A Map Is Not the Journey: How Xanadu Arrived in Japan

by Joel West

I was interviewing a professor at a Keio University faculty club late one afternoon when a somewhat disheveled< I>gaijin came over to inquire about the PowerBook I was using to type my notes. After a brief exchange, he introduced himself as "Ted Nelson." "The Ted Nelson?" I asked? He smiled at the recognition and nodded.

Intrigued that the man who put the "hyper" in "hypertext" was working at Keio's Shonan Fujisawa Campus (SFC) in Kanagawa prefecture, I headed to his office once my interview was over.

A short answer to the question of why Nelson is in Japan -- as with most things in his life for the past three decades -- is "Xanadu," his grandiose vision for a global information retrieval system. But as I realized when I left his office three hours later, when it comes to Nelson there are no short answers.

Beyond hypertext: Xanadu

Who is Theodor Holm Nelson, and what is Xanadu? To understand requires going back to the beginning of his career as a hypertext visionary. It was Nelson who, dissatisfied with the sequential presentation of information, coined the term "hypertext" in a 1965 conference paper. The concept, however, had little impact at the time.

An epic 20,000-word profile in the June 1995 issue of Wired referred to Nelson's 30-year quest as "The Curse of Xanadu." Nelson detests that article, but in his own book he quotes an earlier Economist description of "Mr. Ted Nelson, gadfly, prophet, and self-confessed computer crackpot, with a lifetime's obsession wrapped up in an enormous program called (after Coleridge's unfinished poem) Xanadu."

I first learned of Nelson in the fall of 1987, when Apple's new "HyperCard" program introduced the concept of hypertext to a mass audience, and Nelson enjoyed his 15 minutes of fame as the godfather of hypertext. Later, this same hypertext concept spawned HyperText Markup Language (HTML), which has become the lingua franca of World Wide Web (WWW) browsers such as Netscape and Mosaic.

While hypertext links were always part of his vision, Nelson stresses they are only a part. "No one else understands the other principal feature: transclusion. Transclusion and links are left and right halves; if you don't have both, it's like standing on one leg." If hypertext is what takes you from a document to a related passage, then transclusion is what takes you back and shows side-by-side all the texts that reference a passage.

Nelson -- a Swarthmore philosophy major with a master's degree in sociology from Harvard -- offers an example from written history, which can be hard to follow because of all the various, interwoven threads. Suppose, for example, that you are reading a history of World War II and come to a section about Yalta. The same section (which might include photos, video, and sound segments) would also form part of the biographies of the three world leaders present: Roosevelt, Churchill, and Stalin. You could leave the Yalta discussion to follow any or all of those threads.

The problem with the existing WWW, Nelson asserts, is that the relationships between such parallel threads cannot be seen; they are lost in "spaghetti hypertext." Nelson also believes that the combination of transclusion and hypertext can solve another existing problem of the WWW -- that of copyrights. Information from many providers (e.g., magazine articles and encyclopedia entries) needs to be readily accessible and linkable to realize the full value of hyperlinked information. Those whose business is information, however, won't provide it if it can then be recopied indefinitely for free.

One approach, often used, is to prohibit or surcharge such redistribution. But another is Nelson's proposed method: a doctrine of "transcopyright" in which substantive documents are seamlessly created by employing links to other material. Rights-holders of such material would allow links in exchange for payment upon use, i.e., they would be paid only when a reader actually links to (and retrieves) a copyrighted passage. This structure allows new work to be created by using original sources while maintaining the context. Nelson is currently working on a "nanobucks" method of payments, which would be needed to implement such a rights-payment system.

Nelson holds that even the existing hypertext-only WWW needs a concept of "committed publication" to ensure that information will be present five years from now, just as a conventionally published source is. No one can make enduring hypertext documents if half the links will be moved or deleted within six months, as is often the case today.

From Xanadu to Jipangu

Nelson has led, or inspired, several unsuccessful attempts at implementing his vision as part of "Xanadu," his trademarked name for a hypermedia information clearinghouse. The largest effort was a four-year, $5-million project sponsored by Autodesk, which was eventually spun-off in 1992 when a corporate crisis threw all non-essential projects over the side.

In the wake of his ill-fated efforts in the US, Nelson came to Japan in late 1994. "Basically, I'm in Japan to get funding for the new Xanadu," acknowledges Nelson, because the Japanese have "more money" and are "more sympathetic." (A harsh April 24 article in the Wall Street Journal was subtitled "Japan Gives Hero's Welcome to a Man Too Eccentric for Silicon Valley.")

Nelson in Japan is an unmodified import. He has only smatterings of Japanese language, history, and culture, and he still uses terms like "nanobucks" instead of "microyen." Thus far, he has worked for a year with the "HyperLab" group in Sapporo led by Professor Yuzuru Tanaka of Hokkaido University, who is best known for his IntelligentPad Consortium. Now that the HyperLab has been disbanded, Nelson has a visiting professorship at Keio SFC under the sponsorship of Professor Hajime Ohiwa.

Japan has given Nelson a new start, both in terms of implementing Xanadu and of spreading his vision. For the former, he hopes to get grants from the Japanese computer industry and use students to implement a demonstration text-based transclusion content server at Keio. He's more likely to be successful in information transfer to Japan, spreading his vision to new audiences -- at least those fluent in English. Nelson will teach two classes at Keio's SFC next fall: one on information media, the other a graduate course entitled "Cinema of the Mind: Design of Software and Interactive Objects."

The future

Nelson laments today's computers and Web browsers as crude approximations of what could be. He briefly musters some enthusiasm when I tell him that some US elementary schools are teaching HTML rather than BASIC programming. "I approve, I approve," he says, then qualifies his endorsement: "Though HTML is so awful it's like" -- he sighs -- "it's like building a diamond palace all these years and having someone build a Quonset hut."

Nelson's unfinished software palace is named after the mythical palace in Coleridge's famous poem "Kubla Khan." ("In Xanadu did Kubla Khan, a stately pleasure-dome decree.") It was also the name of the castle of the fictional Citizen Kane, in the film of that name by Orson Welles. Indeed, Nelson, at times, fancies himself another Welles: a director of work performed by others. His major problem is that the closest analog to the director of a software product is lead programmer or perhaps project manager, roles for which he has neither the skills nor the disposition.

The harsh truth is that a shipping product has yet to come from the efforts of the man who was writing about linked text 20 years before HyperCard and 30 years before Netscape. With a bit more discipline, business acumen, and programming expertise, it is conceivable that a small piece of his 30-year head start might have been translated into a Netscape- or Yahoo-type IPO that could have funded Xanadu into Nelson's golden years.

Instead, reality has caught up with Nelson's vision of the 1960s. Even his place as a hypermedia visionary has been pre-empted by such rivals as Nicholas Negroponte, MIT Media Lab director, author, and Wired columnist.

The Gary Wolf Wired article that Nelson hates so much labeled Xanadu "the longest-running vaporware project in the history of computing." And Xanadu's central problem remains: Because Nelson still relies on others, there's no guarantee that the efforts of Japanese programmers will be any more successful than those of programmers in Pennsylvania or Silicon Valley. Yet the failure of the implementations does not disprove the power or accuracy of the underlying conceptualization. The issues that Nelson has been thinking about are central to the future of an information society.

"I am a failure according to the standard I set for myself, so far," admits Nelson. "But I am unhumbled, and the confirmation of my ideas left and right and sideways astounds my enemies and only makes me feel sardonic.

Joel West is a researcher at the Center for Research on Information Technology and Organizations at the University of California, Irvine, and a former columnist for MacWeek magazine.

For information on Xanadu and assorted other items of interest (such as Nelson's retort to the "Curse of Xanadu" article in Wired Magazine), see http://xanadu.net/the.project.