Domain Names, Trademarks and Taxonomies

Disclaimer

This problem, independently conceived, was addressed in RFC 2352.

Background

The Domain Name System translates between host names (such as moglen.law.columbia.edu) and IP addresses (like 128.59.177.251). One registers a domain name (columbia.edu) with the appropriate registry, sometimes through an intermediary, usually for an annual fee. Domain ownership is a sort of property, although it might also be thought of as a lease: the owner (tenant?) of the domain has the right to "occupy," exclude others, subdivide, and sell/assign.

The Problem of Exhaustion

Domains must be unique; there can be only one pizza.com. Fortunately, .com is not alone: it is just one Top Level Domain (TLD). .net, .edu, and country domains (.ca, .us, .cn...) exist. However, because the .com TLD is pre-eminent and so closely associated with the Internet, a non-.com site will not resonate. Commercial sites in .biz or .info face consumer scorn and forgetfulness. When have you seen a Super Bowl advertisement for a .biz?

Exhaustion may not naturally occur soon, but the practices of "cybersquatting" and "domain tasting" may bring an artificially early end to the .com namespace.

Categorizing the Problem Away

The .biz domain is the worst example of attempting to create a new .com. Although .biz includes safeguards for trademark owners, .biz and .com's markets overlap. This creates yet another name for big corporations to buy, as anticipated in RFC 2353 §3(b).

Another approach is to create more generic TLDs, such as .museum and .aero. However, with the search-based navigation popular today, .museum looks odd: why re-train patrons to type danish.national.museum instead of "danish national museum?" .aero has great potential for usefulness because the wide use of IATA airline and airport codes: business travelers might like to be able to type jfk.aero or ua.aero (not registered). However, not all airlines and airports are convinced and .aero coverage is spotty.

The Trademark Wrinkle™

The .com domain, although not under .us, is de facto a US domain and hosts frequent run-ins with US trademark law. Registrars now follow the UDRP, which attempts to resolve disputes against bad-faith users and in favor of trademark holders. As a result, .com namespace overlaps trademark namespace. However, trademarks have features unlike domain names. [US] trademarks must be used in commerce: one cannot speculate. Similarly, trademarks are granted for "goods and services"--a situation which might be mirrored in domains by the introduction of new TLDs (with the problems noted). RFC 2352's assertion that "all names are unique" is not true. Even using a .tm.us system (where the US trademark holder for X gets x.tm.us), there are multiple legitimate claims to delta.tm.us.

"Skadden" is a trademark for legal services. "Skadden Stairs, Inc," a hypothetical escalator maintenance company, may still go by "Skadden." This restriction would only work with DNS if .com were to be abolished (or reserved for conglomerates) and specific domains (.lawfirm?) introduced. This system has many advantages: the owner of nissan.com would use nissan.computers and leave nissan.cars free. Aside from the loss of brand equity (a taking? is this really property or even a state actor?), overhead costs, and mass user confusion, what would be wrong with getting used to yahoo.search or moglenravicher.lawfirm?

Domain names are not always commercial. I own three domains, one of which is used in (negligible) commerce. I could be punted to some version of .name, with its own exhaustion problems, or to a .personal or .homepage. If .com were reserved for commercial use, current .com denizens might undergo token monetization. I once sold a t-shirt for one of my domains on Cafepress: that is entrepreneurial enough. The current bad-faith provisions of the UDRP should prevent this from being abused. I think most problems (except for users) could be worked around, but how can you make such a drastic change when some still cannot use the search box?

Going Local

A subdivided namespace has met some success in the use of country domains. When operating Skadden Stairs in Tuvalu, skadden.tv is yours. Of course, .tv and .to, among others, provide an example of another problem (country codes being usable as English-language generics or typo-squatting as .cm. Sub-categorization has been attempted in the .us domain, and is often used for schools ( ... .county.k12.state.us, combining both types of classification).

Businesses did not flock to obtain .city.county.state.us names before the .us overhaul. Would you go to hamdel.ny.ny.ny.us for lunch? How about hamilton.deli.ny.ny.ny.us? Even the less-unwieldy international-business-machines.co.ny.us suggested by the RFC editor is unacceptable. Any classification which does not eliminate generic .com names is doomed by .com’s primacy, and big corporations will not want to give up the .coms they bought for millions.

Abandoning the DNS?

Can we abandon the DNS? With increased use of search, I should be able to just type "delta airlines." The details are irrelevant: maybe the server would be at delta.airlines.atlanta.ga.us, or maybe it would be at a unique string of symbols. If I type in "delta," I should get a disambiguation page or be directed to the one I am most likely seeking (according to history). An unwieldy unique identifier, however, could make citation difficult or impossible.

A method requiring only sufficient uniqueness to disambiguate the result fails for new businesses: "zombo" would no longer be unique after Zombo Air takes flight. E-mail and other non-interactive services would require adaptation, and any such change would run up against myriad obsolete references.

Conclusion: Boldly Doing Nothing

In the end, it may be best to do nothing (or close to nothing) and let .com continue toward exhaustion. Some anti-speculation reform may help postpone this, although UDRP and its reliance on findings of good and bad faith provide a practical limit. By the time the .com namespace is exhausted, there may be a Web successor. All the alternatives above would cause too much pain to users and entrenched interests. Meanwhile, third-party DNS services can experiment with new taxonomies. If a new method proves popular, users can adapt to it at leisure and the root servers might find themselves out of work before .com fills.

also available: previous version and comments