David F. Gallagher
MIT Technology Review, November 2002

Surfing the Web, as the term implies, is all about forward motion. And like the aquatic version, Web surfing against the tide poses quite a challenge. That’s because the Web hyperlink is a one-way affair: it’s easy to follow links from page D to pages E, F, or G, but the Web’s architecture offers no simple way to see which pages A, B, and C link to D.

Programmers have tinkered with solutions to this problem since the early days of the Web. But interest in solving it has picked up recently with the spread of weblogs, most of which are personal sites full of links and commentary. In the last two to three years, hundreds of thousands of netizens have created weblogs to chronicle their daily lives, discuss the latest news, or share their expertise in their chosen fields. Many webloggers link to each other’s entries, creating threads of conversation scattered across multiple sites and, consequently, a new demand for "backlinks" to see who is linking to what. "I post to my weblog; you respond in your weblog–but without backlinks, I may never know we’re having a conversation," explains Mark Pilgrim, a developer and technology trainer in Apex, NC.

Webloggers say recent experiments with backlinking could benefit all kinds of online publishing. Instead of pointing readers only to sources for the item they have just read, backlinks also point to newer material that item inspired, making it easy to follow a path through the Web’s marketplace of ideas. And because they can be updated automatically to reflect new incoming links, backlinks turn static Web pages into active hubs of related information.

Search engines such as Google are one source of information on inbound links–just type "link:" and a Web address in Google’s search box. But centralized search engines aren’t updated frequently enough to allow the kind of discussion tracking webloggers want. A better route, software-savvy webloggers have found, is to make use of referrers. These are the addresses, sent along with page requests when you click on a link, that show where the link was found. In most cases, such referrer data is available only to the owner of the requested site, but one way to create a feedback loop is to automatically paste the addresses from server logs into the pages themselves.

That’s what software developer Chris Wenham does in his Web magazine, Disenchanted. Wenham wrote a program that puts backlinks at the end of each article, with the most-used links listed first. Occasionally, Wenham adds his own comments or excerpts from the linking pages. "Bloggers want to be part of a community," and such backlinks are one of the easiest ways to gather and share feedback about their ideas, Wenham says.

Ben and Mena Trott, a husband-and-wife team behind a popular weblog publishing system called Movable Type, have enhanced their software with a more sophisticated backlinking feature they call TrackBack. When a weblogger publishes an item that makes reference to an entry on another site, the software sends an announcement, or a "ping," to that site. If the target site is also using the software, it will automatically add a link and an excerpt from the new commentary to the bottom of the relevant entry. The feature sets up "a connection between authors" that is stronger than would be possible with referral tracking, says Mena Trott.

Webloggers say academic and news sites could benefit from backlinking. "It would be nifty if the newspapers did this, so people could get a sense of the discussion going on around certain topics," says Peter Merholz, a consultant with Adaptive Path, a San Francisco interactive-design firm. An online retailer might also use the technology to offer access to consumers’ reviews of its wares, suggests Wenham. But the retailer would have to be comfortable with linking to negative reviews. "You can bet that a lot of the backlinks will go to pages that have gripes," Wenham notes.

Mena Trott expects that more complex uses for systems such as TrackBack will emerge over time. "There’s a lot you can do with it that we haven’t figured out yet," she says. Perhaps the same could be said of the lowly hyperlink.

This article appears in the November 2002 issue of Technology Review and is posted here with (something like) permission.

More articles by David F. Gallagher



There have been a few developments since I filed this story. Earlier this month some comments from Jeffrey Zeldman inspired Mark Pilgrim, Chris Wenham and others to change the formatting of their backlinks so they looked less like a popularity contest. Chris started putting the most recently used links first, instead of the most popular ones, while Mark made the visitor count less prominent.

While he was at it, Mark got his system to automatically add a text excerpt from the linking page, making his "Further Reading" links pretty much the slickest backlinks around.

Meanwhile Ben and Mena Trott have ambitious plans for TrackBack, some of which go far beyond backlinking and so were beyond the scope of this article. For example, it's being used here to bring together posts about knowledge management from a bunch of different weblogs.

Then there is Pingback, which as far as I can tell is meant to be like TrackBack but with less work involved on the user's part.

The backlinks on this page are made possible by a service created by Sean Nolan, which I've been trying out on my site since I started working on this story. It was easy to set up and it's been fun to keep an eye on. (11/03 update: The service moved to Blogtricks and no longer appears on this page.) There's a similar service available at Steven's Web. For those who can run their own scripts there's Dean Allen's Refer.

If you would like to comment on this story or you have a link that belongs here, you can send me e-mail. Thanks. -- David, 10.30.02

Update: Eric Scheid has a nice roundup of links on referrer linking. And Denise Howell did some thinking about the implications for legal research.