CircleID

As an alternative… why build this on DNS?  A Peer-to-peer discovery service could be designed to supplant/extend the DNS directory.  This service could match any cornucopia address to its "proper" IP or DNS address, allowing access to the web content.  The service could be built as a browser plugin that facilitates searching/ranking/modifying names and addresses.  This solution would allow an independent development team to create and roll-out this service without modifying the current DNS structure. 

I'm not sure I see a compelling economic model for being the registry, especially if you consider that ICANN will continue to allow more specialized (indeed, generic) TLD registries.

Who would buy a random-letter-TLD domain when they can get their name at any of the myriad of new, generic TLD registries that ICANN posits will be created starting next year?

On the matter of random-letter TLDs versus meaningful generic TLDs: a meaningful TLD is very attractive if it happens to be in your language and have a relevant meaning. If it also has loose registration permissions, then it will result in a "land grab" for famous trademarks and generic terms (assume the TLD in question is ".shop" and fill in the blanks: "sex.shop" will be registered in the first nanosecond). Consequently, a privileged few will be lucky enough to get the generic term they were after, and everyone else is left to pick up the scraps, as happens with ".com" and the other popular TLDs. The also-rans may well prefer a good generic term from cornucopia to a second-rate term in the new gTLD.

Alternatively, if the meaningful TLD in question has very strict registration policies (offering some sort of guarantee that the registered name bears a meaningful relationship with a real-world entity), then registration opportunities are reduced, and cost is increased. In short, such a domain has a high barrier to entry relative to cornucopia, and so many people will consider cornucopia a better option.

In any case, the existence of cornucopia is not incompatible with meaningful TLDs. If economies are your preferred judge, implement both cornucopia *and* as many meaningful TLDs as you like, and "let the market sort it out".

This is almost exactly the same proposal as one I made back in the late 90s, as I recall, though I had revised it a little bit (Remove domains like 'a00' that might be seen as having intrinsic value.  The goal is they all have the same value, no one is inherently more attractive than another.)

However, I used that mostly as an example, because in reality you don't want strings that can be easily confused for one another in names.  You get a really nasty typo and memorability problem.

You reference one of my pages but I guess are unaware of the other proposals.  In the end I concluded that trademark law actually had worked out a way of dividing up namespaces and assuring all names started out equally valuable—but then gained value as you invested in them.

However, there are other intermediate solutions, including letting anybody who wants be a TLD registrar with any string not found in google (including with embedding spaces at any appropriate point.)

I still maintain the TM system works the best here, as it has a long and well established legal history we can rely on rather than going out in the fog.

I wrote about it here at circleID, it's also to be found at www.templetons.com/brad/dns/

I have no problem with this approach, provided that it is not mandated that this approach be taken to the exclusion of any other (including generic TLDs). This, simply because there are still those who wish to compete (and have, since 1995) with existing registries, and mandating that there be no more generic TLDs would create an unfair monopoly among existing registries.

I fully support a "let the market decide" approach, and would welcome this plan along-side other, new registries.

Brad, while I have your attention, here's a request: please put dates on your essays. It makes them much easier to cite. Also, even now that you've made me aware of your prior work on this matter, I'm unable to find the actual proposal. The key difference between my proposal and your summary here is that I suggest random allocation of TLDs as a solution to the "intrinsic value" problem, where you suggest removing the attractive domains.

Various letter-digit-digit TLDs will have minor attractions relative to others, but the problem can't be generally solved by eliminating the attractive ones, because attractiveness is a culture-specific phenomenon. The Chinese, for example, are going to have a very different opinion about which of the domains are good and which are bad, because they associate meanings with numbers in a way we don't.

The cornucopia is also different from your throw-away suggestion to allow meaningless TLDs (verified as meaningless by lack of Google hits). Once again, this stems from the fact that there is a namespace, and random allocation within the namespace. Cornucopia allows the same generic term to be registered as a second-level domain again and again, and this makes it unique. Ad hoc meaningless TLDs create a few opportunities to register a particular generic term, but cornucopia should be inexhaustible.

As regards trademarks, I have no problem with a portion of the root namespace being allocated to trademark-related name allocations, but I think that basing DNS naming as a whole around trademarks is dreadful. Trademark is both business-oriented and dispute-oriented, whereas cornucopia is egalitarian and aims for peaceful coexistence. Cornucopia therefore seems much more attractive to me, not because I'm a peace-love-and-mung-beans hippie, but because I know I'll be the losing party in any trademark dispute that a typical company-with-legal-department might care to aim at me, purely on the basis of it being not worth my while to mount a defence.

Actually, this point bears a little further emphasis. Most other proposals for meaningful TLDs spend some time talking about the rules relating to who may have what name. One of the main aims of cornucopia is to eliminate that entire stream of argument. The rules for cornucopia are, consequently, extremely simple to both describe and enforce: namely, "take what you want." It's very hard to abusively violate such a permissive rule. Cornucopia registrations have zero overhead in rule-enforcement.

Well, I discarded most of those essays.  In fact my first first proposal, long gone, was what I called inserting a disambiguator into requests for the same name.  At the time, my now rejected thought was that you could ask for a domain with or without a disambiguator (effectively random number).  If you asked for one without a disambiguator, you were taking a risk.  If anybody else showed up with a legitimate request for the same term (which is to say, you couldn't get them enjoined from doing that request in trademark court) then you lost your domain, and both people got a domain with a disambiguator.

The goal was not to have people lose domains unless they were idiots.  But to allow people with truly unique names (kodak, xerox, etc.).  Anyway, it wasn't really that workable a system, and also the numeric or random disambiguators turn out to be a bad idea.

They are a bad idea because they give up on some of the most important goals for domain names people have—you want them to be memorable and accurately typable.  Random numbers fail miserably here.

My 2nd stage of this theme—also long ago discarded—was to use the equivalent of street names as disambiguators.  Elm, Pine, etc.  Memorable and typable but not generally meaningful.

Your proposal could easily be modified to this form, though of course internationalization is an interesting issue.  You could have several classes of random disambiguators—including numeric, roman, chinese, arabic and the other major character-set groups.  Numeric is fair and simple but loses the test.  It destroys the village in order to save it.

My second thrust however is to agree in one tiny way with Esther's otherwise really-wrong essay.  The idea of lots of registrars out there all selling the same commodity, unable to compete except on very basic terms like how little they will make on top of the registry fee—that this is not a good regimen.

THere should be competition, competition on all elements.  Policy.  Fame.  Price.  Service.  Things nobody has yet thought to compete about yet.

I believe the answer to that is to find a way to allocate a set of fully owned TLDs, and to make all those TLDs of equal intrinsic value, and let TLD companies claim them and compete any way they can.

Now there are a few ways to make the TLDs start with the same equivalent value.  (Of course, the idea is that after time, some TLDs attain lots of value through the hard work of the TLD company.)

Random numbers is one.  Random strings another.  Random strings with some phonetic memorability and typability another.  Hunting for strings not found in google is good.

But as I said, the trademark business already worked this out long ago.  If you want to own a term in a context, it can't have pre-existing value as a generic term.  It must not mean anything specific.  If it does, it's generic.

However, I am open to any system of allocation that is fair.

What is not fair is what Esther proposes.  Monopoly ownership of English words to the TLD managers.  Aside from the internationalization question, it's just plain wrong.  The idea that I can't say I'm "safe" unless I please the monopoly owner of the word "safe" in the TLD space.  That's crazy.

On the other hand, the idea that I can't declare myself a 3* Michelin guide restaurant without impressing Michelin—that's a great and time tested idea.  You don't give Michelin a monopoly on rating restaurants by having them (or anybody else) own .restaurant (or even by having many registrars put things in .restaurant first-come first served).  You let them publish their guide, and you let them build a famous name for their guide, and you let them and only them decide what qualifies to be in the guide.

That being the case, Brad, how do you propose to remove .NET from the zone? Surely you can't allow one company to run a generic term, yet constrain trade for all other companies that wish to compete on the same terms.

A propos Mr. Brett Watson's Cornucopia approach (see below) to new TLDs, the idea might have had merit were it not for the fact that the political and economic interests of the key players (we all know who they are!) precludes this. Mr.Watson makes the point that generic [names]are "highly coveted" in the dot-com name space. Highly coveted is right. However, Mr. Watson is not correct where he observes that all coveted generic SLD names are taken in the dot-com name space.

The Achilles Heel of the ICANN regulatory apparatus and, indeed, domain name registrars in general,would be the scary situation where a single private entity captured enough generic SLDs in the dot com name space to mimic the name space itself: in other words a "replica" TLD that nonetheless perfectly resolves in the root. A privately-owned replica TLD or ccTLD in any name space means for practical purposes that ICANN is irrelevant.

In fact, and it is a fact, that this has already occurred.The hegemonic apparatus which is ICANN (and its would-be pretenders such as ITU)is a function of the ideological predispositions of the entrenched players referenced above that a name space necessarily is private property. Notwithstanding the holding of the 9th Circuit Court of Appeals in the matter of Sex.com (Kremen v NSI) the dot-com level one name space is a shared information utility open to anyone sufficiently ahead of the curve to capture the required number of generic domains to make doable the concept of a level three dot-com name space within the level two name space itself. This is the dot-com "replica" scenario.

400 or so generic dot-com SLDs is, for practical purposes, a sufficient number of natural language product and service-specific dot-com master channels to replicate the dot-com name space. When that happens, it becomes feasible for any private entity to replicate the dot-com TLD by selling level three names attached to level two names that mirror (generic) consumer products and services, say, "air", "book","car" and so forth across the commercial alphabet.

The hegemonic apparatus which is ICANN (and NSI in particuar) would have us believe that level three name spaces are anathema to the general public. This is (of course) nonsense. ICANN sanctioned the .NAME registry with exactly that purpose in mind: level three name spaces.

There are tens of thousands, maybe millions, of commercial entities around the world that themselves "covet" the dot-com name space, but they are unable to secure relevant names because the dot-com TLD was and is monopolistic by design.

However, technology (as we all know) challenges that assumption. It is perfectly reasonable, and doable, to assign a relevant and proprietary, (meaning selected by the registrant), domain name in the dot-com name space to anyone who wants one. Anyone means anyone. All that is required are the master generic channels in the form of <myname.masterchannel.com> to achieve that purpose. This, as I said, has already occurred. Such a dot-com replica already exists and perfectly resolves in the DNS.

Any business or private entity can obtain any proprietary dot-com address they wish, for example,<verisign.masterchannel.com> without offending copyright or trademark regulations. This network ...and clarity is important here...in most cases makes it(thereby)feasible for anyone, anywhere to obtain any dot-com address they wish. The word "relevance" means an appropriate (personal/private) level three domain name in a relevant product or service specific level two dot-com master channel, by simply requesting it - and paying for it.

Almost...emphasis on almost...every commercial product or service category required by everyday businesses has already been assigned a generic level-two generic address in the dot-com name space. The process began in 1999. The replica dot-com network sits silent and unused to this day.
++++++++++++++++
Brett Watson
++++++++++++++++
"So why would anyone want a meaningless TLD? The first and foremost reason: generics! Generic terms are highly coveted in ".com" space, and it's safe to say that they are all taken. In fact, they're probably all taken in all the available TLDs that allow registration of generic terms. The cornucopia allows everyone to register their favourite generic term, since there is plenty for all. The meaningfulness of the second-level name can make up for the meaningless TLD to some degree."