Third Voice: Wrong Turn on the Road to Xanadu

Editor’s note: This article was written by me and originally published in the Online Community Report in 1999. OCR went on hiatus in 2004, and when it came back in 2006 only a few of its old archived articles remained on the site. With their permission, I am republishing this article on this site for historical interest, along with a new epilogue explaining where the story is today.

Is it the next wave of online community software, or the digital equivalent of the vandal’s can of spray paint? This is the question that online community builders are grappling with regarding Third Voice, a new tool that lets visitors post persistent notes on sites that are visible to everyone whose browser is equipped with the (free) Third Voice viewer.

Third Voice claims that their software is the Next Big Thing in online communities, bringing together groups of people who share the common interest that each site represents. In one sense, they are correct — opening Web sites to annotation does create a new space in which communities can develop. But there is no way for community-builders to moderate these discussions, leading some to worry that Third Voice is less about community-building and more about letting anyone in the world scrawl on your pages.

To understand Third Voice, let us look to the birth of the idea of annotating hypertext documents. This will demonstrate that the spirit of Third Voice comes, not from the Web, but from an older, grander, more idealized conception of hypertextual communications: Xanadu, the forty-year impossible dream of professional visionary Ted Nelson.

The Web that Never Was

In 1960, before the Web, the Internet, or the personal computer — heck, before we were even done liking Ike — Ted Nelson, then a graduate student at Harvard, had a vision. He foresaw a network in which documents were freed from what he saw as the tyranny of the linear narrative. He saw this system, which he named Xanadu, so clearly that he and his team have spent the last forty years of their lives and millions of dollars of other people’s money trying to make it a reality. (That’s right, they’re still at it — Project Xanadu even has a product now, a “hyperstructure kit” for Linux called Zigzag. You can read the saga of the development of Xanadu in Wired’s archive.)

You would think that the Web revolution would have made Ted Nelson one happy camper — after all, his vision of a hypertextual document library came true, right? But to Nelson and company, the Web is a step backwards, its simple protocols and libertarian approach to version control a slap in the face compared to the elegantly awesome (and, not coincidentally, famously difficult to implement) system that manages such matters in Xanadu.

So Nelson and Xanadu have been left behind by the hypertext explosion. This is partly because Xanadu is much more ambitious than the Web — it attempts to control such problems as link-breaking and copyright in the system, where the Web pushes these responsibilities toward the individual site builder. But it is also partly because the Web is centered around an easy-to-understand metaphor — the page, which, once published, cannot be altered — while Xanadu, in reinventing writing, lacks such a convenient mental hook.

Third Voice, however, alters the page metaphor by taking the power to set context away from the site designer and giving it to any user armed with their software. In doing this, it seems to add to the Web a central concept of Xanadu. Nelson’s system allows document annotation through transclusion — the sideways inclusion of one document within another. So, in Xanadu, if you wanted to critique a document, you would transclude that document within your own. Nelson’s idea was that you couldn’t change the original document itself, but you could create a new version of it, with your remarks as context, that Xanadu would store right alongside the other one. In this manner you could create your own personalized docuverse by seamlessly combining material created by others with your own material.

Why Should You Care?

What does all this mean for those of us tasked with nurturing online communities? If Third Voice doesn’t catch on, it means little; this only really becomes interesting when it becomes ubiquitous. But if it does catch on, it could add a discussion around any content item on your site, without having to build in any technology. Third Voice pushes the technology burden away from the server and onto the client, making it extremely easy for users to participate, as well as for webmasters to set up (since you leave all the setting-up to your users).

Of course, this ease comes at a price — complete loss of control over the discussion. Want to moderate out off-topic notes? Too bad! Want to screen posts with offensive language? No soup for you! The only moderation mechanism is collaborative filtering — users can vote posts up or down, and then screen out all posts that fall below a certain level. But Third Voice has made a misstep in not providing a way for community managers to tie Third Voice discussions more closely into their site. Even an opt-out or notification system (currently you have no way of knowing if your site has been Third Voiced, short of browsing it yourself) would be a step in the right direction.

This is where understanding Xanadu helps us understand Third Voice. In a sense, Third Voice has taken one of Xanadu’s central tenets — dynamic document annotation — and turned it into a weapon aimed at the heart of another — the integrity of online documents. Nelson’s vision of transclusion held that the original document was sacred; annotations could be made, but these would exist in parallel with the original document — you could literally put them side-by-side on the screen and the system would illustrate the changes between them. Third Voice discards this principle; if you are running their client, the only way to see the original, unmarked document beneath the notes is to turn it off. If Third Voice becomes ubiquitous, the documents we create online will cease to exist in their original forms, becoming the property of the world, writhing and roiling with life, never meaning exactly what we intended them to.

Is Third Voice the Next Big Thing in community-building software? Perhaps; but by excluding community-builders, Third Voice has positioned themselves to succeed only if previously silent “communities” that exist around various Web sites are cohesive enough to support rich, involving discussions (which I imagine is true only of a minority of sites). At the same time, they have given us a slice of Ted Nelson’s dream, but with a twist that misses one of the most important ideas behind Xanadu altogether. In short, Third Voice has missed an opportunity to bring users and community-builders together by turning the wild, wild Web into something a little more like the Xanadu promise.

EPILOGUE (December 13, 2006): Funny how things work out sometimes. When Third Voice launched, it seemed like it was going to be a Big Deal; possibly in a good way and possibly in a bad way, but certainly it was going to be Big.

Then they showed it to users, and it wasn’t.

Nobody used Third Voice. Across the Net, millions of webmasters sighed in relief as the prospect of people marking up their sites vanished in a puff of hype. In April 2000, the company shifted its business model and abandoned the original Third Voice product; a year later, they were out of business altogether.

Here’s what Philip Kaplan of said about Third Voice in his 2002 book, F’ed Companies:

Almost immediately upon the launch of Third Voice’s service, controversy broke out. A lot of webmasters didn’t like the technology, as they imagined a world where everyone used Third Voice sticky notes, and webmasters would lose control of their sites.

Didn’t matter, it sucked — especially if you were one of the investors who pumped $15 million into this toy.


Oh, and Ted Nelson is still working on Xanadu. As of this writing, it has been forty-six years since he first described it.

UPDATE (May 23, 2007): Proof of the old adage that history repeats itself, first as tragedy, then as farce: Fleck is a new “Web 2.0” startup that is exactly the same idea as Third Voice, only with AJAX.

If you needed any more proof that Web 2.0 is a bubble, well, here you are!