Skip to content
Thoughtful, detailed coverage of everything Apple for 34 years
and the TidBITS Content Network for Apple professionals
Show excerpts


Adam Engst No comments

The Abstract

Ted Nelson’s worldwide open-hypertext-publishing network, Xanadu, has once again been delayed. The version described in Literary Machines 87.1, etc., has been completed, but put on the shelf due to the absence of some key software mechanisms. The new prototype of the single-user back-end server software is in Smalltalk that will compile down to C to run on essentially all types of machines. That’s the nitty-gritty of the keynote lecture at the first stop of Ted Nelson’s 1990 World Tour (complete with beautifully embroidered black satin jackets), the ‘Multimedia 90’ conference, held in Linkoping, Sweden on September 10th.

Ted Nelson: "In 1987 […] that small fraction of the computer field that knew of Xanadu was very much astonished when they heard that the AutoDesk Company [57% world market share in CAD programs] had actually bought the project, and they’d be even more astonished if they knew how many millions of dollars AutoDesk has put into it since, which I can’t tell you but it is ‘several.’"

They now work on performance and related parameters, so that online deliveries might take place "while the user is still awake." The FEBE (front-end-to-back-end terminal access) protocol has yet to be finalized though. We’re to expect a LAN-version of the xanalogical storage server to be introduced on the market in 1991, with a few front-end programs available from AutoDesk, Inc. (the Macintosh version is being written by Mark $ Miller, so we’re apparently in good hands).

The first public-access Xanadu vending point in Palo Alto in ’93 will be followed six months later by a sister installation at Chico State University, then in some yet undecided "Country Two," in few more American states, then worldwide.

Adam Engst No comments

What is Xanadu?

Ultimately it may take an astrologer or a sun-spot specialist to find a plausible explanation for the remarkable two weeks in the fall of 1960 when Ted Nelson figured it all out. Because that’s when he first defined what may eventually be recognized as the true beginnings of the coming new paradigm, The Age of the Unified Data Structure.

The Unified Data Structure is an entirely new world-class paradigm all of Nelson’s own doing, even though his life achievements up to now have mainly consisted of making visionary waves, giving new meaning to the term ‘vaporware’ and siring probably the most stolen book in history [‘Computer Lib’]. He’s also know for generally muddying the clear minds of inexperienced programming youth. Some may recall a similar accusation that once did in Socrates, bringing him the death sentence in due democratic process by his peers. Or were they really his peers?

Had you been reading this in Xanadu chances are that you’d never finish the rest of the sentence, instead zooming off to dictionary entries on Socrates, source writings on Athens democracy, and collections of commentaries by later contributors. All that and more, the entire written, whispered, telegraphed, and filmed record of the civilization as we know it, instantly available at the fingertips from your own Xanadu home terminal or from a nearby Public Access Xanadu vending store at Desolution Hwy and Fifth.

Because that is exactly what Nelson’s paradigm promises: the tablets of Babylon, the scrolls of Alexandria, the NFL polls of all seasons, down to the preserved napkin-doodles of Einstein, Curie, and Springsteen, all in one LOGICAL, easily accessible place at the end of an existing-bandwidth telephone wire.

That’s Xanadu in a nutshell, and it finally appears to be on the verge of fulfillment after 30 years spent in the realm of gee-whiz ideas. Moreover, what it will eventually confront us with will be an entire new type of literature, a "transclusive fragment writing and publishing system," first defined in those fateful weeks in 1960.

Adam Engst No comments

The New Literature

And what are those mysterious ‘transclusive fragments?’ Ted Nelson has a definition ready for the term he coined two years ago; finally giving The Vision the right generic name. Transclusion is a way to include, to quote, parts of a document without losing its current (or any subsequent) contexts, and without it becoming a physical part of the new text (which could be a movie, hyperfiction document, you name it). In this fashion one might see all newly formulated or recorded texts, data, sounds, pictures as future ‘boilerplate paragraphs’ or fragments, available for viewing, digesting, and transclusion in new works.

And then these fragments will be available cheaply, instantly, and in principle to all, because there will be no one deciding who might or might not be a worthy commentator. In present-day times the possibility of quoting, adding to, or paraphrasing someone else’s work is always a function of access, time, and effort spent searching for the relevant parts, a process that by its very definition limits the possible number of contributions and contributors. It doesn’t have to be that way.

Consider literature. "There is this incredibly powerful instrument called ‘literature’ that was invented long ago, which we don’t see, don’t recognize how powerful the design [of] it really is, don’t think of it as a system, because it is THAT good, we just say ‘oh, that’s just the way it is.’"

But what is this ‘literature?’ "It is a system of interconnected ideas," the accumulated record of humanity, pile upon pile of writings, from the earliest of times. A record that each subsequent generation builds upon, indexes, nails on the doors of cathedrals, abstracts, rearranges, burns at the stake, folds, spindles, and mutilates. Of this literature we’re usually only aware of that thin slice that we’re physically able to interact with, pore over despite overdue notes, make comments in the margins of, wrap a fish in, feel offended on the subway by, clip, file and forget. Nominally it also chiefly means handling documents made out of paper.

Now, when Ted Nelson says ‘literature’ he "doesn’t mean paper, paper documents, and he doesn’t mean TEXT either." All of today’s "halfway" (information-handling) systems work on the assumption that paper is the basis and the desired end result. Nelson thinks of paper as "just an object that [some] information has been sprayed onto in the past […] In today’s offices you’ll get a printout at the end and then some secretary will go over and put some little white paint on something that’s wrong and correct it because getting that paper right is regarded as the objective. And that means that the computer files are never correct, they are always an approximation." Alas, "as long as the paper-sprayed version of a document is seen as the final destination no one really cares about keeping the computer versions of the same information canonical or correct."

Adam Engst No comments

Xanadu Publishing

Then there is the problem of the many modalities available for presentation. Many are available, but none are on speaking terms with each other. Text documents are those made up of words on paper. Motion-picture documents, which we call ‘movies,’ consist of picture sequences that have been recorded on film. Sound documents, which could be words and melody, mumbled by a voice to music on an LP, all are different modes of conveying the information that they contain. Still, all these belong to the same "word-picture-continuum" and to Nelson are of one realm. Therefore we need to have facilities to be able to treat them as such. "That means a paradigm shift which in turn means our being able to deal with change in a new way."

As far as paradigms are concerned, Tomas Kuhn’s work, ‘The Structure of Scientific Revolutions,’ has always fascinated and influenced Nelson. Kuhn tells of "the real arguments between scientific opponents being all about paradigm boundaries. If one sees an existing paradigm as a coordinate space, a finite area, then a radical new idea may be perceived as a paradigm threat, and the distance between the old and the new one termed ‘the paradigm gap’."

Consider WYSIWYG [What-You-See-Is-What-You-Get], "the most inane propaganda, the foolish, defensive response to tie a computer down, the ‘paper simulator’ used to enshrine two-dimensionality of paper on a computer screen" [right on, man!]. By recognizing the limitations of the existing paper-as-record paradigm we prepare ourselves for the coming new literature, one that’s accessible in a uniform and painless way, one that allows us to contribute to it on equal terms, rather than those defined by the technological constraints of production and distribution.

In open-hypertext publishing "anyone will be able to add [publish] a document which links to or quotes from any other [existing] document. Freely. Anybody, or else we’d have to decide at the system level who would be a worthy contributor and who would not, and neither you nor I are fit to decide who that might be. The only alternative is to say that everyone is a worthy contributor, everyone’s contributions are in principle welcomed."

And those contributions may then be in the form of the contributor’s own choosing an essay by someone enhanced with voice comments (How? That’s a front-end input problem.), a video sequence accompanied by blow-ups with notes, a diagram attached to a screenful of data, pointing out your own (r)evolutionary insights, all instantly available on the network from the moment they are published.

These contributions will be available as an ordinary byte-stream, easy to distribute at the speed of the delivery network of the day, which is bound to get faster and faster as technology progresses. Available in a network that might eventually contain most of our ever-recorded intellectual heritage, that might grow to allow unlimited number of simultaneous users, consist of unlimited number of servers, documents, links, transclusions and fragments requested. And all of these fitting within the logarithmic-shape ‘soft corridor’ [LM 87.1 4/2] of the performance degradation curve, so that delivery times will NOT increase proportionally with the size of the ‘docuverse.’

Indeed, if delays doubled in step with the doubling of the available document mass, "maneuvering through this vast and forever growing forest of vines" would become unthinkable. "The way that this curve deteriorates is a fundamental point which had to be addressed in the initial design of the data structure and the algorithms." Thus Xanadu has been "designed backwards from the performance requirements of [such a future] network scale-up" [with allowances for additional delays from servers in space, where "speed-of-light considerations become significant" – LM 87.1 2/57].

Closer to Earth, any published (or MADE PUBLIC) document will be accessible almost instantly from any Xanadu public access location, or from any connected terminal of a suitable type. Obviously, some general-purpose, relatively unsophisticated home computers might be able to run a front-end to Xanadu, but be unable to handle all types of documents (such as animated video). Still, one would most certainly have an option to display named stills from linked video sequences along with the streaming-text data on the same monitor.

Or, rather, on a high-resolution TV screens. Upon taking a college course in ‘Computers for the Social Studies’ during those weeks in 1960 Nelson discovered that "they’ve got it all wrong, these were not some ‘computer terminals,’ these were great MOVIE PROJECTORS, behind whose screens one could create chambers where all the thoughts could be found."

Indeed, the world of movies has a lot in common with that of software design – the latter in itself a highly structured form of creative writing. To be exact, Ted Nelson considers software design to be a branch of cinema. "The cinema-analogy is not an analogy, it is a statement of fact. Software design ought to be taught in film schools. Do you know who’d have made the greatest software designer of the century? Orson Welles, no doubt about it, if he’d understood what it was about. Because writing software requires cinematic imagination with the grasp of the possibilities of writing, a grasp of the possibilities of diagrams, a grasp of the possibilities of animation, a grasp of the possibilities of interaction. And Welles was a superb writer…"

Adam Engst No comments

Setting Up a Stand

Back to our open hypertext publishing. "The notion of a [clearly delimited] document is an important one, really a social and psychological mechanism, fine, we keep that because literature is a system of documents which works. Xanadu will provide the feeder, storage and delivery mechanism that will enrich and electronify this system, with linkage and transclusions providing a representation for the previous implicit [idea-inter-] connections. Before we could say ‘such and such author has said so and so and now I would like to show why and where she is wrong,’ but now in Xanadu you can simply add ‘such and such author has said it’ and bingo!, you can go there and see it right away." Indeed, he thinks of Xanadu as of "that magic place of literary memory where nothing is [ever] lost."

Among the most important aspects of the system is the automatic royalty due on every fragment delivered. "Every document will contain a built-in ‘cash register’ […] but the system only works if the price is low. If the price is high then different users will [use and] hand each other dated [paper] copies. If the price is low it’ll be more convenient for each user to get [same] material anew from the system." Indeed, the cost of fetching and reading a document from the system should be minute in comparison with other methods. And the royalties for accessing that document will be advanced to all the authors of there transcluded fragments, if applicable, in proportion to the byte-content of their respective contribution.

In fact, the very act of ‘publishing a document’ will mean signing a [written] contract with a Xanadu storage vendor, in which the author (i.e., the publisher) explicitly gives permission for anyone to link to, to transclude his or her material freely. Nelson explains that "you have no control over that. However, you have absolute control over the integrity of your document and you can give instructions to the reader as to how they should view it and so on. Of course, since it is sent down the line to the viewer we have no idea whether they’re gonna do that… but that’s OK, the whole point is they’re buying the rights [to view it] every time."

When an author publishes a Xanadu document, he or she pays a small fee to a Xanadu storage vendor for three years’ minimum storage on the disks (on three different servers, for backup and mean distance content distribution reasons). The author decides what gets published, when and where. The author also bears the sole legal responsibility for that publication’s content. If the document includes something that "wrongs other people or wrongs the government, breaks the law, [then it is you, not the vendor] who gets caught." The vendor’s legal position is that of "a contract printer’s or a truck driver’s."

So how does one become a storage vendor, which is almost like getting a license to print money, anyway? The Public Access Xanadu organization, which Ted Nelson still owns, will empower national licensing organizations, which will in turn license (or franchise) individual operators, the storage vendors, franchising being the fastest method to expand without losing control of an enterprise. And here’s where the magic ends and real life begins: "to set up a Xanadu stand you’ll have to put up [some] $200,000 and then HAVE TO WORK PERSONALLY in the stand, 10 to 12 hours a day… we’re gonna go strictly by McDonald’s rule (of personal daily participation by the owner). Different places will handle the problem of food and snacks differently though… also my lawyer reminds me to tell you that this is not an offer to sell, merely a conjectural discussion."

Though "the objective is to create one mighty server for the whole world" it by no means follows that all the servers on the network have to be alike. On the contrary, many different types of servers will be possible, and many will be present: "computers that are set up to deliver certain kind of things, render-servers for graphic images, file-servers for the normal documents and so on," all running the same back-end feeder software, delivering fragments across the network, keeping track of dues. Nor will the Xanadu organization be creating/publishing the literature, filling the network with the food for thought and income-fodder. For that individual entrepreneurs will be needed.

If a future Xanadu vendor believes there is better return in, say, deliveries (sales) of weather-data, fine, let his set up say, a ‘Boreas Real-Time Weather Server’ on the network and start courting weather-data producers to make their results available to the public by publishing them on his server. Then the vendor can attract users of such data, and get them to request the data at whatever intervals they might require, for whatever purposes they might have, in whatever forms or contexts they might desire.

Thus a following flow of income could be envisioned (provided that there is a market demand for said type of data): owners of the weather-images become publishers for a fee proportional to the physical size of their data on the storage vendor’s media. The storage vendor will wish to maximize his sizeable initial investment by making his own premises attractive for the public to visit and appealing to prospective future publishers, who are looking for suitable/genre-specialized storage sites to publish at/rent space from. It is in the vendor’s self interest to try to find potential users for the deposited weather data and to promote use of them, since ultimately he’ll be receiving a percentage on each and every fragment sent to and from his server. Nothing, of course, hinders the publishers from promoting use of their data themselves. The publishers receive royalty on each fragment delivered, proportional to the requested fragment’s size, which accumulates in their account, thus covering the costs of publication and storage and, hopefully, making a profit. The users, finally, get to view/use their data and have a shot at subsequent (part-)royalties on any material that they elect to enhance via linkage and/or publish themselves (for a fee, etc…).

Furthermore: any user without access to a personal terminal will be able to open an account at a local Xanadu vending stand, with facilities for browsing, reading, viewing and printing out the requested fragments (the facilities meaning primarily high resolution, high quality, high speed, ergonometric terminals and peripheral equipment in a "pleasantly painted," futurico-spacey setting, "the bridge of the Enterprise, […] with a pleasant helper in a polyester suit nearby" [not joking]). The monthly bill will then consist of a basic fee, as well as fees for connect time, data delivery (data delivery will include royalty on every fragment), storage fees (if Xanadu disks are being rented) (for the deposition of private data, mail, etc.), and possible publication fees, MINUS royalties (if publications have been read, linked to, or transcluded).

With the system not yet in existence it is difficult to predict the monthly cost for a Joe or an Adina User. Still, as Nelson repeatedly points out, the system has to be affordable to the general public. He’s not worried about lack of potential users either; "his problem is with dealing with the demand [that] he already has… 100,000 people out there who want it tomorrow, TOMORROW. The first XU stand will only have 30 ports [modem lines, with another perhaps 20 terminal points inside the store], and in six months [the network] may grow to at most 500 ports, 1000 ports, which is not enough to service the people he already has, already wanting the service, and certainly not enough to service the number of people who will want it by then." To be exact, "there are more than 50 people, who have already paid 100 dollars each for a Xandle, a user-name on the network" (mentioned in LM 87.1 0/-10), the very same one that has yet to come into being, and then "may yet turn out to be a flop."

Similarly with the critical mass of documents… there is already so much available online in existing electronic networks. Still, he’ll be out there, "preaching and proselytizing to potential publishers, trying to find the most leverage in terms of getting it off the ground. One group [that] he’ll be approaching will be the free-lance photographers, because here is a group [of people] that have a lot of bits to distribute and no existing channels except for magazines. So they have to go through editors, spend a lot of money making portfolios to leave with editors for a time, and maybe the editor looks at it and maybe he doesn’t. So Xanadu publishing gives them an immediate new way to get their photographs out there where other people can see them." Camera owners, do take note.

Adam Engst No comments

PAX Front End Demo

That said, we were then treated to a quickie demo, "made few days ago," in MacroMind Director (I think), projected off a Macintosh with color screen. First we saw how an animated sequence of a 1960 Parallel Textface version might have looked on upper-case alphanumerics-only screens of that time [LM 87.1 4/76], then a static view of a later QFrame, edge-linked text-tiles [LM 87.1 4/77]. Then "a ‘rigged demo,’ where only certain parts function, so you have to know where to point and click; a quickie, very rudimentary demo of a [modern] Xanadu front-end, of which many visualizations are possible." The initial image showed three rectangular buttons arranged horizontally along the upper part of the black viewport, labelled Journal, Projects, and Publications, as well as three vertically placed ones along the left edge, labelled ToDo, Schedule, Coresp (spelled that way).

Clicking on the Projects button on the screen made a menu unfurl, displaying the following items (invisible from any distance, had to work real hard to get it all down; may not be exact):

Show Docuverse
Show Personal Collection
Select Endset
Show Linkset
Renegotiate Specs
Show Individual Link

Next, clicking:

  • ‘Show Docuverse’ displayed a space darkness, filled with small white rectangles of various (4-character-cell at best) sizes.
  • ‘Show Personal Collection’ showed a subset of that; i.e., most of the white specs disappeared.
  • ‘Select Endset’ opened a white square window halfway down the screen, with the name of the selected document (one of possible list of docs?) and the name of author in smaller, separate side rectangles. A third windoid still, below the square one, contained a type of document ‘Technical specifications’ or something similar.
  • ‘Show Linkset’ displayed a collection of thin blue lines, arranged in a fan from the document’s square to the right-hand edge of screen. A small rectangle, superimposed across it, told of the number of recorded links, some 44,600-odd.
  • ‘Renegotiate Specs’ (specs not supplied) made this fan thinner, down to some hundred lines. Finally, clicking the
  • ‘Show Individual Link’ button and then on one of these lines opened another windoid below the main square one, with the linkee’s name and the type of link made to the original text (‘technical comment’). Now, presumably, one could request the comment or some additional information about it (size, date, etc.) from the back-end, had there been one in existence nearby (and if the linkee’s name sounded familiar? trustworthy? or whatever-the sublimal-feeling-selection-method-of-the-day).

That was it. The concluding screen showed large bluish PAX (Public Access Xanadu) letters, with a zooming take of a street in perspective inside the ‘A.’ The ‘PAX’ was framed by the words ‘Welcome Home’ above and ‘Everyone’ below. Weeelll, maybe. Then again maybe not. I wouldn’t know, I’ve got to keep an appointment for a fitting of that damn polyester suit.

all double-quoted contributions by Ted Nelson (29%)
all [LM 87.1 chapter/page] pointers refer to the 87.1 edition

Adam Engst No comments

Further Reading

Literary Machines, book by Ted Nelson describing the Xanadu concept & methodology, latest edition 90.1, new edition coming shortly, postpaid US$ 25 US (US$ 40 foreign) from Mindful Press, 3020 Bridgeway #295, Sausalito CA 94965

Literary Machines, the 87.1 Macintosh hypertext edition on disk, available from OWL International Inc., 14218 NE 21st Street, Bellevue WA 98007

Computer Lib/Dream Machines, by Ted Nelson, a ‘1987 revised & updated’ reprint of the original 1974 edition, Tempus Books/Microsoft Press

For information on the forthcoming Xanadu software from AutoDesk contact Xanadu Operating Company, Palo Alto CA, tel. [+1] (415) 856-4112

To get on the mailing list for PAX developments write to Public Access Xanadu at the Mindful Press’ address above or contact their EC representative Elisabeth Davenport (c/o Department of Information Science, University of Strathclyde, Glasgow, UK tel. [+44] (041) 552-4400 x3700, fax (041) 553-1393

Explicitly referred to in the lecture:

Alfred Korzybski (an eccentric philosopher whose best known work is ‘Science and Sanity, An Introduction to Non-Aristotelian Systems and General Semantics,’ (1933), last ed. Boston, 1980).

Tomas Kuhn (introduced ‘paradigm’ in science; wrote ‘The Structure of Scientific Revolutions,’ Chicago, 1970)

K. Eric Drexler (shortly re-joining the Xanadu development team, wrote ‘Engines of Creation, The Coming Era of Nanotechnology,’ 1986)

Buckminster Fuller, Bertrand Russel – his teen age idols

Xanadu, XU, Xanadu Stand, Parallel Textface, Qframe – trademarks of XOC Inc.

Macintosh – a trademark of Apple Computer Inc.

Snacks eaten by the author during writing supplied by Goteborgs Kex AB.