Tuesday, June 16, 2009

Project Xanadu

This is one of those posts where I really don't want to be mean, but I have to be.

Often the computer systems that we create today don't hold a candle to the things that were envisioned for computers at the very outset. For instance, I'm still in awe of what Douglas Englebart was able to accomplish back in the 1960s. (I've written at length about that in "Why you need VIC CRM".) On the other hand, not all of it was golden, even when it sounds really good on paper.

For instance, take PROJECT XANADU, billed as "the original hypertext project", established in 1960 (which makes this the longest running mostly vaporware project in the history of computers -- and I'll expound on that word "mostly" in due time). I could spend a lot of time telling you about it, or I let its author and architect, Ted Nelson, do it for me:

Mighty big words. And 40 years of effort! When the Israelites gave up 40 years of their lives, the Lord gave them the Promised Land in return. We should expect Big Things of this project. Indeed, Big Things are promised:
"Today's popular software simulates paper. The World Wide Web (another imitation of paper) trivialises our original hypertext model with one-way ever-breaking links and no management of version or contents."
But when we dig through the many, many deep layers of pretention, we learn that Xanadu is basically hypertext and embedding, with micropayments thrown in, ensuring that it will never, ever be used on a large scale.

Nelson is apparently upset that ancient versions of programs don't treat documents the way his vision dictates and that hypertexting doesn't work as his vision dictates and basically that the world of computing doesn't work the way he wants it to. He's got a forty-year old dream which he's been too busy pursuing to use, as the rest of us do daily.

Nelson's vision was revolutionary, 40 years ago. His core complaints with today's computing boil down to obsolete observations. For example, Microsoft Word actually does allow marginal notes. And links. And embedding. Maybe Ted Nelson needs a newer copy. Fortunately, OpenOffice.org does all of the above, and it's free. The world actually works much better than his vaporware vision. Watch the video. Download XanaduSpace 1.0, and don't say I didn't warn you.

Broken Rules

Nelson has a list of 17 rules for documents. (These are also listed in the Xanadu FAQ) XanaduSpace proceeds to break many of them. I'll leave it to you to follow the link and read them. Here are some comments.
  • Rules 1 through 3 are tough to evaluate with version "1.0" of XanaduSpace, since none of the documents reference anything outside of local machine. Rule 3 in particular is highly unlikely.
  • Rule 4 can't be accomplished at all within XanaduSpace, as no editing facility is provided. Instead you must create your documents with a plaintext editor. So much for any criticism about systems that treat documents as paper. Xanadu documents are in fact exactly that... flat text documents. A separate document ("_Connections.txt") contains references. XanaduSpace is a viewer to resolve those references. Well... "resolve" is a pretty strong word. It colors them. Pretty, but most assuredly not the revolution that's advertised.
  • Rule 5 isn't demonstrated in XanaduSpace 1.0. All of the documents are plaintext.
  • Rule 6 is about "transclusions". Now, transclusions are basically quotes. Well... embedded quotes that replace existing text. For instance, I "transcluded" a quote from the Xanadu homepage earlier in this document, though Nelson would likely disagree in that I did it manually. Nevertheless, the effect is identical: I can see the quote; I can follow the link back to the source. Xanadu's "transclusion confusion" is of no benefit other than to obscure what the thing is by replacing the perfectly serviceable word, quote. What would be nice is to be able to reference another file and show the text inline... but you can do that in HTML documents as well by using JavaScript or server-side includes (what Xanadu calls "clinks"). And while it would be very nice to have a simple tag that embeds selected text from another document, that's not what XanaduSpace does, exactly. Really, install it and go look at the source docs in C:\Program Files\XanaduSpace\texts\demo. What you have there is conventional in-line quoting with a separate file called "_Connections.txt" that lists the attributions, which are displayed as colored filmy structures. Transclusions include the entire quoted text from the source doc and text of the destination to be replaced. If the link is broken (i.e. the source doc is changed, then the quoted text in the destination doc is used instead. Not only is this not hypertext, it's not even smart. A changed paragraph in the source doc doesn't automatically update the composite doc; instead the link breaks, and whatever was there (which better be a verbatim quote of the original if you want to maintain the illusion of solidity) is displayed. If you're required to quote verbatim and can't maintain a link if the text is changed, then what's the point of the link?? Attribution should be encapsulated within the source file, so that when you copy it, you don't lose your connections. The _Connections.txt file doesn't even follow the standard described at Transliterature.org. Try it.
  • Rule 7 states that "links are visible and can be followed from all endpoints". This is fine for a severely restricted set of documents, but for something the size and complexity of the Web it's not even a good idea. Really... I want you to imagine for a moment how you would sort out the back-links from, say a Wikipedia page to every single document to which it's linked. That's exactly what's described by this rule, and it's insane. I realize that if the fiction that is Xanadu were made reality, it wouldn't be Wikipedia. But something very much like Wikipedia would have to exist, or the system is worthless. In fact, the whole project, as described, is very much like a Wiki, with the addition of some very bad ideas, which we'll see next. But that kind of confusion doesn't happen with Xanadu, because the links aren't globally shared. Being, as they are, in a _Connections.txt document, they aren't maintained outside of your own little "literaverse", which presumably would be defined by your "Xanadu Server" if such a thing existed. This, of course, means that as you transfer a document to another reader outside of your little world, you must also separately transfer the connections, as well as ensure that he has access to all of the source files if the linkages are to be preserved. Of course, in the demo, XanaduSpace maintains the illusion of unbreakable links in that deleting a source file doesn't result in disappearing text, but that's simply because the text is directly quoted, only to be replaced by the transcluded text. Because _Connections.txt has no source metadata other than the location of the source file and the quoted text, traditional attribution is superior to this weak scheme.
  • Rule 8 states that permission to link a document is explicitly granted by the act of publication, but applying Rule 9, we learn that this permission is contingent on royalty payments. Not only is this one of those Bad Ideas that I mentioned, it presents a legal challenge which I'll address below.
  • Rules 10 through 14 and rule 16 are boring housekeeping. Lotus Notes servers routinely provide all of these capabilities.
  • Rule 15 states that any provider can charge you anything they choose for their services, but those of us with long memories know that Rules 12 and 13 mean you don't know where it's stored. Together, these rules spell, "it sucks to be you." Again, this isn't a big deal if we're talking tiny data worldlets in academia, but connect them into something the size of the Web and we're talking about a system that has nothing to recommend it over what we have, and plenty to argue against it.
  • Rule 17 states that "the Xanadu client-server communication protocol is an openly published standard. Third-party software development and integration is encouraged." Now this one I find intriguing. Look at that standard. (that's right, it's all on the one big web page. Just keep scrolling). It sucks, and I'm not being unkind at all when I say that. Please read it for yourself.
Xanadu a lot of new words. Unnecessary words, to be sure, like flink and clink, but some that are necessary to understand just to be able to show how remarkably bad they are.


For instance, there is the issue of transcopyright. Basically it's a license that says other people can use your stuff, and at that level it's no different at all from the Creative Commons. Quite frankly, in that we already have the Creative Commons, transcopyright is a duplication of effort, and therefore a complete waste of time.

But that's not where transcopyright ends. Integral to the concept is micropayments, which would be automatically deducted from your accounts for amounts below a certain threshhold (five cents is advanced as an example). AND, this micropayment system would be enforced for mere "snippets" of information. This is the legal challenge I mentioned above. It may be a news flash to those who wish otherwise, and who really, really want micropayments for every damned thing, but under the Berne Convention and Title 17 of the U.S. Code, I DON'T NEED PERMISSION to quote a document. I am allowed fair use of that document under certain broad conditions, including news, commentary, educational purposes, or satire. In short, copyright cannot trump my free speech rights. So, for instance, when I quoted the Xanadu homepage, above, I not only didn't ask permission, but before I ever seek permission I will pull down my pants and fart Swanee River.

Now, you could impose a system whereby micropayments are enforced before I could link the page. But that would just encourage me to quote the page conventionally for free, as I am quite nicely empowered to do by the law, rather than link it. Such a system would discourage use of itself, which is double-plus ungood if you're looking to overthrow a paradigm.


Then there is the concept of the "transliterary standard". Ignore for a moment the monumental pretentiousness of the text... this is, after, all written by the man who coined the term "hypertext", so he is allowed a higher level of bombastic declamation than most of us.

But most of the things he describes here are either already available or simply a bad idea. For example:
  • Zotero provides all of the practical "transclusion" capabilities in XanaduSpace. This allows you to pull references from the web (for instance), and include them in Word or OpenOffice.org documents. The bibliographical information and formatting exceeds that described by the Xanadu project.
  • Quotations and comments on a document are commonplace. That's what the collaboration features of MS Office, OpenOffice, or Lotus Notes are for.
  • Ditto for "shared workgroup writing-spaces where different contributions are recognizable."
  • Likewise for "profuse and varied links of many types by different authors, overlapping freely."
  • In fact, all of the what-ifs involving document management are commonplace.
Regarding Better Tools, "pull across editing" is commonplace as well, it's what we know as "drag and drop". In practice, you see something on a web page, you drag it, you drop it in your document. Done. Now, what this doesn't do is provide you with attribution, which is easily fixed by using Zotero. It's not necessary to visually show rubber bands connecting quoted text with the source document so long as it's attributed, and in fact, that would simply be distracting and undesirable. Let's use Nelson's own work to illustrate how distracting that is. Follow the link to his "transquoter page" and mouse over it. Can you imagine anything more annoying as you're trying to read a document? Zotero provides quotes and bibliographical references including hyperlinks to the source documents... a superior system.

Regarding Generalizing the Document, the transliterary standard ignores that layout is often part and parcel of a copyrighted work. There is a reason that formats like PDF strive to preserve the visual impact of a page while still allowing hypermedia. This criticism doesn't apply only to magazines. Imagine how applying the transliterary standards described would effectively rape the works e.e. cummings.

When he presents the concept of "flying documents", and "sworphing" around (swooping + morphing), Nelson takes on the role of the pot in order to call the kettle black. This physical representation (albeit in virtual space), is exactly a case of treating documents as if they are "paper". If you've watched the video you've seen first-hand how wasteful of screenspace it is, and how slow. If you've downloaded XanaduSpace you've experienced how it ignores the mouse and doesn't provide clickable links. "Flying" through a forest of pages is not a good idea. I had a friend who developed a 3-D art gallery for the Web. It was a terrible idea. While it's all very kitzy to say "ooh, here's my art gallery, it's just like being there," it's not just like being there, and it wouldn't be desirable if it were. 3-D representations of 2-D objects distort them, just as they distort the text of the documents in XanaduSpace. And flying or walking around is slow and tedious compared to teleporting from place to place, as hyperlinks of the Web allow you to do. Computers shouldn't model reality, they should improve it. Nelson knows this. It's sad that he didn't embrace his own knowledge in developing XanaduSpace.

Summing up

I'll finish this up by commenting on a quote of Nelson's:

Tekkies think that electronic documents and the World Wide Web are something completely new and that they own it, exactly the way every generation of teenagers thinks they've invented sex and it's their secret.

But it's not new and they don't own it. Word processing and the World Wide Web are not intrinsically new. They are literature.

Nelson couldn't be more wrong. While it's true that literature has been around as long as the alphabet, it's also true that each generation... each individual... categorically does own its own literature, lock, stock, and barrel. Given that, like every generation of teenagers, Ted Nelson doesn't own anything but that which he creates, the admonition above is patently silly, especially from a man who takes great pains to develop and promote a system that would enforce "intellectual property" rights even when they conflict with Constitutional free speech. Would William Shakespeare, or Cicero, or even Hammurabi berate him in similar fashion? And would he have the hubris to take offense?

What we have in Xanadu is too little, too late, too weak, too flawed, too misguided, and too duplicative, too primitive. All XanaduSpace provides is an arguably pretty view of flat documents. It's 40 years of sizzle and no steak at all. Ignore it.


Post a Comment

<< Home