11.4 Free Expression vs. Controlled Vocabulary | Ruling the Root: Internet Governance and the Taming of Cyberspace

< Day Day Up >

At the heart of the controversy over global rights to names are two distinct and incompatible ideas about domain names and the function of the domain name system. One view sees domain names as a highly flexible naming framework that gives users tremendous freedom to adopt names and naming conventions, and use them to express and advertise messages and identities in a public space. In this view, the DNS protocol is just a framework for coordination. It is the users who autonomously select the names and give them meaning through their uses; the protocol merely ensures that they are unique. The naming regime this produces has no overall organization-it is self-organizing-and offers no guarantees of authenticity. The results are sometimes confusing. But the system as a whole leaves room for creativity and innovation and, more important, is highly responsive to what the broad masses of Internet users want to do with names. It was this freedom, after all, that created the global market for domain names.

The opposing view-the one that animates WIPO and other international organizations, many trademark holders, and national governments-strives to make domain names into what information scientists call a controlled vocabulary. A controlled vocabulary is a system of classification and naming wherein each term has an official and precise meaning. A controlled vocabulary presupposes an authority with the ability to make binding determinations as to what names are associated with what entities. The Library of Congress index or scientific taxonomies for classifying plants or chemical elements are examples of controlled vocabularies. As the examples suggest, controlled vocabularies can be extremely useful for a specific purpose. They are also rigid and constraining, and cannot be used successfully for anything other than the purpose for which they were designed.

Which approach to domain names-coordinated free expression or controlled vocabulary-is better suited to the Internet? I argue in the following sections that the DNS protocol answers this question for us. The DNS is a system of coordinated free expression; it cannot be made into a controlled vocabulary without drastically altering its functions.

11.4.1 Seven Deadly Assumptions

The effort to turn domain names into a controlled vocabulary is founded on a series of assumptions about how domain names are used, what they signify, and how they are interpreted by ordinary Internet users. Those assumptions are given in the following list. (The supporting commentary and footnotes refer to legal decisions and statements that show that these assumptions are widely held and commonly asserted.)

The DNS is a directory. This assumption posits that the purpose of DNS is to guide users to specific kinds of content, Web sites, or services. As a corollary, end users search for what they seek on the Internet primarily by consulting lists of domain names or by guessing domain names.
Authenticity. Domain names are (or should be) 'authentic.' To possess a domain name is to posses an official, authorized relationship to the named person, place, organization, or thing. A stronger form of this assumption holds that for any given name, it is possible to know which applicant has the most valid claim to it.
Hierarchy doesn't matter. Domain names are not really hierarchical. It does not matter whether a character string is registered under .com, .to, .net, .org, .blat, .xxx, or anything else. A name must be protected in all top-level domains; otherwise it has no meaningful protection at all.
Nonuniqueness. Domain names need not be unique. If a registered name looks something like a name that someone has rights to, including misspellings or words in combination with a trademark, then it ought to be held by the rights holder, or at the very least, not held by someone else.
Domain names are trademarks. Every domain name points to an ecommerce Web site, an offering of goods or services. Domain names are not used to express ideas or refer to things.
Domain names strongly influence content interpretation. Internet users' interpretation of what they encounter on the Internet and the Web is closely linked to the semantics of the domain name. Thus, if a domain name address leads users to information or content different from what they expected to find, they will be hopelessly confused. As a corollary, in adjudicating domain name disputes the actual content of a Web site is not as important as an analysis of the text of the domain name itself and whether it can be construed, in isolation, as somehow impinging on the scope of a mark.
Global visibility. The mere registration of a domain name guarantees the registrant a substantial public audience. The name or site does not have to be advertised or promoted to have a significant impact; indeed, it does not even have to be visible on the Internet or associated with an operational Web site or email account. Millions of users will spontaneously type the name into their browsers, without any prompting or advertising.

All the preceding assumptions are problematical. Many are simply false. Some are half-truths, while others stand in direct contradiction to how domain names function technically. Taken together as a package, they constitute an attempt to reconstruct domain names into a controlled vocabulary.

Consider, first, assumptions 1, 2, and 6: that the DNS is a directory, and the purpose of the directory is to steer users to officially sanctioned information correlated with the name. This set of assumptions is the most fundamental one behind the push to make DNS into a controlled vocabulary.

It is embedded in many court and UDRP decisions. The second WIPO report boldly states, 'The placing on the domain name register of a distinctive name, such as gretagarbo.com, makes a representation to persons who consult the register that the registrant actually is, or is associated with, the person whose name is registered and thus is entitled to use the goodwill in the name' (WIPO 2001, para. 139).

This view of domain names is fundamentally inaccurate. The WIPO statement, for example, contradicts what we have already established about users' adoption of identities on the Net. The many AOL users who chose some variant of the name Greta Garbo are making a statement about themselves -their personality, likes, and dislikes-not representations to others that they are Garbo. Nor is it likely that many users who see the name interpret it as such, given the context.

Moreover, the theory that the DNS is an authoritative directory reveals a basic ignorance of how the protocol actually functions. People do not find things on the Internet by 'consult[ing] the register of domain names.' The domain name 'register' consists of resource records scattered around half a million name servers in different parts of the planet. To compile and consult that list, one must pull out zone files using complicated software, and the resulting list would consist of nearly 35 million second-level domain names; the .com zone file alone would contain over 23 million. That simply is not how ordinary users find things on the Internet. The notion that domain names are used for 'searching' confuses searching techniques with locators, two completely different functions.

When users type in a domain name to locate a site, it is usually because they already know the domain name and the nature of the site they are headed to. That is, they are using the domain name simply as a lookup tool. Although some users try to find sites by guessing an organization's domain name, this is done as a last resort after other methods have failed. The vast majority of users rely on search engines and portals. They locate content through hyperlinks that they receive from email or see on other sites. They bookmark links in their 'favorites' file. Or they copy down or remember specific names that they have seen advertised.

As for assumption 6, the words 'Greta Garbo' typed into the popular Google search engine bring over 38,400 hits. Interestingly, none of the top ten listings returned by Google have domain names that include the labels 'gretagarbo,' 'garbo,' or 'greta.' The URL that arrives at the top of the heap is <http://www.mdle.com/ClassicFilms/FeaturedStar/star53.htm>, a tribute to Garbo put up by a fan club for silent movies. ^{[18 ]}Similarly, on Yahoo! and HotBot, the most highly ranked content on Garbo is under domain names like netcomuk.co.uk, home.hiwaay.net, or bombshells.com. In the majority of cases, there is no correlation between the content of a Web site and the semantics of the domain name. At gretagarbo.com, on the other hand, ownership appears to rest in the hands of Garbo's heirs or licensees. At that site one finds a rather slow and poorly organized site selling jewelry. Although the connection to Garbo is 'authentic,' is it valid to assume that anyone using 'Greta Garbo' as a keyword for searching, is looking for that particular line of jewelry?

Users who employ 'Greta Garbo' as a keyword may be interested in communicating with other people who are fans of Garbo. They may want to buy a book about her, find a picture of her, or find out which retail stores sell copies of her movies. For all we know, a user may be trying to find out whether MTV has produced an episode of Celebrity Death Match (a cartoon using animated clay figures) that pits Garbo against Madonna. Given what we know about the Internet and the incredible variety of content and materials available there, it is presumptuous to claim that we know what people who type names into their browsers are looking for. There are at least as many different objectives for searches as there are searchers.

Consider, next, assumption 3 on the list. It is a fact that DNS names are hierarchical. Nevertheless, the assumption that the semantics of the top level do not matter is becoming an increasingly common part of the jurisprudence of domain name law and the UDRP. If one has a legal right to a name in one TLD, the theory goes, that right should extend across multiple TLDs, because users cannot be expected to differentiate among different top-level domains.

WIPO used this argument to support its policy of name exclusions for international organizations. A special top-level domain, .int, is reserved for legitimate international treaty organizations. WIPO recognized that Internet users 'can have reasonable confidence and trust as to the genuine identity of the organizations registered in .int, and of the validity of the information provided by those organizations' (WIPO 2001, para. 102). The WIPO report also argued, however, that the mere existence of valid registrations in the .int domain is not sufficient because abusive registrations can still take place in other top-level domains. In essence, WIPO is arguing that the top level of a domain name doesn't matter. The same assumption shows up frequently in UDRP cases and domain name litigation. In one well-known British case, a judge took away the bt.org domain name from speculators and awarded it to British Telecom even though British Telecom already had the bt.com domain and the acronym BT could be used by many different legitimate organizations.

The assault on hierarchy is now being pushed into the second and third levels. The second WIPO report, for example, argued for excluding country codes from the second level on all new top-level domains, because users are unable to distinguish between domain names like company.uk.com and company.co.uk. And some trademark hawks are beginning to seek to assert rights in third-level delegations.

Consider, next, assumption 4. Uniqueness is the most significant requirement of domain name assignment under the standard protocol. But to DNS, 'unique' means any difference in a character string that can be recognized by a machine. Uniqueness to a machine is not the same as differentiation by a human being. People might use any one of several different names to denote an organization, idea, or product, and they may not be able to distinguish between different spellings of the same word. In response to this problem, many brand holders have attempted to register every possible permutation of their names, multiple misspellings, as well as domain names that include the trademarked term along with generic terms, such as fordcars.com, fordmotors.com, ford-source, and so on. Both UDRP panelists and courts have often upheld their right to reclaim such domains.

Here again, however, the desires of trademark owners are fundamentally at odds with the nature of DNS. The protocol allows any unique character string to be registered. The giant telephone company Verizon, formed via a merger of GTE and Bell Atlantic, learned the futility of resisting DNS's reliance on uniqueness. Just before announcing its merger and new name, Verizon purchased close to 500 domain names, including not only verizon.com and verizonlongdistance.com, but also verizon-sucks.com, and several misspellings of the brand. Later, the publishers of the hacker magazine 2600 tried to register verizonsucks.com to operate a site for consumer venting. Upon discovering that the name was not available they registered verizonreallysucks.com. The humorless telephone company sent them a cease-and-desist letter accusing them of trademark violation. Undeterred, the 2600 group went on to register VerizonShould SpendMoreTimeFixingItsNetworkAndLessMoneyOnLawyers.com.

The point of this story is that it is impossible for a company to prevent someone from incorporating its name into a domain name in some way. Registering a few common misspellings (or using the UDRP to recover them if they have been registered by others in bad faith) makes some sense. But the DNS supports too many variations to make it possible to preempt criticism or capture all possible references to a company or product. Any attempt to protect massive 'clouds' of names will be pointless unless draconian and undesirable restrictions are placed on the use of DNS.

All this assumes, of course, that the possession of these domain names is important and valuable. Here, too, the case for a controlled vocabulary is based on highly questionable premises. As noted, the idea that the majority of Internet users find their way around the Internet by typing hundreds of different variations of domain names into their browses flies in the face of everything we know about user searching behavior. Contrary to assumption 7, the registration of a domain name is no guarantee that a significant number of users will be attracted. Popular Web sites that make money require expensive promotion, high-quality content, lots of links from other sites, and good word of mouth in the press and among users. What evidence we have suggests that simple, generic terms in the .com space do generate traffic, but there is also ample evidence that that type of random traffic by itself cannot sustain an online business. ^{[19 ]}

Controlled vocabulary advocates also assume that users who type in the domain name of a company and find something they did not expect-say, a protest site rather than the company-will not be smart enough to look elsewhere. They will become completely diverted and lost to the company forever. This notion is implausible on its face. It is like saying that someone who has incorrectly dialed a telephone number will not correct the error and redial.

^{[18 ]}The domain name in this URL, mdle.com, refers to M. David Lewis Enterprises, an organization that has no official relationship to Garbo or her estate.

^{[19 ]}Cecily Barnes, 'Catchy Domain Names Lose Their Luster,' CNET News, October 16, 2000.

< Day Day Up >