Xanadu Light

Copyright (c) 1990-1993 Adam & Tonya Engst.
Originally published in TidBITS#204, 29-Nov-93

The high point of Hypertext '93 was of course the talk given by Ted Nelson after the reception in his honor. Nelson is a thoroughly engaging speaker, and he devoted much of the first half of his talk to providing the audience an overview of the 32-year history of Xanadu, Nelson's electronic publishing world view. I won't attempt to summarize that history since a bit of it exists in TidBITS #30 and Nelson's books, including Computer Lib/Dream Machines (one book) and Literary Machines, are required reading for anyone in the field.

What interested me was the reaction Nelson received in the crowd. I don't mean the public questions and comments, but the asides and looks various members of the audience traded during the talk. Members of the hypertext community seem to view Nelson with a complicated mix of awe and devotion (after all, he is the father of hypertext) combined with an almost cruel pity and ridicule. I suspect this mockery, which was seldom voiced loudly, but was evidenced in eye-rolling and smirks, stems from the fact that despite his long involvement with hypertext, Nelson has never shipped a product. Xanadu has been vaporware longer than many of us have been alive. The reaction concerned me, because even though Xanadu has yet to appear, that fact is independent of Nelson's ideas, just as much theoretical physics is more or less independent of practical application at the moment. It may mean that he's a theoretical hypertext scientist, but there's no shame in that. I sensed a vague paranoia in Nelson, but one that is probably justifiable if his ideas have received similar reactions (and most likely, even worse ones) in the past. It's a shame, and let me attempt to convey his concepts in relation to the new Xanadu, now called Xanadu Light. Much of this information comes from the handouts Nelson provided with his talk.

To bring you up to date quickly, it seemed as though the hope for Xanadu lay with Autodesk, the CAD giant that purchased it back in 1988. Unfortunately, after investing five years and five million dollars, Autodesk dropped the project in 1993. Nelson didn't say specifically, but I have the impression that all that development effort remained at Autodesk; all he managed to get back was the trademarked name. In large part because of that, I suspect, Xanadu Light is now based on garden-variety database programs and using the Internet for worldwide access. Nelson mentioned something about searching for stuff via Gopher and then telnetting in or using a dialup BBS to actually retrieve the information - I'm sure a custom front end would appear quickly.

Within Xanadu, people can have three roles - readers, publishers, and suppliers. As a reader, you connect to the entire Xanadu universe by connecting to one Xanadu supplier. You can browse hypertext links indefinitely from document to document. No records are kept of your hypertext trail or of the items you send for, and you can keep what you receive (a receipt token helps you file it for future reference).

As a publisher, you may link to, comment on, or append information to any published document. Quoting documents by what Nelson calls a "transclusion pointer" automatically links your document to the original and pays the original publisher for the data, and although you have no control over who links to your documents, the documents themselves are kept inviolate. Everything is handled by links. You may publish anything within the law (which Nelson notes is going to be a big issue in the future), and you take responsibility for the contents of anything you publish, just as in traditional paper publishing.

As a supplier, you can locate your business anywhere and allow your customers to connect to you in any way. You can charge what you like for storage of published documents and for connection time, and you have complete control over credits and payments. In an attempt to avoid the mega-companies that currently dominate publishing (apparently there are about 40 "important" publishing companies out of a set of some 70,000), Nelson specifically designed Xanadu on a franchise system. Anyone can set up as a supplier with some hardware and a connection, and anyone can set up as a publisher.

In brief then:

The publisher pays for storage, the reader pays for delivery, along with a small per-byte royalty. Nelson recommends rates in the range of 1/10,000 of a cent per byte for text, perhaps one cent per minute of video.
The reader may send for any portion of any document and pays for just that portion, not the entire document. However, since the rates are so low, there's no concept of browsing and then choosing what you want to buy. You pay for everything you see.
Anyone may quote anything in the Xanadu network by transclusion (virtual inclusion - it's a hard concept to convey without an illustration, perhaps think of it as a publish & subscribe type link) from another publisher's document. Royalties continue to flow automatically to the original publisher of information.
Anyone may publish links to anything in the Xanadu network (but remember, original documents remain inviolate, so you don't have to worry about your data being corrupted by virtual graffiti).
Every document has an owner, the publisher, and that person pays for its storage on a Xanadu host machine.
Every link is also owned as a part of some document.
Connecting to one Xanadu node connects you to all nodes, and thus all documents and data objects. This inherently implies some sort of global name space for objects, I would assume.
All data structures are welcome and connectable; there are no closed objects. This will prevent what Nelson calls the "Balkanization" of electronic media, where the data objects are inherently proprietary and isolated.

Copyright always comes up in these sort of discussions about Xanadu, but the system handles copyright and royalties automatically and unobtrusively. Since every document has a known owner, and since there's no reason why you wouldn't quote something as opposed to retyping it (it's thinkable, but I imagine it would become culturally taboo to do so), any owned data will always remain owned. Royalties (set by the publisher) flow automatically from the reader to the publisher on a per-byte basis, and give the reader the right to backup and one printout as well, although there's no reason alternative arrangements couldn't be made.

Xanadu Light, then, is essentially four public database tables, plus content bytes stored in standard and nonstandard files. Each document lists its contents in a public table, and users may query the database using standard queries or SQL queries for more complex searches. As I understand it, some sort of client software would be responsible for presenting this information and allowing you to browse and search among it.

From Nelson's handouts, then, here are the four database tables.

Grand directory of all documents (public table)
----------------------------------------------------------------
Author | Title | Document | Date of     | Owner | Size (may
       |       | type     | publication |       | be misleading
       |       |          |             |       | in hypermedia)
----------------------------------------------------------------

Sequential pieces of a document (royalty bytes)
  Note that a document may include part of any other document,
  simply by including that part in this table.  Permission to
  do so is assured by our publishing contract.
-------------------------------------------------------------
Type  | Owner | Author | Publisher | Where  | Size | Royalty
of    |       |        |           | stored |      | per byte
piece |       |        |           |        |      |
-------------------------------------------------------------

Document's outbound links
  A document may contain any number of links of any number of
  types.  Each link connects to particular sets of bytes in
  this or other documents.  Note that link contained in one
  document may connect material between two others.
---------------------------------------------------------
Type of link | left endset (bytes, | right endset (bytes,
             |  node, document)    | node, document)
---------------------------------------------------------

Document's inbound connections (harpoons)
  This table records all the links and transclusions citing
  this document from elsewhere.  Since these connections are
  made by the choice of others, the others pay for their
  presence in this table.
-----------------------------------------------------------
Type of connection | left endset (bytes, | right endset
(transclusion, or  |  node, document)    | (bytes, node,
  link of whatever |                     |  document)
  type)            |                     |
-----------------------------------------------------------

I realize that's not a totally satisfactory explanation of it all, and Nelson didn't intend it to be. However, I would like to say that this article is a perfect example of what Xanadu would be good for. Rather than try to recreate ASCII tables, I could merely have quoted them so that you all saw the originals, and so that the royalties could go directly to Ted Nelson. As it stands, I'm going to have to hope that this article stirs enough interest among folks who are in a position to help out with Xanadu. For more information and contracts, send a self-addressed, stamped, envelope to:

Xanadu On-Line Publishing
3020 Bridgeway #295
Sausalito, CA 94965 USA

Nelson said he had to give up on email when he found himself with over a thousand unanswered email messages in his mailbox, although I wonder how answering snail mail is any easier - I'd drown if I got 50 pieces of personal snail mail every day.

Among other various comments, two stood out. When asked what he thought of the World-Wide Web, which was developed at CERN in Switzerland and which provides hypertext browsing of documents spread over the entire Internet, Nelson said that he thought it was an excellent step forward, and suffered from only two major problems. First, the Web is not fine-grained enough, and second, you can't follow its links in both directions, which Nelson claims is a necessity. I'm not quite sure how to explain the criticism of the Web not being fine-grained enough; he didn't explicate further. The second comment was classic Nelson. When someone brought up CD-ROM publication, he responded, "CD-ROM is pre- Columbian. When you get to the edge you fall off."

Non-profit, non-commercial publications may reprint articles if full credit is given. Others please contact us. We don't guarantee accuracy of articles. Caveat lector. Publication, product, and company names may be registered trademarks of their companies.

This text was originally wrapped as a setext, and hand-converted to HTML by avatar@xanadu.com (Andrew Pam) of Xanadu Australia. For more information send email with the single word "setext" (no quotes) in the Subject: line to fileserver@tidbits.com. A file will be returned shortly. For information on TidBITS: how to subscribe to our mailing list, where to find back issues, how to search issues on the Internet's WAIS, and other useful stuff, send email to: info@tidbits.com

Otherwise, contact us at: ace@tidbits.com * CIS: 72511,306
AppleLink & BIX: TidBITS * AOL: Adam Engst * Delphi: Adam_Engst
TidBITS * 1106 North 31st Street * Renton, WA 98056 USA