category. This single move would have wiped out the Internet addresses of hundreds of thousands of people who use Juno's free e-mail service, even though the owners of juno.com would not have been guilty of trademark infringement under normal circumstances. A rising chorus of critics deemed NSI's behavior hasty, vague, and excessively prejudicial on behalf of trademark holders.
 

The indemnification provision in Postel's draft was coupled with a request that ISOC fund a committee that would study the trademark dispute resolution issues and develop guidelines for an organization that would handle the assignment of iTLDs to new registries. ISOC's board accepted the idea, requesting that he flesh it out with a business plan for the committee's work.
 

The effort to solve the TLD problem was soon linked to another pressing issue. Since early 1995, aware that DARPA funding would eventually be withdrawn, Postel had been suggesting ways that closer formal ties with ISOC could assist in generating alternate sources of revenue for the IANA. These ties would be used to help legitimize some sort of fee schedule for the number and parameter assignments the IANA had been performing for the Internet community without charge. His first two drafts raised the prospect that new commercial TLD registries pay a fee to IANA prior to inclusion in the root. He initially suggested a $100,000 fee, which raised a hue and cry, so he reduced the suggestion to $2,000. The idea of paying such fees divided the alternate registries. One group viscerally opposed such a tax, and resented the idea that they should be subject to Postel's authority. The other was ready to invest what was needed to get in on the "ground floor" of the registry business.
 

The next step was to try to develop registry evaluation procedures. The IANA's attempt to do this in the summer of 1996 through direct consultation with aspiring alternate TLD operators like Simon Higgs and Christopher Ambler of IODesign only worsened the acrimony, raising charges of favoritism. Alternate registries like Kashpureff's AlterNIC and Karl Denninger's MCSnet participated in a simultaneous counter meeting called by Iperdome's Jay Fenello in Atlanta, where they hoped to establish a confederation they called eDNS (e standing for enhanced). In any case, the meeting with IANA's representative on July 31, 1996 turned out disastrously. Its everlasting legacy is a hotly disputed tale of the attempt by Ambler to get immediate authorization to go online with his registry, .web. Ambler gave Bill Manning, an IANA employee, an "application" which included an envelope containing a $1,000 check (Cook 1996; Stark 1997). The envelope was returned unopened the next day, and on August 2 Postel issued statements that no commercial TLD registrations were being accepted. But the damage was done. Things looked sloppy. Though the majority of the community remained deeply loyal to Postel, his once unassailable reputation had been sullied. Ambler proceeded to accept registration payments and offer service in .web as an "experimental" TLD, and moved closer to the Alternate camp.
 

Other events of import were taking place around the same time. Increasing interest in the Internet and the DNS issue had prompted the convocation of various panels and policy conferences on the topic, primarily in the US and Europe. A June 1996 meeting sponsored by the Organization for Economic Cooperation and Development (OECD) in Dublin, brought together Don Heath, ISOC's recently-hired CEO, Robert Shaw, the ITU's advisor on the Global Information Infrastructure, Albert Tramposch of the World Intellectual Property Association (WIPO), and David Maher, an intellectual property attorney. Maher was affiliated with the New York-based INTA, and was slated to present a paper on the issue at ISOC's annual Conference, INET, the next week in Montreal. He had expedited the widely-reported transfer of macdonalds.com from journalist Joshua Quittner, and had also been participating in the newdom discussions since January. This began a process of idea exchanges that continued at OECD workshops that summer in Geneva, and in September at a conference hosted by the Harvard Science and Technology Program's Information Infrastructure Project (the conference produced a book: Kahin and Keller 1997). Throughout these meetings, Shaw was outspoken in his opposition to Postel's plan, arguing it would only set up a series of mini-monopolies, replicating the existing problem of subjecting registrants to "lock in" by predatory registries, while making a unified dispute resolution policy even harder to implement. Shaw was not a veteran Internet "insider," but had acquired a well-informed technical background during his years running the ITU's internal networks, and he contributed a substantive overview of DNS management at the Harvard conference (Shaw 1997). Heath was eventually influenced by Shaw's arguments, and finally began calling for a "blue ribbon" panel of experts to rethink the issue. By September 1996, Postel's proposals had been refined to take a more conservative approach, such as starting with thirty new iTLDs building up next to one hundred twenty, and then three hundred. Like NSI, these would operated as combined registry/registrars. Postel also published lists of the new suffixes that were being proposed, though he made no comment regarding which of them might be given priority.
 

One of the chief organizers of the Harvard Conference, Brian Kahin, was both a professional academician and a quasi-officer of the U.S. government, working simultaneously as Lecturer at Harvard's Kennedy School while also chairing the Working Group on Intellectual Property, Interoperability, and Standards of the U.S. Advisory Committee on International Communications and Information Policy. That committee reported to the U.S. State Department and was primarily concerned with copyright issues, so DNS questions were of great interest to its members. After the conference, September 23 1996, Kahin wrote to the co-chairs of the FNC asking them to clarify whether the U.S. government had any claim to ownership over 1) the IP address space; 2) the .com .net and .org TLDs, and; 3) the root.
 

When the FNC Advisory Council met in Washington the next month, October 21-22, DNS issues were a high priority. The report requested by Kahin had not yet been completed, but the attendees nevertheless dealt with DNS issues at length, and passed the following resolution: "The FNCAC reiterates and underscores the urgency of transferring responsibility for supporting U.S. commercial interests in iTLD administration from the NSF to an appropriate entity."⁽²¹⁾ The FNC was also monitoring ISOC's activity. In early October Heath had floated a plan to create a study group called the International Ad Hoc Committee (IAHC). Official public word of the IAHC came on October 22, in an ISOC press release declaring an intention to appoint a nine member panel to "resolve controversy . . . resulting from current international debate over a proposal to establish global registries and additional international Top Level Domain names (iTLDs)"⁽²²⁾ The FNC requested a seat at the table on that same day. According to the minutes: "While not endorsing the [Postel/ISOC] RFC, FNCAC members urged NSF and the FNC to seek membership on this advisory committee, in recognition of the government's historic stewardship role in this sector.⁽²³⁾
 

Pressure to act was increasing. Little real progress had been made in the past year and the impending expiration of the Cooperative Agreement was now less than 18 months away. Paul Vixie was growing impatient and was pushing hard for a solution.(Stark 1997). He is a force to be reckoned with as a result of his outstanding technical contributions in BIND and elsewhere, a selectively strategic willingness to engage in controversy (he is a celebrated opponent of unsolicited commercial email), and a deep loyalty to Postel. On October 31, 1996, concerned that the momentum would be lost in another round of online drafts and argumentation, he wrote to the main IETF mail list under the heading "requirements for participation:"
 
 

The next public step was taken on November 12, when eleven (rather than nine) IAHC panel members were announced. Under the October announcement, IANA, ISOC, and the Internet Architecture Board (IAB) were each allowed two appointments. (The IAB is a body of Internet luminaries that offers direction to the IETF and is available as an appeal body to settle disputes in case anyone claims that the IETF's standards making procedures had been violated.) Postel's two selections were IETF members who had been extremely active and often highly disputatious in the e-mail lists associated with the DNS controversy. Dave Crocker had worked directly under Postel years before, and had also been an IESG Area Director (AD) for DNS concerns. Crocker heads the Internet Mail Consortium and is particularly involved in Internet faxing technology. He also has a name that is "famous" in the technical community, since his brother Steve is one of the most prominent Internet engineering "founding fathers," and had authored, among other things, the very first RFC in 1969. In his writings, Dave Crocker had been consistently outspoken on the theme of how the Internet must address the interests of people outside the United States. Postel's other appointment was Perry Metzger, the youngest member of the panel, a security specialist who had chaired the IETF's working group on Simple Public Key Infrastructure (PKI) and an outspoken voice on the newdom list since its inception. Since it was likely that some sort of public key technology would be used to authenticate DNS registrations, his expertise seemed desirable. Metzger is also a effusive advocate of the Libertarian Party, and a loud critic of the U.S. government's restrictions on the export of software products which employ strong encryption algorithms.
 

Heath's selections for ISOC were David Maher and Jun Murai. Maher had relatively limited practical knowledge of the details of Internet engineering, but his background included a stint as counsel to the American Bar Association on telecommunications matters. His efforts to assimilate into the Internet's culture included attending an IETF meeting wearing a T-shirt touting the PGP encryption format. PGP had been a famous bone of contention between techno-libertarians and the U.S. government, so this was a graceful and astute move to overcome the antipathy that many of the engineers held toward lawyers, especially those who served the trademark industry. Murai was an ISOC board member and a computer scientist who had played a significant part in building his country's Internet infrastructures. The IAB's appointments were Hank Nussbacher, an Israeli-based engineer who had been highly instrumental in building IBM's presence as an ISP there, and Geoff Huston, founder and President of Telstra, the leading ISP in Australia. Huston was also an ISOC Board member and served as its Treasurer. The IAB had initially sought to appoint Simon Higgs, recognizing his early and active participation in the discussions, but he wavered. As the first TLD applicant in September 1995 (for .NEWS), he believed there would be a conflict of interest. One of Postel's initial choices, Christopher Ambler, had declined for the same reason

Single appointments were granted to the ITU which selected Robert Shaw, the WIPO which selected Albert Tramposch, and the INTA which selected Sally Abel, an attorney. At the behest of the FNC, Heath added George Strawn, the FNC co-chair from the NSF. (It may be worth noting here that NSF funding and oversight had been so critical to the Internet over the years, that some IETF insiders called it "daddy.") Having an officer of a U.S. agency on the panel added an aura of stature and legitimacy to the IAHC. Heath became chair, resulting in an eleven member panel. Heath later stated that he had also wanted to add Barbara Dooley of the Commercial Internet Exchange (CIX) and industry lobbying group, but that this was resisted by Crocker.
 

The IAHC conducted its deliberations in a manner far unlike what is normally seen in ISOC or the IETF. Work took place in closed session with a staff counsel present, and participants kept no official minutes or other formal record of their meetings. What is known of the debates within the IAHC has been reconstructed through follow on conversations and interviews; the main elements of the following discussion are generally known and unsurprising. Postel's series of drafts promoting the creation of countervailing registry/registrar entities like NSI was put aside in favor of Shaw's concept of a centralized non-profit registry fed by a globally dispersed network of commercial registrars. Much of the IAHC's time was spent drafting a document that would be used to constitute the appropriate formal organizational arrangements, and designing procedures which would be used to select the registry database operator (DBO) and vet the registrars. The first draft, issued December 19 plan called for soliciting applications from around the world, requiring that registrars provide evidence of capitalization, insurance, creditworthiness, and number of employees. These applications would be audited by the New York accounting firm Arthur Anderson. A lottery would then be used to determine which of the applicants would be accepted, under the restriction that there would be four registrars in each of seven global regions designated by the World Trade Organization: North America; Latin America; Western Europe, Central and Eastern Europe, the Baltic States and the Commonwealth of Independent States; Africa; the Middle East, and; Asia⁽²⁵⁾
 

The new registry's computer would be designed to provide expedited access to outside trademark authorities in order to speed up searches for specific strings of characters that might indicate the existence of potential trademark conflicts. There would also be a sixty day waiting period on the registration of new names. Though many of these provisions clearly served the interests of the trademark community, the plan also satisfied the minimum demands of the Internet community. It constituted a mechanism that could serve to take possession of the NSI registry and introduce competitive practices into the registration business when the Cooperative Agreement expired. NSI would be allowed (even encouraged) to participate in CORE as a registrar, and with a tremendous head start in name recognition, but would lose its monopoly advantage.

(This section needs considerable elaboration, refinement, and correction)

The last IAHC session in mid December dealt with the question of adding new TLDs. By now the IAHC had determined to use the term gTLD (for generic, and hinting at global) instead of international TLDs. According to Dave Crocker, trademark interests in the group wanted a "go slow" approach, with no new TLDs added at all. There was powerful motivation for this. In a proactively defensive move, many owners of "famous names" like Tupperware had embarked on a policy of registering in alternate TLDs, whether or not those registries had any visibility. This was to ensure that no cybersquatter could get the name first, and hold it hostage if that registry was later added to the root. Rather than enrich a proliferating number of zone operators and squatters, the trademark industry preferred to solve the question of dispute resolution in the existing TLDs, and avoid the complexity of dealing with new ones. Crocker and Metzger backed down from Postel's recommended first wave of thirty, and the group chose the number seven as a compromise. With Crocker standing at the whiteboard, they settled on .arts, .firm, .info, .nom, .rec, .store, and, .web. Crocker then asked the four attorneys in the room if the IAHC should be concerned with challenges from IODesign regarding any potential conflict with .web. He was advised that the company had no legal standing. Since Crocker had so insistently called IODesign's registry a "pirate" TLD during the preceding months, he was unlikely to be deterred by any other argument about the need to avoid conflict. Crocker later said he was unaware of the existence of the alternative .arts registry run by the Canadian company, Skynet. This was a fateful decision, akin to driving a car through an intersection when you are sure you have right of way, despite seeing another vehicle in your path.
 

In the IAHC's preliminary report issued December 17, 1996, no mention was made of any additional gTLDs that might be introduced further down the line. This strengthened the impression among the IAHC's increasingly infuriated critics that trademark interests had dominated the process, easing the ability of global brands to defend their names in a highly constrained TLD space. The IAHC's Final Report, issued on February 4, 1997, spelled out the parameters of the new DNS institutional and policy framework. The registrars in CORE would be required to pay a $20,000 entry fee (only $10,000 in less developed regions of the world) plus $2,000 per month, plus whatever fee would ultimately be charged per registration once the system went into operation. CORE would be incorporated as a non-profit organization in Geneva, overseen by a Policy Oversight Committee (POC) made up by nine members identical in composition to the initial plan for the IAHC--two appointments from ISOC, IANA, and the IAB, plus one each from the INTA, WIPO, and the ITU. Disputes between registrants and trademark holders would be arbitrated by a new structure called Administrative Challenge Panels (ACPs), organized through WIPO's Mediation and Arbitration Board. The ACP model was evidently drawn from countries like Sweden where litigation is relatively uncommon in trademark dispute resolution. Finally, all signers of the gTLD-MoU could participate in a Public Advisory Board (PAB), which would monitor the POC. Participating registrars would have to sign a separate CORE-MoU. The ITU would serve as a depository for both documents. IAHC members expected that CORE would be ready to start service by the end of the year, but a crash effort would be needed to create and test the Shared Registry System (SRS) software on which everything depended.
 

The gTLD-MoU was signed on March 1, 1997 by Heath and Postel, acting for ISOC and IANA respectively. Heath then began working with the ITU to organize a signing ceremony in Geneva at the end of April. That ceremony was critical to the ambition of making the MoU a fait accomplis. The more varied and independent support that could be enlisted into the PAB, the more legitimacy CORE could claim. In the interim, various IAHC members, especially Crocker, undertook a globe-trotting public relations campaign to promote the new system.
 

Amendments and adjustments began to appear nearly right away. Howls of protest led to elimination of the 60 day wait. Complaints from the European Commission led to the end of the 28 registrar limit, thus allowing reduction in the entry fees to $10,000. IODesign's Christopher Ambler, who was by then employed by Microsoft, sued the IAHC, Postel and others, claiming his company had prior claim to .web. He dropped the complaint before a final ruling was issued, but the judge added a statement highly favorable to the IANA's position. Most observers conclude that the court was ready to dismiss the suit, so, by withdrawing the complaint "without prejudice," IODesign retained the right to sue again elsewhere another time. This demoralizing outcome left the alternate registries more divided and in a far weaker position than they had been a year before; an effort to meet in Atlanta and revive their confederation was poorly attended.
 

In late March 1997, at the IETF meeting in Memphis (which is where I first encountered this issue), the "buzz" was that the MoU was designed to satisfy the interests of "big business" interests like MCI, DEC, AT&T, IBM, and UUNET that wanted assurances about the stability of DNS management. If those key Internet functions were to be moved from NSI, they would have to be transferred to a group made up by accountable professionals, rather than a diverse group of overworked volunteers, neophyte graduate students, or ramshackle entrepreneurs. Most IETF members were willing to defer to Postel, Vixie, and the other DNS "wonks" who endorsed the plan. Technical viability was the primary concern, and the SRS was now said to be "do-able." It was left up to the implementors to go do their thing. Most people involved in the IETF's standards making process had learned over the years to accept compromises that were aimed at reducing dissatisfaction by moving forward on "rough consensus," rather than wasting time trying to completely eliminate all complaints and misgivings. This made it easier to get on with more interesting new ideas. Technologically, TLD questions had become rather stale. That part of the DNS was now a policy matter, officially outside the IETF's purview.
 

In the closing days of April the US State Department leaked a memo from Madeline Albright expressing "concerns" about the ITU Secretariat acting "without authorization of member governments" to hold "a global meeting involving an unauthorized expenditure of resources and concluding with a quote international agreement unquote."⁽²⁶⁾ The signing ceremony hosted by the ITU in Geneva on April 29, 1997 failed to generate a groundswell of support that would indicate the MoU had forged the desired level of consensus among Internet stakeholders. ISPs and prospective registrars were slow to join the plan. The only nation-state to sign on was Albania. PSINet, which had initially supported the IAHC, denounced the MoU and called for a global Internet convention with Vice President Al Gore as moderator. Undaunted, the IAHC reconstituted itself as the Interim Policy Oversight Committee (iPOC) which was to manage further institution building processes until mid-October, 1997, when a formally selected POC would begin its term of office.
 

Diplomatic tensions were smoothed over at the annual convocation of the ITU Council in late June. Some member states publicly "regretted" the short notice given regarding the April signing, while others praised Secretary General Pekka Tarjanne for his initiative. The Council Chair, Argentina's Maricio Bossa, was tasked with "carrying out an inquiry into the substance of the MoU and the ITU's role." The US delegate, Richard Baeird was reported to have had strong misgivings about the MoU, but in public announced that even though the US had not endorsed any plan, "the momentum of the April meeting should not be lost."⁽²⁷⁾
 

The undying controversy and a mounting anti-MoU lobbying campaign supported by NSI raised the attention of the Clinton Administration. The first comprehensive attempt by the White House to deal with issue was presented in a paper by Karen Rose, a domestic policy advisor. This alerted other administration officials to a wide range of related technical resources over which Postel exercised authority. J. Beckwith "Becky" Burr (formerly of the Office of Science and Technology Policy), and Brian Kahin became particularly concerned with a new plan developed between the IANA and NSI for the allocation of IP addresses. NSI was hoping to divest itself of the IP number assignment business by constituting a new organization called the American Registry for Internet Numbers (ARIN). Many participants in the DNS controversy considered this to be a separable issue, but Burr and Kahin believed they were inherently connected.
 

The redelegation of authority for allocating large blocks of these critical and finite resources had been progressing in manner far less controversial than the DNS debates, though the issue was arguably much more significant. It is common to think of radio spectrum as a good analogy for IP allocation, especially amid rising awareness of the increasing scarcity of large IP blocks. The U.S. Federal Communications Commission has been selling rights to exclusive control over spectrum at auction, but the NSF was demanding nothing for the IP blocks, even though U.S. government grants had led to the creation of that resource. The disbursed IP blocks had great potential value as a private asset, and quick availability was critical to businesses planning for rapid growth. The power to allocate those resources implied great influence over markets. Randy Bush argued that management of the IP space should be treated as a stewardship and not as a business. Rather than putting the blocks up for auction, the various number registries should only charge fees sufficient to maintain their continued administrative capacities. ARIN, with Postel and Bush as board members, was to be modeled after the Asia Pacific Network Information Center (APNIC) and the Réseaux IP Européens Network Coordination Center (European IP Network--RIPE NCC) to which Postel had already delegated large blocks of address space. ARIN would also control IN.ADDR.ARPA, a significant technical feature of the DNS that was not itself the subject to much controversy and was therefore deemed appropriate to link with the numbers registry. The most vocal opposition to the ARIN/APNIC/RIPE framework emanated from Jim Fleming, an irrepressible gadfly who was pushing a variety of radical technical proposals that did not conform to the IETF standards process. Burr and Kahin were able to delay the remaining transfer to ARIN while they acquainted themselves with the details, but they were derided as poorly informed interlopers by Gordon Cook, a prominent Internet journalist who seemed to harbor particular disdain for Burr.⁽²⁸⁾ The escalating controversy drew in Ira Magaziner, the senior White House advisor for Internet affairs. He had been preoccupied since the beginning of the year with developing the administration's "Framework for Electronic Commerce" which was announced on July 1, 1997.⁽²⁹⁾ This freed him to focus on DNS issues.
 

Events began to accelerate as players jockeyed for position. NSI announced an Initial Public Offering of stock shares which raised about $50 million in capital for future investment and acquisitions. In a fit of pique, Alternic's Eugene Kashpureff exploited a bug in the DNS that allowed him to divert traffic from NSI's InterNIC website to his own, where the captured websurfer would encounter a written protest and a working link back to the InterNIC site. NSI was not eager to publicize the weaknesses of the DNS, or NSI's own security, but after a second event, the company filed a restraining order. Now a mini-celebrity, Kashpureff claimed he had discovered how to black out entire countries from the Internet. He mellowed, however, when the U.S. government opened investigations into wire fraud, and his actions were denounced throughout the technical community as an irresponsible disgrace. Vixie and many others were already angry with him for trying to fragment the root. With this act of unabashed arrogance, he had crossed the line, stealing time from thousands of unsuspecting people, and undermining the "running code" of the DNS. A once sympathetic journalist, Ken Cukier of Telecommunications Week, sent him a message, "Eugene, Nice hack. You asshole!"⁽³⁰⁾ Learning contrition, Kashpureff publicly apologized to NSI and the Internet community, and promised to assist NSI plug its security holes.⁽³¹⁾
 

Also in July, the U.S. government initiated a Notice of Inquiry (NOI), administered by the Department of Commerce's National Telecommunications and Information Administration (NTIA). The primary thrust was to solicit public comment on the expiration of the Cooperative Agreement. The NOI accepted hundreds of E-mailed submissions through mid-August.⁽³²⁾ As would be expected, many of the comments revealed the material interests of the individual or group making the submission, but many also carried a rather grandiose and idealistic perspective. The concept of creating a new structure for global Internet governance was motivating serious flights of fancy, like the following proposal for the opening an Internet Constitution offered by an otherwise staid commercial association:
 
 

An Open Internet Congress was held in Washington, DC that summer along a similar theme, though it was actually a poorly attended front for an NSI lobbying effort focused on denouncing the MoU. Partisans to the controversy continued berating each other online, and even met during the IETF meeting in Munich that summer, but were unable to settle anything. NetNames, a leader in the Mouvement, tried to promote its own conference to reconcile the warring parties, but NSI refused to attend. IODesign offered several times to sell itself to CORE and was rebuffed. All the while, new registrars were signing up to join CORE. Some, like British-based NetNames had years of experience providing registrar services and assisting clients with the peculiar complexities of other national registries around the world. Reputable applicants also included Mindspring, a large American ISP which sought to become a registrar as a way of adding value to its hosting services. But other prospective registrars had little if any background, suggesting that quick buck artists had entered the process and might corrupt it. The MoU had initially stipulated that no registrar should accept pre-registrations from clients, but some had done so anyway, demanding payment for names that were not expected to be visible on the Internet for months. That stipulation was removed so that no registrar would be placed in a disadvantage in relation to the others. In addition, a computerized round robin registration process was designed, so that when CORE went online, each of the registrars would submit names one at a time in turn, until their queues were exhausted. Opportunists exploited this rule by selling priority positions in their queues for non-refundable fees reaching thousands of dollars. Crocker dismissed the sleaze as the inevitable result of creating an open market for registrars. Caveat emptor. He found something positive in that: This supposedly proved that POC was impartial, and had not exerted biased influence over the registrar selection process.
 

Members of the U.S. House and Senate started showing interest by the autumn. Over two days of hearings, during the first live Internet "webcast" from Congress, Postel, Heath and others answered questions about the DNS, CORE and the official expiration of the Cooperative Agreement, now only six months away. Speaking under oath on September 30, 1997, Postel was treated with respectful deference. The witnesses were not sworn on the next day of hearings October 2, however, which prompted a considerably harsher exchange. Andy Sernovitz, the vehemently anti-MoU lobbyist who had organized the Open Internet Congress that summer relentlessly denounced CORE, referring to it as a conspiratorial "Swiss cartel" that was engaged in a "power grab" to "take over" the Internet. During a recess he engineered the release of a report that NetNames had been doing business with Libyan government, hosting the Libyan TLD .ly, and assisting Western companies register within that zone. This prompted a sensational furor when testimony resumed, and Heath was ill-prepared to respond. It was later confirmed that no laws were being broken, but Sernovitz had succeeded in tainting CORE with the image of Moammar Khaddafy. NetNames, frightened by the American nativists, stopped providing service for Libya. Its U.S. offices, incidentally, had always been instructed to refuse to provide registration services for Cuba, North Korea, Iran, Iraq, and Myanmar. Sernovitz's nationalistic appeals started to resonate with some the committee members, prompting the Chair, Mississippi's Charles Pickering, to insist that registrations in .com should be restricted to U.S citizens. Heath was caught off guard by the ferocity of the attacks that day, and was unable to mount an effective response, other than stressing that the U.S. government's continuing delay would impede the growth of the Internet.
 

(A number of significant events occurred in the last months of 1997 and the beginning of 1998. They are abbreviated here.)
 

The Pickering hearings bolstered CORE's opponents, who redoubled their efforts to influence Ira Magaziner. In mid-October the Senate Judiciary committee held hearings focusing specifically on the trademark issue. At the end of the month, Kashpureff left the U.S. to work in Toronto, causing the FBI to suspect he was fleeing possible prosecution for wire fraud. He was arrested by the Mounties, and was held in a Canadian jail there until extradition just before Christmas. The Mouvement continued to formalize its structures, electing David Maher to head POC, and two Internet engineering veterans, Alan Hanson, and Kent Crispin, to head CORE and PAB. The opening round of CORE registrar applications included 89 companies from over a dozen countries. The NTIA report responding to the public comments submitted during the summer was initially due in early November, but was repeatedly postponed as Magaziner, Kahin and Burr continued their consultations with Internet "stakeholders" in Washington and in meetings around the world. That report was finally presented as a Notice of Proposed Rulemaking (NPRM), released in draft form online on January 28, 1998 and published in the Federal Register on February 20. Also known as the Green Paper (because it is not considered ripe enough to be a White Paper), it laid out a plan for recreating the IANA as a privately regulated non-profit corporation.
 

Much to the dismay of the Mouvement, and to the delight of its opponents, the NPRM asserted full authority over the IANA, and did not mention the IANA's relations with ISOC or the MoU. In fact, the IAHC/MoU/POC/CORE/PAB process was not mentioned at all. The NPRM plan for the DNS called for five new registries, each initially limited to serving a single TLD under exclusive management. NSI would be allowed to maintain its control of the three TLDs .com, .net and .org, but, as with the CORE plan, registrar functions would be separated. Some suspected that NSI had anticipated this outcome, since it had already reorganized itself to comply with the NPRM guidelines, creating a new subsidiary called WorldNIC. Two other alternate registries .per, run by Iperdome's Jay Fenello, and Ambler's IODesign indicated their willingness to split off their registrar functions in compliance with the NPRM. One prestigious CORE registrar, Mindspring, defected, and others began hedging their bets, submitting their .web registrations to IODesign.
 

The release of the Green Paper coincided with one of the most widely reported events of the entire controversy. On January 28 Postel initiated what he later called a test of the root system, directing all secondary root server operators (excepting NSI and U.S. government entities) to use his server B for their primary service. He continued to pull (download) NSI's zone file from Server A, so no interruption of normal service occurred, but if he had wanted to add the CORE TLDs into his own NAMED.BOOT file, which defines the TLDs to be made visible on the Internet, it would have been easy to do. The rapidity with which these operators followed Postel's directive was considered further evidence of his ability to command their trust and loyalty. But this faith in him was not universal. Critics charges he had "hijacked" the root. Magaziner learned what had happened soon after arriving in London for a meeting regarding DNS dispute resolution hosted by Prince, PLC. He called Postel immediately and told him to reverse the situation. Postel did so, announcing the "test" was over. Magaziner later commented during the conference, without using Postel's name, that manipulating the root to add TLDs without the U.S. government's permission would be a criminal offense.⁽³⁴⁾
 

It is noteworthy that Postel did not use the word "test" in his initial email, titled "root zone secondary service" but instead referred to a "small step" in the transition of "management arrangements."
 
 

Since the word "test" was never used in the first note, one can only speculate whether Postel might have been thinking about making an omelette without breaking eggs. Transitions of this sort had been major events in the early years of the Internet, before terms like "Web" and "email" became household words. In 1984, adding a hierarchical structure to the DNS under invented names like .com and .net had been a disruptive event, but only among a small group of people, and well within the capabilities of the computer experts who were running the system. Those previous changes provided a great example of what economists call collective action and coordination problems. The adoption of TCP/IP was another example, and the upcoming transition to IPv6 will provide another. Breaking NSI's monopoly on .com without breaking .com itself presented a tremendous challenge. The IAHC and CORE had risen to the task, devising a way, at least on paper, that would have fundamentally overhauled the machinery of the root zone without requiring any need for action or awareness by the vast majority of the Internet's users. The DNS concept first conceived by Paul Mockapetris in 1984 was technically sound enough to enable such a transition in 1989, and could again, but the potential for monetary profit had grown too high for the resources to be redistributed without a fight.
 

Summary
 

What I hoped to make clear in this section was that both sides to the controversy favored market competition, but that they had very different notions of how to evolve this out of NSI's U.S.-sanctioned monopoly. The IANA and the "inner circles" of the Internet technical community held to a notion of stewardship that they believe granted them clear authority to overrule the market-based preferences of entrepreneurs and private corporations. They hoped to exercise this power through a single, non-profit global organization. In trying to broaden the inner circle which had been controlling DNS policy as a benevolent Internet aristocracy, they reached out directly to those organizations most available to lend their project a formal global imprimatur. On the other hand, NSI and a host of new business entrants were able to convince U.S. policy makers that their shared belief in free enterprise and competition mandated a solution which would expose virtually all Internet management functions to the play of the market.
 

Both camps agreed on the concept of fostering open markets, and competitive behavior, but they disagreed on the mechanisms which could best construct this. The Internet's stewards believed the most effective approach was by reinforcing the stabilizing, deterministic hierarchical authority of the IANA over the root. This would provide a reliable way of enabling market behaviors across the rest of the system, while providing greater security against potential deleterious effects: monopoly rent-seeking, market failure of a TLD, and an escalating proliferation of TLDs which might overburden the technical managers of the primary root. They also sought to create an environment for dispute resolution that promised to overcome what they considered to be the limitations and disadvantages of litigation under U.S. jurisdiction. There are material incentives for the veteran Internet community to favor this approach. The cyber-libertarians who have a high profile in that community are engaged in a campaign to diffuse technologies that they believe can be used to construct a liberal order capable of securing property rights without relying on the coercive power of the state. They feel that some central but limited mechanisms like the DNS can be used to propagate those technologies. Moreover, participants in the IETF standards making process come to the organization as volunteers, but they are generally employed by companies which have much to gain from the expansion of the Internet. Switching equipment sold by CISCO, high end computers sold by Sun and DEC, and connectivity services like those leased by MCI, UUNET, and BBN are all increasingly in demand as more people use the Internet to do more things. Not surprisingly, past and present MCI employees are well represented in ISOC and the IETF. CISCO employees are exceptionally prominent in the IETF, and there are rumors (which I have not yet investigated) that CISCO has begun providing funding for ISI--the USC program which houses the IANA. In other words, the more that can be done to expedite Internet expansion, the more these companies stand to gain.
 

Several other confluences of interest helped forge the Mouvement alliance. Trademark interests represented in the IAHC wanted assurances that their investment in brand identities would not be degraded by the ascendance of a new global agency empowered to regulate the assignment of character strings on the Internet. Moreover, the ITU anticipated there would be less call in the future for its services as a venue for intergovernmental negotiations, and was therefore seeking to redefine itself as an entity prepared to serve the interests of transnational business (Fuchs and Koch, 1996; Hamelink 1994).
 

The Mouvement's opponents generally desired to maintain the Internet's stability, but they were more highly motivated by the heteronomous principle of marketization. Thus, the anti-CORE alliance brought together a diverse group of participants. Startup entrepreneurs--the so-called "pirate" alternate registries--had undertaken high-risk investments in pursuit of high-reward payoffs. NSI was jealously defending its existing market advantage. Most importantly, the U.S. government was still deeply influenced by the Reagan-Thatcher legacy of deregulation and privatization. Open market rhetoric was constantly employed in the pronouncements of the "stop the gTLD-MoU" camp. Such themes resonated with U.S. government officials, and were attractive to CATO Foundation allies like Milton Mueller, and academic who became a prominent partisan in the controversy. The private lobbyists employed by this group were also much more highly skilled in their public behavior, and were far more familiar with the ways of Washington.
 

Internet or Americanet?
 

This section will close by briefly discussing the arguments made by the individual who should rightly be considered the most effective CORE opponent--Tony Rutkowski, a former assistant to the the Secretary General of the ITU, and former Chief Executive of ISOC. After leaving the ITU, Rutkowski became highly critical of it, pointing out it had long been in adversarial relationship with the Internet (Malamud 1993) and claiming that conspiratorial plans were being hatched behind closed doors there. Returning to the U.S., he was later appointed first chief executive at ISOC, working closely with Vint Cerf on setting up the organization. Rutkowski now criticizes the ISOC board as a group of individuals who share an unspoken, quasi-religious faith in bits and bytes which he considers to be out of step with the general public. (His own philosophical point of view is highly influenced by fashionable theories of chaos and complex systems. Rutkowski owns the domain name chaos.com and has used fractal patterns to explain how the Internet works⁽³⁵⁾). After leaving ISOC, Rutkowski tried unsuccessfully to move the IETF from under ISOC's umbrella to a new organization.
 

Rutkowski has argued that ISOC and the IETF supporters of CORE are naive, and are being manipulated by Shaw and other "loose cannons" in the ITU who have acted outside the law by issuing the gTLD-MoU as an ITU instrument without the consent of member states. Rutkowski's public demeanor and the quality of his writing is generally superb. He is by far the most skilled communicator who has participated in the "DNS wars" (and he was also one of the most fully responsive sources that I interviewed). He testified before the Pickering Committee, and has met frequently with Ira Magaziner, both in Washington and overseas. Rutkowski's website at wia.org is an essential resource to anyone who wants to explore this controversy. Therefore, it is remarkable that he has taken the surprising position that the Internet is essentially a U.S. domain.
 

Rutkowski has consistently argued in the online discussion groups that the Internet was and would remain primarily an American phenomenon. This prompted occasional discussions regarding host counts, national origin of domain name registrants, relative saturation of third level domains by country and rates of growth of Internet usage in and outside the US. He stuck to this argument tenaciously, even when other CORE opponents disagreed with him on these points. Most observers believe that Internet growth outside the U.S. has already outpaced growth within this country; and that between thirty and forty percent of new registrations in .com, .net, and .org come from outside of the United States. The U.S. market for new user accounts will eventually become saturated, slowing the growth of that base here while it is still accelerating elsewhere. This expectation had much to do with CORE's efforts to promote new TLDs supported by registries outside the U.S. Also, much of the credit for the Internet's exploding popularity in the 1990s should go to non-Americans. One of the first search engines, Archie, was developed in Canada. The World Wide Web was designed by a British citizen working at a European-funded laboratory in Switzerland. And the original graphical Web browser, Mosaic, was created by a Swede studying in the US. Internet Relay Chat (IRC) was developed in Finland, as was Linux, a popular operating system for name servers. ICQ, an important chat technology, was created in Israel.
 

Despite the overwhelming evidence that the Internet is growing as a global phenomenon, Rutkowski stuck to the premise of perpetual American dominance. This was central to his comments regarding trademarks within his response to the NPRM. "The Trademark Dilemma," he wrote, was a minor issue. Why? "Domain names are used primarily for corporate identification and branding in the United States."
 
 
 

He continued, "To call for complex and unnecessary global processes outside the U.S. to deal with what has primarily been a problem among U.S. parties, is little more than a calculated attempt to impede the rapid pace of Internet use and assimilation in the U.S." By asserting that the Internet is primarily a US phenomenon, and then seeking ways to accelerate its expansion here, he excuses the implementation of any policy that would impede the growth of Internet use outside the US. Thus, the Administrative Challenge Panel (ACP) venue created by the gTLD-MoU is deemed inimical to U.S. interests. Terms are defined and policy is structured to induce a presupposed outcome. His agenda seems clearly advantageous and prejudicial to Americans, and would probably be rejected from an equitable multilateral arrangement. Rutkowski's skewed facts may nevertheless appeal to American policy makers who need arguments to justify an Internet policy which is now receiving increasing opposition from other governments. The NPRM's limitation of new gTLDs to five also have been acceptable to trademark interests who consistently favored restricting growth in the use of domain names. The upshot of U.S. policy will be to divert pent up demand for new domain name registrations back into national TLDs. To a small degree, this will indeed inhibit the growth in the use of domain names outside the U.S., as Rutkowski evidently would prefer. CORE's registration policies were considerably more liberal than that used in most countries outside the U.S, and its prices were expected to be lower.
 

One of the most ironic outcomes of the delay in adding new generic TLDs to the Internet is that a few national TLDs have been given over to commercial operators. For example, Turkmenistan, hunting for foreign capital, and blessed by history with the familiar English language designation .tm, allowed Western-businesses to begin selling registrations in that zone. Various tiny island entities like Nueue and the British Indian Ocean did the same, creating open markets for .nu, and .io. All these registries committed to using WIPO's ACP structure to facilitate dispute resolution. It remains to be seen how effective an ACP can be if and when a real dispute is placed before it.

Analytical Framework
 

It is now appropriate to lay out the theoretical parameters which will inform later analysis.
 

To develop an understanding of the social construction of society is to speak about how we establish and maintain practices that coordinate human activity across space and time (Giddens 1991). If one accepts the constructivist argument that rule-based interactivity defines the ontological center of the process by which agents and structures are simultaneously co-constituted, then an understanding of social practice involves speaking about the rules people use to mediate the allocation of resources (Onuf 1989, Dessler 1989, Roberts 1996). This argument grants no prior, separate, or independent standing to either agency, rules, or structure. Furthermore, the properties assigned to identity, the values assigned to resources, and the preferences granted to competing rules are all seen by constructivists as arising endogenously out of interactivity, rather than given exogenously by the environment. Consequently, if one wants to conduct empirical research within the constructivist framework, it is appropriate to focus on rules, providing actual accounts of how they are produced (Prügl 1998 exemplifies this approach).
 

What then, might be the rules of global civil society, if any, and how are they generated? Since constructivism is not especially concerned with globalization, I believe an answer compatible with the framework would consider practices which meet an additional set of conditions: 1) Such practices would be conducted more often by more kinds of people dispersed over more places over time; 2) Such practices would be drained, if not devoid, of a territorially-based national or political social character (Reich 1998), and; 3) People who adopt such practices would be reflectively aware of participating in the making of an emerging global social structure. A corollary to this last proposition is that the spread of globalizing practices would be accompanied by an abatement of localized conventions.
 

Common understandings of globalization are found in the words of Sandra Masur, a prominent representative of the US corporate community, who advocates "across-the-board liberalization of trade in goods, services, and investment." (1991:102, also cited by Rupert:118). But the focus on global marketization--whether Francis Fukuyama's exalted liberal universalism or Robert Cox's triumphant oligopolistic corporatism--doesn't capture the full intentionality that the term "constructed globalism" is intended to convey. Yes, it is true that "the competition state is essential to the globalization process" (Cerny 1996: 136), and that states around the world are lowering trade barriers and loosening regulatory standards to attract capital (Strange). Another common understanding is echoed by Jan Aart Scholte, who characterizes globalization as an "extension" of modernization that leads to the world "becoming one place" (1996: 43). Again, marketization is central to that process. As Scholte puts it, "A certain degree of globalization was in train well before the term was invented"(47). Recent expositions of globalization increasingly point out the need for students of international relations to "pay less attention to boundaries of states and more to the flows and fractures that run across those boundaries" (Dalby 1996: 39). Studies of this sort would focus on production and consumption flows of specific goods and technologies rather than the gross product of particular territories. There is nothing inherently incorrect in this kind of perspective. The objects we generally find at hand in everyday experience originate from a widening number of nations, but these products are increasingly distilled of local character. If "exotic" character is discernable, it tends to be part of a blend, or an artifact of branding and commercial appeal. The study of these objects might indicate where they were manufactured, and can motivate lessons on geography and the use of maps as we track the movement of goods and services, but it does little to help us understand how other people live and exercise their free will.
 

More sophisticated conceptions of globalization look at the "reconfiguration of space time dimensions of social organization" (Rosenau ??). In this vein, Manuel Castells, a critical observer of the Information Age has undertaken a multi-volume study of modern business practices. He sees the global economy as increasingly interdependent, asymmetrical, and selectively diversified. The consequence is "an extraordinarily variable geometry that tends to dissolve historical, economic geography" (1996: 106). More importantly, Castells focuses on the forms of techno-economic change that serve as a precondition for global marketization. What prompts the creation of the mechanisms which enable this time-space compression? If globalization is the intensification of consciousness of the world (Robertson 117), how is that consciousness expressed?
 

The purpose of my third proposition is to emphasize the importance of evaluating how people decide to participate in a global society, rather than simply buy and sell across borders in an international economy. This proposition stresses the reflexively monitored, self-regarded intentionality of globalism. The point of making a distinction between the construction of globalism and conventional understandings of the globalization is to discuss affiliation rather than commodification. One might be tempted to argue that the principle of global trade liberalization satisfies this third proposition: The principle is well known, is widely accepted as common sense (Rupert), and is institutionalized through an array of intergovernmental agreements. The resulting form of rule, however, is primarily heteronomous, constituting a dynamic web of arrangements in which all agents presume themselves to be free and autonomous, but which in fact confines agents by their reciprocal commitments to each other. Under such circumstances, the likelihood of maintaining commitments within a particular interaction is highly contingent on the play of unintended consequences and interruption by other commitments. That is how markets work. They create a world of many places.
 

The stricter test I am applying looks for the global constitution of token values that are stable and organized hierarchically. Such values would correlate to clearly exposed standards of measure and compliance. Openly stated rules of this sort provide a community's citizens and officers with persistent, logical and empirically verifiable ways of coordinating interactivity without needing to stipulate situational contracts. The payoff is to reduce the potential intrusion of unintended consequences and intervening variables into social interaction. Its demand for standard compliance limits peoples' choices with regard to those interactions, but enables it in many other ways. Enablement is fundamental in all concepts of globalization. That is why the modern push for the mobility of capital, labor, and commodities is so often equated with globalization: The reduction of trade barriers between states enables interactivity across borders. Moreover, to the extent free trade is endorsed as a common sense principle world wide, and people tell each other that it should be maintained as a global norm, one can claim that trade liberalization complies with a constructivist test for hegemony. But trade liberalization shows little evidence of hierarchical rule. Violations of trade rules are policed by relatively weak intergovernmental organizations and arrangements, if at all. So reproduction and actualization of any semblance of free trade hegemony depends primarily on heteronomous interactions.
 

My tests for globalized practices can be reduced to the terms ubiquity, non-territoriality, and intentionality. To put it more simply, global practices are those which were designed to work globally and are used that way. This will reveal the kinds of behaviors which perpetuate highly standardized values and centrally organized loyalties that can outlast situational associations. The two classic archetypes of this socially constructed behavior are time keeping and monetary accounting. I intend to argue that the use of mail is socially constructed in similar ways, and that most Internet-based communication is a sophisticated form of mail. This will add to our understanding of what is at stake in the emergence of the DNS as a global addressing system.
 

Global Tokens
 

It is not uncommon to speak of globalization with regard to mechanisms that facilitate modern forms of work and commerce. Clocks serve as the stellar example of technical artifacts which humans use to coordinate their behaviors by referring to a globally operative authority. The mercurial flow of money between most of the world's currencies demonstrates that national borders barely vitiate the fungibility of financial instruments. And the Internet enables the instantaneous exchange of digitized information among a rapidly growing population of users. All these mechanisms easily satisfy the first condition of constructed globalization by demonstrating movement toward ubiquity, but only time keeping strongly satisfies all three.
 

For example, deeply embedded practices such as obedience to Babylonian horology, and the Greenwich meridian are now culturally neutral, and no longer seem tied to any particular imperious political force. One rarely finds any local alternative to the synchronization of hours and minutes which can be displaced, let alone offer resistance (Saudi Arabia is perhaps the strongest site of whatever resistance remains). The use of the Gregorian calendar is not as ubiquitous, but it has finally become standard in Europe and the Americas, and its prominence elsewhere continues to grow. Time keeping clearly meets the third condition, given the diplomatic effort which went into establishing a universally shared system of longitudinal numbering (when the Greenwich meridian was resisted by the Americans and the French), and the sophisticated technocratic effort that goes into supporting accurate time keeping today.
 

On the other hand, despite the ubiquitous presence and the high fungibility of money, almost all of it is denominated by currencies that are issued by sovereign states. Important and powerful institutional arrangements like the International Monetary Fund exist and are able to exert considerable pressure within specific states, but the IMF's power is constrained by the willingness of client states to abide by its restrictions, as well as by the willingness of governing states to contribute capital and expertise. Other global financial organizations, like the privately-funded International Credit Insurance Corporation favored by George Soros are simply proposals that might receive more attention if a worldwide financial crisis gave rise to a rethinking of the current structure. One can speak of institutions which police criminal activities related to global money laundering, but these institutions arise from the self-help activities of individual nations (such as pressure exerted by the United States) or from reciprocal multilateral arrangements. The best evidence for the rise of homogeneity and reflexivity in the use of money may be found in the expanding availability of credit card and debit card instruments which hide underlying currency transactions from the person who presents the card, but these are employed by a very limited strata of the world's population. So money flows globally, and integrated financial markets do often prevail over national monetary and credit policies, but there is no convincing demonstration that money is coordinated by a central global agency.
 

Like the metrics of time, weights and measures were highly refined on a global scale near the end of the nineteenth century. Decimal metric standards are now coordinated through the Bureau International des Poids et Mesures (BIPM) in Paris. Like time keeping, many nations switched to their current system of measure at the moment of traumatic dynastic change. The Soviet Union provides an example of this, though the United Kingdom is a notable counterexample. The world's largest national economy, the United States, does not use the metric system as its primary standard of weights and measure, but the metric system is officially recognized, and extremely precise mechanisms are used to keep the legacy English system and the metric system commensurable.
 

If one wished to evaluate other technically constructed global standards in detail, many artifacts would be available, including workplace norms, market regulation, professional credentialing, identity documents like passports, environmental standards, airport traffic control and air safety, regulation of title such as patent and trademark disputes, and so on. This type of scholarship is most frequently associated with individuals in the orbit of the Society for the History of Technology (SHOT), which has its own school of constructivism. Nevertheless, these researchers have not generally been concerned with policymaking at the global scale.
 
 

Comparison of Globally Constructed Standardized Tokens
 

	Time	Money	Weights and Measures	Mail
Ubiquity	Nearly total (highest)	Nearly total.	Dual Commensurate	Nearly total. (Lowest)
Homogeneity  and Non-territoriality	Total. Use of hours and minutes is nearly universal. Gregorian calendar still advancing, used commensurately with local and cultural calendrical systems.	Mixed. Currencies denominated by sovereignty or multilateral groups. But fungibility between currencies is high.  Low but rapidly growing with regard to information based instruments.	High. Stabilized dual system, U.S. is significant site of resistance to global metric practices, but fungibility of measures is nevertheless certain.	Mixed for physical mail. Some border controls apply, and stamps are nationally denominated.  Low but rapidly growing for electronic mail as user base expands.
Global Reflectivity and Intentionality	High. Vested in professional astronomers paid by host governments, but who make decisions independently.	Low. Various attempts to create unitary global institutions have been attempted without success.	High. Metric system is globally recognized.	Medium. Physical and phone addressing is aggregated nationally, but exposed globally.  Internet expansion and the DNS controversy reflects an effort to aggregate globally.

 
 
 

The table above shows the direction of the next phase of research on this project. Upcoming work will discuss the social construction of addressing systems. This will be integrated with an analysis of debates between the supporters and opponents of CORE over the question of public goods. The MoU had declared as its first the principle that, "the Internet Top Level Domain (TLD) name space is a public resource and is subject to the public trust" (§I.2.1). CORE's opponents denounced this formulation, responding that the Internet's root and TLD space should be treated as a "shared private trust." These competing assertions regarding the disposition of resources will help to clarify the very different strategies that members of the two camps used to promote a common goal of global marketization. I have argued here that CORE supporters advocated what they considered to be a more direct route, through the technical construction of globalism.
 
 
 
 
 
 
 
 
 
 
 
 

Bibliography
 

Braden, Bob. "The End-to-end Research Group: Internet Philosophers and 'Physicists'" Presentation to the 41^st Plenary of the Internet Engineering Task Force. http://www.ietf.org/proceedings/98mar/slides/plenary-braden/index.html
 

Burch, Kurt, and Denemark, Robert A., eds. 1997. Constituting International Political Economy. Boulder, Lynne Rienner Publishers.
 

Castells, Manuel. 1996. The Rise of the Network Society. Information Age, 1. Blackwell.
 

Dessler, David. "What's at Stake in the Agent Structure Debate?" International Organization v4 n3 Summer 1989 pp.443-66.
 

Diamond, David. "Whose Internet Is It, Anyway?" Wired v6 #4 April 1998 pp172-7, 187, 194-5
 

Dalby, Simon. 1996. "Political Geography and International Relations after the Cold War." in Kofman and Youngs, eds. Globalization: Theory and Practice.
 

Fuchs, Gerhard, and Koch, Andrew M. 1996. "The Globalization of Telecommunications." in Kofman and Youngs, eds. Globalization: Theory and Practice.
 

Giddens, Anthony. 1984. The Constitution of Society: Outline of the Theory of Structuration Berkeley, University of California Press.
 

________1990. The Consequences of Modernity. Stanford: Stanford University Press.
 

Hafner, Katie, and Lyon, Matthew. 1996. Where Wizards Stay Up Late: The Origins of the Internet. New York. Touchstone.
 

Hamelink, C.J. 1994. The Politics of World Communication. London: Sage.
 

Kahin, Brian, and Keller, James H. eds. 1997. Coordinating the Internet. Cambridge, MIT Press.
 

Kahin, Brian, and Nesson, Charles eds. 1997. Borders in Cyberspace: Information Policy and the Global Information Infrastructure. Cambridge, MIT Press.
 

Kubalkova, Vendulka, Onuf, Nicholas G., Kowert, Paul, eds. 1998. International Relations in a Constructed World. New York: M.E. Sharpe.
 

Kofman, Elanore, and Youngs, Gillian eds. 1996 Globalization: Theory and Practice. London: Pinter.
 

Malamud, Carl. 1993. Exploring the Internet: A Technical Travelogue. Englewood Cliffs, PTR Prentice Hall.
 

Masur, Sandra. 1991. "The North American Free Trade Arrangement: Why It's in the Interest of U.S. Business." Columbia Journal of World Business. v26 n2. Summer 1991 pp. 98-103.
 

Onuf, Nicholas G. 1989. World of Our Making. Rules and Rule in Social Theory and International Relations. Columbia: University of South Carolina Press.
 

Prügl, Elisabeth. 1998. Feminist Struggle as Social Construction, in Kubalkova, Onuf and Kowert.
 

Reich, Simon. 1998 "Globalization of.. (not sure of title) International Organization Winter 1998.
 

Roberts, James C. "The Rational Constitution of Agents and Structures." in Burch and Denemark, eds. 1997.
 

Rony, Ellen, and Rony, Peter. 1998. The Domain Name Handbook. High Stakes and Strategies in Cyberspace.
 

Rupert, Mark. "Contesting Hegemony: Americanism and Far-Right Ideologies of Globalization." in Burch, Kurt and Denemark, Robert A., eds. 1997
 

Scholte, Jan Aart. "Beyond the Buzzword: Towards a Critical Theory of Globalization." in Koffman, Eleonore, and Youngs, Gillian. 1996. Globalization: Theory and Practice. London. Wellington House.
 

Sernovitz, Andy. 1997. "The US Govt. Is *not* Supportive of gTLD-MoU." July 27, 1997 http://www.gtld-mou.org/gtld-discuss/mail-archive/0375.html
 

Stark, Thom. "What's in a Namespace?" Boardwatch Magazine. May 1997 pp.74-79.
 

Referenced Web Sites
 

The Internet Society. http://www.isoc.org

Rutkowski, Anthony. http://www.wia.org
 

Referenced E-mail Lists
 

domain-policy@lists.internic.net

com-priv@lists.psi.com

domain-policy@open-rsc.org

gtld-discuss@imc.org

newdom@ar.com

ietf@ns.ietf.org
 

Sources

The research for this paper included public and private e-mail exchanges, interviews and discussions with, among others, Tony Rutkowski, Einer Steffarud, Richard Sexton, Jay Fenello, Christopher Ambler, Robert Shaw, Patrick Faltstrom, Christian Huitema, Tim O'Reilly, Dave Meyers, Bill Simpson, Bob Moskowitz, Dave Clark, Don Heath, John Gilmore, Jon Postel, David Maher, Rob Austein, Chuck Gomes, Dave Farber, Alan Hanson, Hugh Daniels, Bill Flanigan, Robert Fink, Susan Harris, Donald E. Eastlake, III, Dave Crocker, Perry Metzger, Karen Rose, Greg Chang, and Erik Fair. I have tried to keep my reports of these discussions and observed incidents faithful to my best recollection, and to draw quotes from written materials when possible. Therefore, all responsibility for any misrepresentations of their views and comments is my own.
 

Craig Simon

University of Miami, School of International Studies.

PO Box 24-8911

Coral Gables, FL 33124

(305) 667-6141

cls@flywheel.com

1. http://www.isoc.org/isoc/mission/

2. A. Michael Froomkin, "Internet/Habermas" unpublished draft, and not yet formally citable.

3. "Overview of the IETF," http://www.ietf.org/overview.html

4. NSF Cooperative Agreement No. NCR-9218742 http://rs.internic.net/nsf/agreement/agreement.html

5. RFC 940

6. My thanks to Robert Shaw and Gosta Roos, Chairman of ISO 3166 Maintenance Agency, for much of this information.

7. RFC 1034, 1035

8. Cricket & Liu Se also "comp.protocols.tcp-ip.domains Frequently Asked Questions," http://www.users.pfmc.net/~cdp/cptd-faq/

9. Wired May 1994.???

10. Newdom Archives 1995q3 http://www.iiia.org/lists/newdom/1995q3/0001.html

11. Newdom ibid.

12. E-mail to Rick@uu.uunet.net ccd to ISOC trustees. http://www.wia.org/pub/postel-iana-draft13.htm

13. Ftp://rg.net/pub/dnsind/relevant/draft-ymkb-itld-admin-00.txt

14. The document was provided to me by the author.

15. ⁰At paragraph 3.C of the Cooperative Agreement. See also, Vincent Cerf, "IAB Recommended Policy on Distributing Internet Identifier Assignment and IAB Recommended Policy Change to Internet "Connected" Status." August 1990. RFC 1174 http://ds.internic.net/rfc/rfc1174.txt.

16. Milton Mueller argues that cybersquatting is not disreputable, but reflects a normal play of market forces which should not be impeded.

17. ⁰Http://www.iiia.org/lists/newdom/current/0233.html

18. Presumably .xxx would attract pornographic content, .nic for Network information centers, .med for medical service providers, .ltd for general businesses, .lnx for users of the Linux computer operating system, and .exp for experimental.

19. Wired article and Matt Marnell

20. ⁰draft-postel-iana-itld-admin-01.txt

21. "Draft Minutes of the Federal Network Council Advisory Committee (FNCAC) Meeting" http://www.fnc.gov/FNCAC_10_96_minutes.html

22. ⁰ISOC Press Release. 1996 "Blue Ribbon International Panel to Examine Enhancements to Internet Domain Name System." http://www.iahc.org/press/press1.html

23. Ibid.

24. Newdom 1995q3

25. "Regions as defined by the World Trade Organization (WTO)" http://www.iahc.org/docs/countries.html

26. ⁰Cited by Andy Sernovitz in "The US Govt. Is *not* supportive of gTLD-MoU." http://www.archive.gtld-mou.org/gtld-discuss/mail-archive/0375.html July 27, 1997 See also Margie Wylie "U.S. concerned by ITU meeting." April 29, 1997. http://www.news.com/News/Item/0,4,10198,00.html

27. ⁰Beaird's title is Deputy Coordinator and Deputy Director of the Bureau for International Communications and Information Policy

28. Cookreport http://www.cookreport.com/

29. "Memorandum for the Heads of Executive Departments and Agencies" http://www.pub.whitehouse.gov/uri-res/I2R?urn:pdi://oma.eop.gov.us/1997/7/3/6.text.1

30. Personal communication.

31. Eugene Kashpureff, "AlterNIC Presentation for ISPCON." Boardwatch October 1997, pp88-91.

32. National Telecommunications and Information Administration, "Comments on the Registration and Administration of Internet Domain Names." http://www.ntia.doc.gov/ntiahome/domainname/

33. ⁰"Comments of the Internet Service Provider's Consortium" August 18, 1997 http://www.ispc.org/policy/dns-comments.shtml

34. Magaziner's statement was discussed on the domain-policy by Tony Rutkowski, who had attended the conference. This was also confirmed for me by Greg Chang of Ira Magaziner's staff.

35. Anthony M. Rutkowski, "Considerations relating to codes of conduct and good practice for displaying material on Internet." 1997 http://www.wia.org/pub/UNHCHR-paper.html

36. http://www.ntia.doc.gov/ntiahome/domainname/130dftmail/03_23_98-5.htm