Dan's Domain Site

Structure: Alternate Roots

Among the most controversial issues in the ongoing saga of the expansion of the domain name space is the issue of "alternate roots." Here is some information on this concept, and my opinion about it.

What is the 'Root'?

Since the domain name space is hierarchical in nature, there has to be something at the top of the pyramid, and that is what is called the "root". This is a "top-level server" that is defined as the ultimate authority on what domains actually exist and where their locations can be found. Actually, for redundancy sake, there are several root servers (with hostnames within the domain root-servers.net), but they are kept in sync. A new top-level domain doesn't really exist on the Internet until it is added to the root servers, so that any system anywhere on the net that is seeking that domain can find out from the root where the specific DNS servers for that domain lie. (Or, to be more precise, it finds out from the root where the DNS servers for the top level domain it's seeking lie, and then finds out from those servers where the DNS servers for the second level domain are, and so on. Thus, to look up host.subsite.example.com, a DNS query would go first to the root servers, then the .com server, then the server for example.com, and finally the subdomain DNS server for subsite.example.com, which can finally report on what the IP address of the hostname host is.) In practice, not all DNS queries need to go to one of the root servers (those servers would be overloaded if they did), because of caches at many different levels in the system allowing queries to be resolved at other levels, but top-level change does ultimately propagate from the root.

Thus, the operators of the root servers have a great deal of political power over the domain name system. Presently, these servers are operated by Verisign, but their policies are determined by ICANN, the organization set up to administer Internet naming and numbering schemes. Since ICANN has attracted a great deal of criticism (much of it highly deserved) for its biases towards large impersonal bureaucracies and against individual Internet users, various people have come up with the idea of "fighting back" against ICANN by setting up alternate roots.

What is an 'Alternate Root'?

Setting up an alternate root turns out to be a very simple matter. The Internet has always been sort of a "do-it-yourself" thing, not centrally controlled or administered like a proprietary online service. You can actually make your own computer recognize any weird domain names you want, whether they officially exist or not. For instance, if you're running Windows 98, you can open up a file named "HOSTS" in your Windows directory using any text editor and edit it to insert entries for domains of your choice -- show.biz, for instance -- and make them go to any place on the net for which you know the IP address (a series of numbers that's the actual address of anything on the net -- domain names are just a human-understandable front-end to them). Of course, if you do that, the new domains you created in this manner work only on your own computer, not for anybody else. If you administer an office LAN, you can set up phony domains there which work for anyone in the office. If you're the system administrator for an ISP, you can set up domains that any of your customers can reach. But they still won't work anywhere else in the world, for people not connecting through your ISP, as long as the standard roots don't recognize them. And, if there's a real site at the address you're using, and it's at a different location than the place you've redefined your host table to point, then you can't access it; the rest of the world will see a different show.biz than you do.

What the "alternate roots" have done is set up servers that recognize a bunch of new top level domains as well as the "standard" ones like .com. That part is easy. The hard part is getting people to use those servers instead of the normal roots. Actually, it's the ISP administrators that need convincing to switch roots, since most normal users connect to their ISP's servers for DNS resolution, so if those servers in turn use an alternate root, then the users would have access to any new domains that are on that alternate root (but wouldn't have access to any that are on the normal root but not the alternate one).

Who would do such a thing?

Several alternate roots exist, some created as a technical experiment, some as a political protest, and others in an attempt to make money by selling registrations in alternate TLDs. (Perhaps some of these root servers were created out of a mixture of all three of these motives.) However, none of them have any official status with the governing bodies of the Internet. That doesn't automatically stop them from operating, however, since any official status that anybody has on the Internet is the result of informal consensus rather than force -- nobody holds a gun to anybody to tell them which root servers to use, but the "standard" ones are the "de facto" roots because they're in traditional use and are the defaults built into server software (so the administrator has to make an explicit decision to use a different root if he so desires).

As of 2010, due to government crackdowns on alleged "piracy" sites and other controversial sites that sometimes include domain seizures, some are attempting to create a new alternative DNS system based on BitTorrent, with domains such as .p2p.

What's my take on all of this?

OK, so where do I stand on this? While I have a good deal of sympathy for the protestors, and have a general libertarian dislike of monopolies, I still think there are some things that are best off being under a single, consistent system. Addressing schemes are one of those things. It may peeve supporters of alternate roots, some of whom have written nice things about me and my domain name site in message boards, but I have to come out against alternative roots and in favor of living with the current root (even while working to get change in some of the more objectionable policies regarding this root).

You see, a naming or addressing system only makes sense if everybody uses it consistently. If every telephone company had a different idea of how the country and area codes ought to be allocated, so that if your long distance service was with AT&T, "1-212" would reach New York City, but with Sprint the same prefix would reach Los Angeles, then telephone numbers would be in a state of chaos. Nobody would be able to give out their number and have any assurance that they could actually be called on it. But fortunately, the phone companies of the world have all agreed on a consistent system. I don't know a lot of details of how that system is administered (this is done under the auspices of the International Telecommunications Union, a treaty organization), but whether its decisions are evenhanded or politically biased, and whether they're made by wise statesmen or corrupt idiots, it might still be better to stick with the system (while trying to reform it) instead of just coming up with "alternative root phone number systems", picking various prefixes that don't seem to be used yet as country and area codes in the current system, resulting in phone numbers that may or may not be dialable depending on what local and long distance phone company you're using, and which may or may not eventually conflict with the same number assigned to a different customer by a phone company using a different "alternative root". Throw in some saber-rattling by alternative numbering operators where they claim to own trademark rights to the number prefixes they managed to grab first, and threaten to sue the original system if it ever uses them itself, and you've got a really big mess.

In fact, in the early days of telephones, there were various competing phone companies that didn't interconnect their systems, so you couldn't call everybody. Similarly, in the early days of commercial online services, customers of each of them could email other customers of the same service, but not those of other services. People preferred to be able to reach everybody, though, so a system eventually arose where the systems were interconnected. This required a unified naming or numbering system.

When the World Wide Web was invented, one of its key elements was the URL, which, depending on whom you ask, stands for "Universal Resource Locator" or "Uniform Resource Locator". But if domain names contained in URLs can be part of any of an unknown number of different roots which may resolve to different locations, the result can't be termed either uniform or universal.

Certainly, anybody can set up their own network, using the same protocols as the Internet and with an overlapping set of participants vis a vis the Internet -- that's what intranets and extranets are. Within such a non-Internet network, the creators can use any naming system they wish, including the creation of top-level domains that don't exist on the Internet. They run the risk, however, of winding up with names that clash with names on the Internet, if not now, maybe in the future as more TLDs get added. That's just the risk they have to take. If the intranet or extranet is kept completely separate from the Internet, with no computers needing to communicate with both networks at the same time, then such clashes might not be harmful, but if there is to be interaction between both networks, it's a good idea to keep the namespaces from overlapping. But one can never be sure this will be maintained given the expansion of the Internet's namespace.

Alternate roots basically define extranets of their own, whose members are those computers that are using that root rather than the standard Internet roots. These extranets might be gatewayed to the Internet, since the alternate roots also incorporate the existing legacy TLDs, but they're not part of the Internet themselves. And once the Internet adds a TLD that conflicts with one in the alternate root, as is happening now with .biz, then users of the "extranet" (alternate root) will find they can't get to those sites, and thus don't have a full Internet gateway any more.

Thus, though I don't have any great liking for ICANN or the policies which it has adopted, I must stick with the unified system which they presently run, and don't consider the alternative roots to have any sort of legitimacy as addressing systems for the Internet. Sorry, alternative root supporters.

Alternate Roots and Intellectual Property

Then there's the issue of whether an alternate root gains property rights to the TLD names it introduces. Some of the alternate roots are making such a claim, but it's dubious. The U.S. Patent and Trademark Office has declared that TLDs have no source-identifying significance, and hence can't be trademarked. Anyway, if it were possible to gain property rights this way, then this would lead to a "land rush" for TLDs just like the land rushes for second-level domains in the current and proposed TLDs, with every sleazy Make Money Fast operator vying to get some sort of "alternate root" operational containing every character string they can think of as TLDs, so they can stake a claim to them. Unlike the past and present land rushes, there wouldn't even be a governing body that could make policy decisions that set the rules for such a rush, but rather, everybody would just grab what they could, and sue one another when their claims conflicted. This might be fun for speculators and profitable for lawyers, but very confusing for everybody else, and not likely to result in a logical namespace.

Even with the present bunch of alternative roots, there are some squabbles about "who got what first". One of the alternate registries has a .biz domain, and has been complaining that ICANN and Neulevel are "stealing" it now by creating a TLD of that name in the official system -- but yet another alternate .biz registry has come out of the woodwork, claiming that they were around even earlier than the other one, and that they are the ones whose name was "stolen". But just how does anyone gain "rights" to a TLD, if it's to be done by some manner other than the ICANN approval process? Do you just have to be the first to set it up to be resolvable, even by a very small subset of the Internet? What if you just set it up within a private LAN, or on the servers of a small local ISP? Or do you have to promote it nationally or internationally? To what extent? A very big can of worms is opened, and as soon as anyone actually succeeds in asserting any such rights, it will bring all the exploiters out of the woodwork to jump on every other suffix they can think of.

...but they can be fun, anyway...

After saying all that, I have to say that (back when I first wrote this page) I'd been browsing the forums at new.net (no longer up), and think that many of the people involved in that alternate root were nice, fun-loving people -- the community that has rallied around their building and using names in an alternative namespace brings to mind the spirit of some of the hobbyist computer networks of an earlier era (e.g., FidoNet in the 1980s) more than it does the present-day Internet with its bureaucrats, lawyers, and quick-buck artists. I have nothing personal against them. If they enjoy living their "cyber-lives" in an alternative namespace from that of the "real" Internet, great for them. But they perhaps should hope that new.net's alternative names never do catch on and become the dominant naming system of the Internet -- because once they did, all the bureaucrats, lawyers, quick-buck artists, and other undesirables would be flooding in, and before long it would be just as bad as the "real" Internet, and made worse by namespace clashes between them and other alternative roots. I believe that any claim new.net may make to having solved the namespace problems of the Internet is really only because they're sufficiently small and insignificant that the outsiders who would corrupt their system don't care about it enough to try. That's how the Internet stayed so pure in its earlier days -- nobody but the computer geeks cared about it. New.net provides a fun playground for its members, and may be useful as a testbed for ideas about such things as the launching of new TLDs, but its ideas won't necessarily be implementable on the "real" Internet.

So, where does XTNS fit into all this?

An announcement a while back introduced something called "XTNS" that is supposed to be a set of "extended namespaces" to expand the domain name system. Initially, it will be marketed to large corporations, where each can get its own TLD and put anything it wants into its namespace, but later there may be TLDs available for public registration. So what is this, another alternative root?

No, it's not. In fact, what they're creating and selling are not domain names at all, not even alternate, non-ICANN-sanctioned ones like those of new.net and the other alternative roots. They've simply licensed the right to create groups of names containing dots within the established RealNames system. This is a keyword system that was introduced with great hype a few years ago, but which few people actually use or have even heard of, despite the fact that they managed to get Microsoft to embed it in recent versions of MSIE.

Adding more namespace, with some kind of hierarchy vaguely resembling domain names, doesn't really make any sort of improvement to this system, or make it usable in any of the ways real domain names are other than as something that can be typed into a browser -- you can't use them in email, FTP, or even in Netscape or Opera browsers. The proponents of XTNS realize these limitations, and make some sort of vague promise of being able to do something resembling email with the new names, some day -- not through your actual email program, but through some scheme with the IE address bar. Once again, those who don't use IE are left out. They're creating a vendor-specific Internet, not the open one I know and love, where I can choose to use Netscape, Mozilla, Opera, or Lynx if I wish. (See my Brand X Browsers Page.)

In fact, RealNames, and the XTNS extension of them, are in the same boat as AOL keywords -- a proprietary namespace that only works if you're using the "right" browser, platform, or connection provider, and is not part of the openly documented standards that connect the entire Internet. All of these systems, then, are a step in the wrong direction, from an open Internet to a closed one.

XTNS got to see for themselves the pitfalls of making their whole scheme depend on a proprietary feature of a specific vendor's browser -- a Microsoft change to how their browser handles "errors" (nonexistent domains typed into the address bar) made the XTNS names fail to work at all, making the whole scheme inoperative. But XTNS forged ahead with internationalized names in other languages' character sets, which apparently worked... for a while... But eventually Microsoft ended its deal with RealNames, and that company went out of business, scuttling all schemes based on this proprietary system.

Mobile Domains

Then there are a few schemes out there to assign domain names for the use of mobile telephones and other wireless Internet-capable devices. The theory is that these things will use a range of services different from normal landline-based Internet connections, and thus one could create an alternate namespace just for them without conflict with other Internet sites that don't need to use the same services. MWonder is one such scheme, with its own set of domains and subdomains, which have numeric equivalents to facilitate use with phones. For this to work, the mobile Internet providers (e.g., cell phone companies) will have to support this system, I think.