January 2001

Where in the World?

A new scheme unites the Internet with geography.

By Alexandra Stikeman

The Web can be a world unto itself, but there are many times when it would be nice if you could understand its correspondence to actual geography. Say you want the Web sites of all the museums in Boston within a few miles of your apartment, or directions to the shoe repair store nearest to where you're standing with your Web-enabled cell phone. Some search engines and online directories can provide the information, but the power and accuracy of your search depend on how many sites that search engine has indexed, or how many businesses have registered their addresses with the directory.

What if, instead, there was a consistent way for all Internet-navigation tools to keep track of businesses' and organizations' real-world locations, as well as their Internet addresses? That's just what a new Internet architecture proposed by Menlo Park, Calif.'s SRI International aims to do.

Last fall, the Internet Corporation for Assigned Names and Numbers (ICANN), a nonprofit corporation responsible for the assignment of Internet domain names, considered a new series of proposed top-level domains (TLDs) to supplant the now-crowded original ones: .com, .org and .net. Among the more than 200 TLDs suggested by companies from around the world was SRI's proposed .geo-a central component of the research outfit's plan for a new geographic Internet infrastructure.

Here's how the .geo system would work: When a store first registered its domain name, say www.GoldenGateSuits.com, it would also have the opportunity to register its latitude, longitude and a description of the store's Web site with one of several companies called "GeoRegistries." The GeoRegistry-call it Acme in this case-would store the information on a server dedicated to the specific geographic area that encompasses the suit store. That server would have its own domain name, along the lines of "acme.28w49n.2w7n.120w30n.geo." The first part of the address (acme) indicates which GeoRegistry owns the server and the lengthy middle part of the address specifies the geographical region covered by that computer.

An Internet user, however, would never have to see this rather unwieldy name. He could simply go to any .geo-enabled search engine, type in the key words "men's clothing store," and in a separate box, "Pacific Heights, San Francisco." The search engine would then translate the neighborhood and city name into a geographic location, contact all the GeoRegistry servers responsible for that area, and find all the Web sites registered to men's clothing stores in the Pacific Heights area of San Francisco.

One of the forces behind SRI's proposal was the desire to put control in the hands of the businesses who put their Web sites up on the Internet and hope to be found, instead of in the hands of intermediaries such as search engines or online business directories and reviews.

"All of these sites give you a very narrow view of what's in the area you're interested in," says Yvan Leclerc, director of the .geo initiative at SRI. And with the rise in demand for location-based information services for wireless devices, technologists are starting to grapple with the issue of how to organize and store geographical information in a systematic way that everyone can follow.

As TR went to press, ICANN had not yet announced its list of approved TLDs. But SRI's proposal has already raised a number of eyebrows in the Internet community. Some have criticized it for adding yet another layer of complexity to the Web. Others have questioned whether SRI would have a monopoly over the naming system.

Regardless, "the basic idea is a very intriguing one," says Joseph Ferreira, professor of urban studies at MIT and an expert in geographic information systems. "Sooner or later something like this is going to catch on. If not a .geo, then some other scheme."