I see that Steven beat me to the punch with his post on James Gleick's New York Times Magazine article on namespace clashes. Gleick has some good points, mostly variations on the meme that the Internet undoes any notion of locality. Still, a few contrarian thoughts occurred to me while reading it, and I'd be happy to hear from anyone who can show me where my reasoning is going wrong.
First, some of the small-namespace issues that Gleick describes are self-created. Drug names make for a good example. This is not a namespace that is intrinsically crowded. If we were willing to give drugs names like "Frederick Terwilliger" or "Antidepressant Charlie Romeo Foxtrot Victor" or "Flatuenza," there would be no scarcity of drug names. The only reason we have sound-alike pairs such as ZANTAC/XANAX is that drug companies have decided to give drugs names that draw upon only a small subset of the English phonemic arsenal. Their reasons for doing so are mixed. Names with obvious meanings are out, for legal reasons (the FDA probably wouldn't let you call your anti-depressant "Cures-All"); names that don't sound pleasant ("Flatuenza") are out for marketing reasons. In any event, this party isn't packed because the room is too small, but because everyone is huddled together over by the DJ.
Does this observation have any policy implications? I dunno. I don't know whether it cuts in favor of broad or narrow protections and zones of exclusion around particular trade names. But I guess it suggests to me that the problem of namespace collisions is not always a stochastic problem. The ill-fated meeting of Ratan N. Tata and Bodacious-Tatas.com is one of those great examples of cross-linguistic irony that makes legal onomastics fun. But sometimes collisions happen because everyone wants a piece of a particularly nice word. "Apple," "Madonna," and "Crazy Horse" -- I'm not at all surprised that these combinations of letters have had serious collisions, whereas "Flatuenza" hasn't. Maybe law and policy should treat these attractive combinations differently than they treat the unattractive ones.
Which leads me to my second thought. We've long since learned that a one-size-fits-all single-ownership regime is a mistake for many words. Trademark has (or, at least, used to have) decent notions of splitting up a mark among different users: if the uses didn't overlap geographically or in product lines, they could coexist. When it comes to domain names, at least, it might be that insisting on a single "owner" of a name is a mistake.
No, seriously, hear me out. Yes, a given domain name can really only belong to one registrant: that's a precondition purpose of having a properly useful DNS lookup system. But if two contenders both have reasonable claims to a phrase, why not force the one who gets the actual domain name to include a prominent "if you were looking for . . ." link? For ease of administrability, say that whoever gets the domain first can hang on to it, and must include a link upon demand of anyone with a demonstrable protectible commercial interest in the name. Cass Sunstein would be proud. The truly bad faith cases -- not the legally fictitious "bad faith" of the UDRP but the real life "bad faith" we associate with low-down rotten no-good swindlin' varmints and typosquatters like Amazoon.com or sellers of fake Rolex watches -- you could still have a UDRP-ish process for, but you'd start with a strong presumption in favor of the existing registrant. A very strong presumption.
Yes, this is giving into the typosquatting terrorists; more of them will be able to hang on to their bogus domains. But this just ties into my third thought, which is, so what? As I keep on thinking while reading about DNS controversies -- UDRP, SiteFinder, new TLDs, etc. -- fighting over domain names is fighting yesterday's battle. Search keywords are what matter today; domain names are just the human version of bookmarks. Even for web sites I think I remember, I type the relevant keywords into Google rather than into my address bar. The SiteFinder/IE technique of doing a search on failed DNS lookups is doing things exactly backwards -- I'd rather feed things into a search engine and then go to the domain name if the search engine confirms that I guessed right. Nor, from my observations of others at work and play online, do I think I'm alone in my preferences.
I'm not claiming that I know how to handle trademark disputes where search engines are concerned. (That's a fascinating question, one to which I've devoted plenty of mental cycles, without much to show for it. It mixes up First Amendment claims, economic efficiency arguments, rules of thumb about technical architecture, some particularly tricky jurisdictional wrinkles, and a whole mess of cross-cutting legal interests. It may well be the Gordian Knot of Internet law. Or perhaps in a couple of years the solution will seem obvious and we'll have a good laugh about how confusing things looked from our silly 2004 perspective.) Instead, I'm just saying that the rise of a search culture makes domain names less important, just as the rise of a domain name culture made IP addresses less important. A domain name, after all, is an arbitrary signifier slapped on to some server. The name could be anything, any word or phrase at all -- semiotically, all that matters is that the linkage from name to server work. Of course, this relationship looks an awful lot like the relationship between trademarks and the goods to which they refer -- the mark itself can be arbitrary.
But we have tons of experience from the trademark world just screaming at us that this arbitrariness is the key to why trademarks do and should work. Indeed, it's been quite reasonably argued that only the purely arbitrary aspect of a mark should be protectable. From this perspective, there's no particular reason why the arbitrary connotations at work in domain names should have to line up with the arbitrary connotations at work in trademarks. They're arbitrary, aren't they? There seems to be no intrinsic reason within the logic of trademark that compels the conclusion that the right to own all domain names that are similar to your mark is part of the bundle of rights that using your mark in commerce wins you. We could instead argue that domain names are just another kind of "commerce," like selling sugar, or books, or rapid document delivery, and that only domain names that are confusingly similar to someone else's domain name are verboten.
But even that's a digression. To put my position in perspective, I come down on the side of live-and-let-live when it comes to domains not so much because I want to peel back trademark rights (although often I do) but because I don't really see the deep and eternal harm in not having the precise domain name you want. That's why I don't care about opening up the TLD namespace (bracketing the tough issues of governance and ICANN/government/private control, that is). In a search engine age, the difference between flatuenza.com, flatuenza.nu, flatuenza.aero, and flatu.enza is minimal -- and no more or less so than the difference between any of those names and flatuenzical.com, flatuenzo.com, and e-flatuenza.com.
Which brings me -- where else -- back to my first point. It's only a certain desire for euphony in the suggestiveness of our otherwise arbitrary marks that makes us think we need certain names for our services. I look at language and I see something adaptable and fluid. I'd rather ask our natural talent for creative wordplay to carry a little more water and cause a little less disruption by evicting incumbent registrants from their domains. After all, after Ratan N. Tata evicted Bodacious-Tatas.com from its erstwhile home, how people would looking for the particular brand of porn served up by evicted site find it? Not though the DNS system. No, they'd go to a search engine, which would serve up bodacious tatas aplenty.