DNS articles
Brad Templeton Home
Brad Ideas
(My Blog)



Jokes / RHF

Photo Pages

Panoramic Photos

SF Publishing








RHF Home

Copyright Myths

Emily Postnews


Burning Man

Alice Pascal

The Rules for Guys

Bill Gates

What's Wrong with other Proposals

What's Wrong with other Proposals

There are a lot of answers and alternatives to DNS out there, some better than others. What are some of the issues with them?

Multiple Roots

Some people have looked only at the most contentions problem, driven by the goal people have of creating names that are a simple word or phrase that people can type into a web browser.

They correctly conclude that real, searchable directories are the right answer to this goal -- and lots of them, ideally in competition.

Thus when you type in "apple," your tools and the directory you use should offer you a choice of "Apples" from Apple Computer to Apple Records to the Washington State Apple Board. Today you mostly get apple.com -- Apple Computer. Your tools might provide different answers or different sorting based on your own preferences, and there might be a competitive market.

All great, but people still want a naming system that's 100% reliable with unique names. I want to put an E-mail address on a business card and it should get to me with 100% reliability. The person mailing me shouldn't be offered a menu of other people when using my address.

But at the same time I want my address to be readable, memorable, and easy to type without errors. That's also true when I put a URL in print, or say it on the radio.

A searchable directory is a great feature, but it's only part of the problem.

Multiple DNS roots

Some have gone further to suggest that an entry, to be unique, consist of the name of a dirctory company, and then some string that company will recognize. Much like today you might say "Look me up in the GTE Yellow Pages" to give somebody a reliable way to find you. In fact, my system is very close to that, done with DNS.

But it must be reliable, so there must be only one of each domain name, and somebody has to either hand out TLDs to distinct parties, or, if multiple parties can register in the same TLD (as is the case today), somebody has to keep the TLDs in line to avoid conflicts. That's a root, no matter what name you call it.

DNS is not a directory, let's stop pretending it is

Advocates of this philosophy mean directory in the sense of a domain name or browseable directory like Yahoo. DNS was never meant to be a directory like that, and many of the problems come from people trying to use it like that, taking the philosophy of "You want IBM? That would of course be at ibm.com."

As such, advocates of this say that we should forget about DNS, and build real workable directories for when people want to find things, and leave DNS as a mostly invisible to the public level of pointers and indirection.

This is wrong for two reasons. One, while DNS should have been used as a way to type in names and get results, it has a lot of other purposes and goals which are useful and desired. People still want E-mail addresses that are meaningful and memorable, and while E-mail directories are useful, they want them in contexts that are not directory-like, such as in seeing the reply address on an E-mail, or reading one off a business card.

Secondly, for better or worse, people and a lot of software are wedded to DNS. Browsers have already moved to offering true directory functions when you type things in their location box, and that hasn't altered the problem. Getting rid of DNS for human consumption is a long way off.

Lots of new TLDs

Many have criticised ICANN for not just handing out lots of TLDs so that there is no more scarcity. This is a positive move, but if the new TLDs are generic, that just gives more ownership of generic terms to more people. Lots of new TLDs is the answer, but brand-name ones only.

Allowing anybody to get a TLD isn't the answer because in effect that just moves the ".com" problem up one level. TLDs would quickly become as scarce or more scarce than .com domains, because they are even better.

Making them expensive helps reduce the number, but in a very elitist way.

NEW.NET and other alternate ICANNs

Many parties have attempted to set up their own alternate NICs to supplant the ICANN/NSI one, and offer new TLDs within that. Most have been small ventures. They tend to create new TLDs and leave the others alone rather than attempt to replace .com. Only sites that use their new list of root servers can see these new TLDs, the rest of the world can't.

One new venture to do this is new.net, which is selling domains in a lot of new TLDs. Their hook is that they partnered with a series of major ISPs to make sure that many people could use the new TLDs from day one, and all people could use them by appending ".new.net" to access through the ICANN domains. Still, their names are far from universal and not yet popular.

But these efforts are just a power grab. Most of these parties just want the chance to own and sell domains in generic TLDs. They're a new ICANN, possibly with nicer policies, but really just another group trying to own what shouldn't be owned.

Some of them also will lead to splintering the net. If two TLDs with the same name exist (as is already the case with some of the new TLDs ICANN is adopting) and get any serious use, you get a fractured net, where site.biz for some users is different from site.biz for others. This benefits almost nobody.

Slow addition of new TLDs

This is the course ICANN is taking. It also is flawed, for they are issuing generic TLDs, and giving them outright to registrars, without even the sharing system worked out for .com. The slow pace assures that the artificial scarcity of the original system continues, and in many cases just encourages domain holders to register "their" domain in all the TLDs to block others from using it. This, of course, is a goldmine for the registrars, collecting money for selling cups of water from an infinite pond.

Single registry, many registrars for all TLDs

This was the solution put in place for the legacy generic TLDs (.com, .net and .org) to reduce NSI's monopoly.

But in fact it still gives NSI a monopoly, and a $7 per registration gravy train. That NSI was gouging the public at $35/year became obvious when competitive registrars took their prices down to around $10, with $7 of that going to NSI just to add a record in a database. (I can buy a book from Amazon.com for $5 and they manage to store lots about me in their database and give me a book to boot.)

But the real issue is that while nobody has a monopoly on being a registrar, if the TLDs are generic, then people still get ownership of generic words and phrases in the second level domains (2LD). If we have the .museum domain, then the first registrant will get "history.museum" and nobody else will be allowed to use that name, even though it is a generic phrase nobody should monopolize. It doesn't really solve much that they can buy it from any registrar in a multi-registrar system.