Mediamatic Magazine Vol. 8#1 Avon Huxor 1 Jan 1994

Xanadu: The Conversation of the Digital Text

One persistent idea in science fiction has been that of an enormous computer library containing all the books ever produced in digital form. It would be a contemporary Library of Alexandria, and one that being distributed across the world's computers, could not be so easily destroyed in one act of cultural vandalism.


Xanadu: The conversation of the Digital Text -

Slowly but surely, this vision is becoming reality thanks to the rapidly growing Internet community, one that interconnects millions of machines and users. It is currently possible, for instance, to gain access from your own terminal to Project Gutenberg, a non-profit effort to put as much literature as possible into a machine readable form (Krol 1992). Texts available include the Complete Works of Shakespeare, Moby Dick, Paradise Lost and Alice in Wonderland. Another group has set up a similar service to make available many classic poems by Brontë, Burns, Byron, Eliot, Frost and Yeats, amongst others. Project Dante goes a little further than just containing the canonical text, and contains reviews of the Divine Comedy by various historical (but nevertheless noted) writers.

Library of Babel

One ongoing project that is often put together with this idea is Ted Nelson's Xanadu. He himself, however, counters the view that the grand vision of Xanadu is to build a database the size of the world, based on existing printed texts. It is, rather, a publishing network for anyone's documents, in which users can combine and link together in whatever way they see fit (Nelson 1988). Xanadu is an extensive and ambitious application of the idea of hypertexts: Indeed, the term hypertext was coined by Nelson himself. In hypertexts, the text is no longer single, closed and stable, but instead various texts are linked by pointers, allowing one text to comment on, explain, critique, and relate to other texts in whatever ways seem interesting. It is a system that is more concerned with marginalia than the canon, encouraging the questioning of the texts that lie within it.

One problem with this vision that has been exercising Nelson recently is the immense amount of material that would be generated. In Xanadu, much text is created in response to existing material. They could be footnotes, quotations from, or modifications to, existing material, and this process could apply to the comments as much as the original texts. Of course, not much can be done about this if the aim of the system is to encourage reaction from readers. But the growth of textual material has another source, one that Xanadu specifically addresses. Nelson (1982) gives the example of his great-grandfather who (allegedly) considered that the phrase sea of troubles in Hamlet should have been siege of troubles. In traditional publishing he could publish a revised edition with this modification. That is, there are two equally sized (and very large) texts to be stored. In a digital world, both the accepted and his great-grandfather's versions could be made available on line (would Project Gutenberg be pleased to hold great-grand-daddy's? Would they throw out Shakespeare's Hamlet in preference for The Spanish Tragedy?). This basically doubles the amount of storage space needed. Of course, one solution is to reject the new version, keep to the canon. In Xanadu, another approach has been found, the revised version would consist of pointers to the accepted version, plus a note about the modification, a note that is not aimed at the reader, but allows their computer to reconstruct his great-grandfather's version from Shakespeare's. Clearly this takes up very little space. Seen in such a fashion, Xanadu seems to risk becoming Borges' Library of Babel, in which infinite material exists, but it is more often than not comments upon comments, and slightly different versions of texts. For the Library of Babel itself contains both several hundred thousand imperfect facsimiles: works which will differ only in a letter or a comma (Borges, 1970) and amongst much else the Gnostic gospel of Basilides, the commentary on that gospel, the commentary on the commentary on that gospel, the true story of your death, the translation of every book in all languages, the interpolations of every book in all books.


The true impact of a system like Xanadu, however, could be more prosaic. The exchange of comments between living authors and critics that the systems facilitates begins to appear like a recorded form of temporally stretched speech. The accumulation of remarks and comments becomes a spatially and temporally distributed conversation. Indeed, with the impeding introduction of the technology to convert speech into text, speechwriters - actual recordings of our spoken comments - might become possible. Miller et al. make the point that when Xanadu was begun, it was seen in terms of extensions to writing, but that increasingly, they recognise that the final systems will embody many features of speech. What Xanadu really offers is not a just new form of publishing, but a new form of conversion. Unlike speech, it is a conversation mediated by digital words. By stringing together a series of existing text fragments, a reader can create a new discourse. It is as if we have a new vocabulary, not of individual words, but whole sets of them, that represent and embody an idea in a fashion not unlike the way words are found in a dictionary. Xanadu would allow complex meanings to be written and read in larger chunks - whole sentences rather than words.

Certainly I know from my own writing that one is forced into positions of the absurd. If one uses someone's ideas, one must (and no doubt quite rightly) do one of two things: quote directly and give the source, or rephrase the original author's words into one's own. Usually the original is better, but current publishing convention is unhappy with excessive direct quotation, forcing yet more rewordings. How nice to be able to take larger elements directly and just put them into sequence, just as I put individual words in order to convey a sense. Many might argue that such a position is inherently uncreative, due to the cut-and-paste functionality of the systems described above. To them it seems unreasonable that by restructuring existing material we can come up with anything really new. But, as Ong (1971) reminds us:// from More to Shakespeare, adult Tudor authors turned to collections (e.g. Wit's Commonwealth, 1597, or A Treasury Or Storehouse of Similes) for ideas, phrases, illustrations, and even plots, just as they had done when they were schoolboys. The most resounding and most quoted passages of Shakespeare are generally reworked versions of what anyone could find here. Like Alexander Pope a century later, Shakespeare was less an originator than a consummately expert retooler of thought and expression.//

We would not suggest that Shakespeare's failure to reproduce The Spanish Tragedy was a failure of memory when he produced Hamlet instead. So I make no apology for including such a long quotation, for I could not put it better, and indeed given a wide range of source material the whole essay could possibly be put together from such fragments. There is an equivalent with digital sampling in music: not only simple notes but whole complex musical structures can be assembled together to create a work that takes on an identity of its own, in addition to showing it roots.

So what Xanadu offers is a contribution to the appearance of a secondary orality. Not the obvious return of technologically distributed speech through the telephone, radio and television, but a conversational text. A conversion engaged in through Xanadu, where the machines that comprise it act as a community memory of higher level language, one not of single words but of whole ideas, a language that the human mind alone could not hold. In this sense, Xanadu naturally retains characteristics of the written word. Consider the case of the grapholect known as Standard English (Ong 1982), a grapholect being a transdialectical language formed by a deep commitment to writing. Standard English has access to at least a million and a half words, available through dictionary and thesauri. A simple oral dialect may typically only have access to a few thousand words. Maybe it could be the same now with complex ideas. We have a limited resource in our natural memory, but conversational computer-mediated texts can provide an extended external and public memory space in which they can be represented by standard, conventionalised, but not necessary formulaic, forms.

Just as digital music was initially employed to simulate conventional forms more easily, before developing a sampling aesthetic in contemporary dance music, who can imagine what a fully functional conversational hypertext such as Xanadu could offer for the expression of ideas.

Jorge Luis Borges, `The Library of Babel', in Labyrinths, Penguin 1970.
M.S. Miller, E.D. Tribble, R. Pandya & M. Stiegler, `The Open Society and its Media'. To appear in J. Lewis & M. Krummenacker (eds.), Prospects in Nanotechnology: Towards Molecular Manufacturing, Wiley.
Ted Nelson, `A New Home for Mind', in Datamation, March 1982, pp. 169-180.
Ted Nelson, `Managing Immense Storage', in Byte, January 1988, pp. 225-238.
Ed Krol, The Whole Internet User's Guide and Catalog, O'Reilly & Associates, 1992.
Walter Ong, Rhetoric, Romance and Technology, Cornell University Press, 1971.
Walter Ong, Orality and Literacy, Methuen, 1982.