Logging onto the Internet is billed as a universal experience. Yet for years, below the radar of most people’s online experience, there have been sets of Internet domain names accessible to some people but not to everyone. Before the introduction of an official .biz domain, an independent .biz had existed for six years—and for a time the two survived in parallel, so that typing in a .biz address might send you to different sites, depending on which Internet service provider you had.
And now countries using non-Roman alphabets, notably China, are pressing for full domain names in their native scripts, prompting concern about whether the domains will be accessible to users of Roman alphabets. Will the proliferation of alternative domain names make the Internet more user-friendly and global, or will it accelerate a fracturing of the Internet that may have already begun?
The universality of the Internet lies in the domain name system, which translates between the names we humans understand, such as google.com and ieee.org, and the numerical addresses of the computers hosting those domains. To retrieve a URL or deliver an e-mail, a user’s computer needs the address of the relevant remote computer. Often a local server, known as a domain name server, will already know it. But when the local service doesn’t, there’s a defined way to find it out: start at the end of the address and work backward.
That final, or right-most, part of a domain name, such as .cn, .uk, .com, or .org, is called the root; a set of 13 root server operators maintains the authoritative list of servers for so-called top-level domains, Those top-level domain servers, in turn, can offer more specific address information for ”google,” ”ieee,” and the like. Once a request gets to the second-level domain, local servers for those domains can direct the original Web request or e-mail to the right Web or mail server.
”There shouldn’t be any kind of local name that works only in some places, from some ISPs,” says Paul Vixie, one of the designers of the system. Hence, a single body, the Internet Corporation for Assigned Names and Numbers (ICANN, in Marina del Rey, Calif.), is responsible for assigning top-level domains. The entire system is predicated on each domain’s being unique. There can be only one .biz, for example, for the same reason that a postal service can’t have, say, two different states or provinces represented by the same abbreviation.
And yet, for a while at least, there were two different .biz domains. Karl Denninger, a network consultant, created a .biz top-level domain in 1995. It was later administered by Atlantic Root Network Inc., based in Georgia. Then, in 2001, ICANN created its own .biz. Some ISPs had to make a choice: keep sending queries to the independent root servers used by Atlantic Root or switch to the official one. The root operators eventually gave up the old .biz domain, reverting to the ICANN-approved domain, which is run by NeuStar Inc., of Sterling, Va.
Fears of Internet fragmentation are largely based on concerns about alternative roots. But roots that try to tamper with existing top-level domains are likely to be marginalized—or worse—and therefore would not impinge much on universality, says Karl Auerbach, a former ICANN board member who runs his own top-level domain, .ewe. ”If any root system operator were not to carry .com, [the operator] would simply lose visibility,” he says. If another .com were substituted, says Auerbach, it would likely be sued successfully for trademark infringement.
Most roots system operators add new top-level domains to ICANN’s list. New.net Inc., an El Segundo, Calif., company, has deals with EarthLink and some other ISPs, giving their users access to top-level domains including .shop and .travel. A new entry to the field, Netherlands-based UnifiedRoot S&M BV, offers custom top-level domains.
Some roots server operators offer the same top-level domains, and if this trend ever became widespread enough to really matter, obviously the effect would be to confuse users and erode the Web’s universality.
But competing roots have never enjoyed a widespread following. ”When was the last time somebody contacted you using a new.net name?” asks Milton Mueller, of Syracuse University’s School for Information Studies, in Syracuse, N.Y. If other new roots did become popular, the core top-level domains would likely remain universal, says Auerbach, but servers or users could select the extras they wanted. He says he once used a service called Grass Roots, now gone, which monitored all available roots and allowed operators to choose the top-level domains they wanted to serve.
China and some Arab countries have begun testing domain names in their native scripts. How to configure root servers to recognize such domain names is a touchy political issue, and as a result, countries could set up their own roots. Arab countries have tested domain names in Arabic by routing those queries to their own root servers, for example. But for now, the Chinese and Arab groups say they intend to maintain the officially recognized root, with only ICANN’s top-level domains.
Currently, the Chinese names can be translated into coded strings of ASCII characters, so they are intelligible as Roman script and thus available across the Internet, according to the China Internet Network Information Center, in Beijing. If China adopted its own roots, similar protocols could let servers translate between roots, says Mueller. Countries sharing alphabets would have to agree on how to register names, but China, Japan, and Korea have already published guidelines for their overlapping scripts. Coordinating several hundred roots, each offering domain names in a different language, would certainly be difficult, says Mueller, but ”not as difficult as some Internet purists have led us to believe.