IAHC-Draft David Collier-Brown Category: Informational Private Person Expires May 30, 1997 November 1996 Forms of Top-Level Domain Names: An Appreciation. Status of this Memo This document is an IAHC-Draft. IAHC-Drafts are working documents invited by the Internet International Ad-Hoc Committee. IAHC-Drafts are draft documents and may be updated, replaced, or made obsolete by other documents at any time. It is inappropriate to use IAHC Drafts as reference material or to cite them other than as ``work in progress.'' Introduction This is an appreciation of the situation [1], a structured document setting out relevant facts and courses of action to consider in considering the form new top-level domain names can take. The aim of this paper is to provide in brief the material factors and plausible courses of action which anyone setting out to debate the form of future top-level domain names will need to consider. The paper is limited to discussing issues centered around the form of domain names and their interaction with trademark and trademark litigation, as that is the significant non-technical problem faced by the community at this time. It is the intention of the author to advance discussion, not to exclude or propose particular courses. This paper does not attempt to provide evidence for a decision on whether the names should be in the root domain (making them top-level names) or in some domain such as .com, making them second-level names. The community seems to favor top-level names, but there has not been substantiative debate on this point. Factors Trademarks The primary factor in this discussion is the issue of trademark on names which organizations wish to use as domain names. There is a rich history of jurisprudence on the subject: for our purposes, a trademark holder which wishes to establish a presence on the 'net may wish: a domain name identical to one or more of their trademarks (referred to in the rest of the document as [identical]). protection against dilution of their trademark from the use of a similar name as a domain name by another party on the net (referred to as [dilution] below). Companies registering domains in the United States have evi- denced such desires, with numerous lawsuits currently underway. This is exacerbated by the conflict resolution mechanisms used by NSI, the U.S. InterNIC operator: it should be noted that this problem has not yet become significant in any of the national top-level domains. A further variant is that some trademark holders, notably of very well-known names, have expressed an intention to register their name in all applicable domains, to proactively prevent infringement or dilution in any context. This is a known problem in the allocation of 800 numbers (referred to as [all domains] below). Inferrability A secondary factor is the desire of the companies registering domain to have their domain names ``guessable'' from their company names (referred to in the following as [guessable]). For example, a company named ``SeaTech'' might wish the domain ``seatech.com''. With a name of this form, persons made aware of it would assume that it was doing business throughout North America (and indeed the world), and that one could send mail to @seatech.com and find their web page at www.seatech.com. If the company does not have a short, unique name, it is to their advantage to have a name which can be found by a domain name system or web search. For example, the American Airlines web page can be found in a few seconds as ``www.americanair.com'' by an Alta Vista search for ``american airlines''. Courses of Action 0: Do Nothing We could, of course, do nothing. This would leave us with a problem in the .com domain, and place the responsibility for solving it on the NIC operator. Based on their lack of success to date, this is probably inadvisable. (Course 0 is always ``do nothing'', and is conventionally included to ensure that all reasonable courses of action are considered, including a failure to make a decision, which is implicitly an unintentional decision to do nothing.) 1: Revert to First-Come, First Served We could revert to allocating domain names in .com and the other existing domains to the first person requesting them, without other conditions. This would not directly address either of the above problems, but would remove the NIC operator from the center of the debate, as they could merely disclaim responsibility for trademark problems, require the parties to go to court, and agree to abide by the result. In the case of either identical or very similar names, the trademark holders would have to ``fight it out'' in court. [identical][dilution] As there would only be a single international .com domain (and some similar national domains), the ability of a company which holds a trademark in the airline business to distinguish itself from one which holds the same trademark in the trucking business would be limited. Litigation would presumably continue in the international domain and to a lesser degree in national domains. Companies desiring a presence in all domains would need only to establish a single international domain, in .com, and might attempt to establish domains in national domains, .edu and/or .org [all domains] Company's domain names would be in one sense hard to guess, as there is no rule about which trademark holder would get the domain in .com. In another sense, they would be easy to guess, so long as you wanted the site of the winner of the race to register. Other trademark holders might elect to add distinguishing words to their domain names, such as ``americanair'' or ``allied-vanlines''.[guessable] There would be little or no effect on domains looked up via search services. 2: Synonyms We could establish more domains in parallel to .com, such as .biz, an existing experimental top-level domain. Companies desiring the ``same'' name as another could simply register in .biz as opposed to .com, so that allied.biz would be united vanlines and united.com would be united airlines.[identical] This would not address either the trademark or the dilution problem, however: there would be ``two'' uniteds, with nothing to distinguish between them in the mind of the customer save which synonym for ``business'' they were registered under. It is likely that at least one allied would be displeased with this lack of differentiation and ask for relief on the basis of misuse or dilution of their trademark.[dilution] Companies inclined to defend their trademarks strongly would attempt to register the same domain in all synonymous domains. This would be a brand-new point of contention between trademark holders in different businesses.[all domains] Company names would be hard to guess, as there would be a component of the domain name which would not be predictable (which synonym) and which would have to be memorized or looked up by the customer.[guessable] There would be little effect on domains looked up via search services, and more people would find it necessary to use such services. 3: Meaningless Names We could establish domains whose names were meaningless: for example, we could have .AAA, .AAB, .AAC and so on, and allow or coerce companies to register in randomly-chosen ones. Companies desiring the ``same'' name as another could simply register in a different domain, so that allied.aaa might be united vanlines and united.bbb might be united airlines.[identical] If we allowed companies to register in the one of their choice, many companies would choose domains meaningful to them, with many desiring to be in the first, last or most meaningful domain for theri business based on the accidental combinations of characters. For example, we might see particular companies preferring to be in .SEX or .XXX. This would not address the dilution problem, however: there would be ``two'' uniteds, with nothing to distinguish between them in the mind of the customer save a random string. It is likely that companies would be displeased with this lack of differentiation and ask for relief on the basis of dilution of their trademark by someone in another domain.[dilution] Companies inclined to defend their trademarks strongly would attempt to register the same domain in all random domains. This would be (like .biz/.com) a new point of contention between trademark holders in different businesses.[all domains] Company names would be hard to guess, as there would be a component of the domain name which would be random and which would have to be memorized or looked up by the customer. [guessable] As before, there would be little effect on domains looked up via search services, and more people would find it necessary to use such services. 4: Meaningful Names We could establish domains whose names were meaningful business categories drawn from existing trademark categories established by national bodies or international agreement. For example, we could have .soap, and names like lever-brothers.soap for the cleaner-related branches of Lever Brothers. This is also called a controlled vocabulary, and is commonly used in library searching and categorization (eg, the U.S. Library of Congress system). Companies wishing to have identical domain names, if in different categories, would simply register their desired name. We could have american.buslines and american.airlines without difficulty. If the categories are over-broad, we would have problems with clashes, such as if bus-, truck- and air-lines were all grouped together under the .transport domain.[identical] Companies fearing dilution would suffer directly only if the categories were over-broad. They could suffer dilution if the general public was inclined to assume that the same ``united'' operated united.vanlines, united.airlines and united.buslines. This appears far-fetched. [dilution] Companies desiring to protect their trademark by defending it broadly would find it difficult and expensive to justify registration in a very large number of arguably inappropriate domains. If the bozotron company wished to proactively defend its good name by registering bozotron.soap, while not manufacturing soap, there would be room for conflict if a soap company with bozotron as a trademark desired the bozotron.soap domain. This in turn implies that the company attempting to protect its trademark would have to ensure no-one else was able to register a trademark in any category, a significantly difficullt taske even in a single jurisdiction.[all domains] Domain names would be relatively guessable, given a small enough set of categories and some experience with the system by the customer. Persons unsure of the business categories would need to look those up with a search service. Persons who do not know the business's category, only its name, would have to use search services, as above.[guessable] 5: Obfuscated Meaningful Names We could establish domains whose names were meaningful business categories, but refer to them by the category number, thus giving us C10, C02 and so forth. This has a result indistinguishable from random names, save the names will not chance to match meaningful short words. If the obfustication were progressively less effective, such that categories could be inferred to increasing degrees from the names, it would come more and more to resemble meaningful names. 6: Unconstrained Names We could establish domains with unconstrained names: any organization could propose a name and have it made a top-level domain. This is in fact the former case within .com, with the limitation that only nominally commercial entities could be registered. Films were seemingly taken to be businesses, however, so the limitation is not absolute. This has a result indistinguishable from reverting to first-come first-served in com, save that the load on the root name servers would be remarkable, as most enquiries would have to go to them. 7: Partially Unconstrained Names We could allow names to be drawn from a vocabulary which is constrained only in that it distinguishes different kinds of business. This might start with the existing categories and subcategories established in international trademark law, and be expanded by reference to the literature of trademark with more and more precise terms. For example, transportation might be an initial category, from which companies might well ask that trucklines, buslines and airlines be added, sufficient to distinguish their businesses. Companies doing business in several specific subcategories would have the option of registering in each (x.trucklines and x.railways) or in the parent category (x.transport). Companys seeking to distinguish themselves would ask for registration in the category that most accurately describes their business, to avoid being mistaken for another, like-named business. [identical] Companies concerned with dilution might choose more specific names, to avoid anyone mistaking them for someone else. In effect, the category they choose would become closely associated with their names (eg, the Unix[tm] Timesharing System) and would appear to the customer to be part of the name.[dilution] Companies wishing to defend their trademarks strongly could regard the above as sufficient. If not, they have the option of trying to register their name in all domains where they believed there was a chance some-one might use the same name. This would be expensive and time-consuming, though: the International Olympic Committee might well find itself faced with registering olympic.restaurant.brampton.on.ca in one known extreme case. [all domains] Names would be guessable, but less so than with a more controlled vocabulary. A search service with a knowledge of the requisite synonyms would be straightforward to provide, however. [guessable] 8: Combinations Finally, we might elect to try combinations of courses based on their relative strengths and weaknesses. It is plausible to have constrained or partially constrained names with a special category for companies who are both inclined to protect their trademarks strongly and who are internationally known. If Ortel found it necessary to defend its name against all similar names (eg, Nortel), then it might ask for a higher-level domain than all other companies. In the root if necessary, or in .com if all the categories were to exist within .com. This would presumably be of interest primarily to the owners of ``famous'' names, such as Disney. Conclusions The author invites the reader to choose from among the above, to try to add new courses or to work out the undiscussed issues that would arise from choosing any of the above. He positively disclaims having a conclusion. Yet. Appendices Appendix A: Appreciations of the Situation This is a document used by the military to teach junior officers ``to use their heads for more than hatracks''. Writing appreciations is taught as a means of clarifying one's thoughts about difficult problems. An appreciation has 1. a single aim, stated without the use of ``and'' or ``or'', 2. limitations, which may constrain the problem 3. factors, each of which affects at least one course of action 4. courses of action, each of which is affected by some or all of the factors, and 5. a conclusion. Readers are encouraged to look for missing relevant factors, missing courses of action, factors which are not used by any (or some) of the courses and courses which do not follow from the factors which supposedly affect them. Dave Collier-Brown / davecb@hobbes.ss.org