AI Online - October 1988 column as published in Information Today

Graham Bell, Edison, David Sarnoff... Ted Nelson?

by Daniel Gross

There are some people whose achievements give them an indisputable place in the Information Age Hall of Fame. Even those who are not the insiders of the information industry are quick to recognize that electricity, telephony and television have radically changed the course of civilization.

Sometimes it is more comfortable for us to believe that these discoveries or inventions were inevitable, part of some kind of technological Manifest Destiny. Someone like Thomas Edison had but to tap the collective subconscious to come up with his next gadget, we think. But it is more likely that a certain blindness of hindsight makes us see it this way.

The telephone is a masterpiece of design simplicity. Its superiority over the telegraph is obvious. Telephones vary widely in shape and functionality, but anyone above the age of six can figure how to use nearly any telephone in a few seconds. Why would we wish to teach anyone Morse code when this alternative is both easier to use and more efficient?

The world of personal computing, however, seems bereft of this basic logic. The continued use of keyboards means access to computing requires the development of a special skill -- typing. As if this initial barrier were not enough, software developers frequently chose to organize and store information using paradigms which simplify the computer's work rather than the user's.

While the advent of pointing devices -- mice, touch-screens, light pens -- and the so-called intuitive user interfaces of the Macintosh and other windowing environments alleviate the symptoms, by no means do they constitute a cure.

Many pundits of the personal computing "revolution" would have us believe that the cure is computer literacy. Other pundits are more lucid. MIT professor Joseph Weizenbaum once made the accurate, if somewhat cynical, observation that "computer literacy is a disease that was invented when it became necessary to market the cure."

In the one historical parallel at our disposal, we note that Gutenberg's moveable type did little to promote the learning of Latin. On the contrary, the newfound economies of scale in book production led to publications in the lower-class, "spoken" tongues -- such as French, German and English. While those who wished to read still had to learn, the aristocrats' monopoly over literacy was broken. The medium met its users halfway -- a happy compromise.

But Ted Nelson would argue that personal computers are more than a medium. They are creative tools, or more precisely __Literary Machines__. This, incidentally, is the name of Nelson's book, which describes "the legendary and daring Project Xanadu, an initiative toward an instantaneous electronic literature."

Theodor Holm Nelson coined the word "hypertext" over 25 years ago. Although the concept and implementation of hypertext have become fashionable only in the past five years or so, Nelson has spent most of his adult life confronting a very dull industry with a very sharp pencil. In collaboration with Douglas Englebart, the inventor of word processing, Nelson gave the first demonstration of a working hypertext at a time when computers and white lab coats were still by and large inseparable. The demonstration also featured a little makeshift device which rolled across the table and moved a cursor on the screen -- the first working prototype of the mouse.

Xanadu is the trade-marked brand name of Nelson's own hypertext product, and his company, which recently licensed some of its hypertext technology to Autodesk Inc., one of the largest computer-aided-design software developers in the world. A product is expected some time in 1989.

Lotus Development Corp., makers of the 1-2-3 spreadsheet, has begun to aggressively promote its Agenda hypertext product. Apple is so bullish on hypertext as an organization scheme that it gives away HyperCard with every Macintosh it sells. OWL International now has versions of its Guide program, the "purest" implementation of hypertext of the three, on both the Macintosh and the IBM PC. A version of __Literary Machines__ has even been hypertexted to be read with Guide.

In spite of the sudden interest in the technology, it seems that Nelson's original point is being lost in the shuffle. __Literary Machines__ is as much a political manifesto as it is an attempt to explain the basic mechanisms of hypertext. It is also a painstakingly precise description of the Xanadu system's underlying mathematics.

The basis of hypertext is the notion of links. Information is stored in small, easily manipulated components which can then be interrelated. A specially-designed user interface lets you navigate through these pieces of information in much the same way you might jump from stone to stone to cross a stream. The stones are data, and the jumps are links. The stones may have nothing in common, but you are free to jump between them as you please. Of course, a link may be made which has no value or meaning -- the equivalent, perhaps, of falling into the stream.

Xanadu's handling of links is extremely flexible. In particular, links can be made to and from the links themselves -- in other words, links are treated as data. This makes it possible to give the links types, values, or meanings. If meanings are applied in a consistent manner the hypertext becomes a semantic network, which I have described in previous columns as an efficient knowledge representation scheme.

Ted Nelson is a very patient man. In the early 1960s he demonstrated the technological feasibility of information organization schemes for which the need is only now becoming apparent. That original demo addressed the two key problem areas I described above as the greatest barriers to democratic computing: the user interface and the organizational paradigm applied to the data.

The most significant difference between Nelson's approach and others' is ultimately one of intention. Xanadu magnificently embeds its intentions in its design assumptions. While Agenda, HyperCard, and Guide fall into the category of "personal information managers," Xanadu is basically an operating system for an information service. The book even includes preliminary contracts for eventual Xanadu franchisees -- people or organizations who will make information publicly available in conformance with the Xanadu architecture.

In effect, Xanadu is to information what the old AT&T strategy was to telecommunications: universal service. Unimpeded, unmitigated, unadulterated access to information. This goal fits in with Xanadu's's hypertext link numbering system which gives each byte of data in the entire "docuverse" a unique address. This addressing scheme makes it possible to create a link between any two items on any two host systems in the Xanadu network.

While no single aspect of Xanadu can be described as artificial intelligence, the resulting web of interconnected data could very well be described as smart information. Knowledge, in other words, which appears to the user to know a little about itself. Scroll through some text, use the pointing device to highlight a person's name, press a button, presto! You are now reading biographical information about this person. Point to a historical reference, and in a puff of smoke you are reading news reports of the event. More links are created and more information is added as time goes on. By an organic process, a sort of cybernetic Darwinism, the information gets smarter.

By his own admission, Nelson's current role in Xanadu is more that of an evangelist than of an implementer. It is difficult to say for certain whether the two hats can ever be worn separately in the world of high-tech. He is eloquent, and there can be no question that he believes what he says.

There is really no way to tell whether Nelson is benefitting from the transient popularity of his ideas or whether the world is really coming around to his point of view. An old French saying tells us that one can only write obituaries for the famous, and never birth announcements. Hypertext, if it is the basic tool of the information age, is only in its infancy -- so young that no one is sure whether it will stand the test of time.

The Xanadu project faces the greatest of all obstacles: inertia. Why try to understand a new way of dealing with information? Why create a worldwide network for information? Do we really want everyone to have access to all this knowledge? What's wrong with the way we do things now?

Nothing, if you like Latin.