Not Evil, Just Smelly

October 6, 2004

Edd Dumbill

One wonderful and frustrating aspect of the XML community is its members' willingness to reassess themselves and their endeavors, often entertaining deep divisions in views, and yet not self destruct into a puff of rhetorical logic. This was evidenced in a rather illuminating discussion last week, caused by the re-examination of a critique of SGML and HTML from 1997.

Stuck on Nelson

Though he has never, to my knowledge, roamed personally among the pastures green and angular of the XML community, Ted Nelson's ghost looms large over us all. Sergio Rodriguez drew the attention of XML-DEV to Nelson's self-censored "XML is Evil" essay, which Nelson regrets "is too inflammatory to publish in its current form." The page instead points to a "very mild" version of his argument in an article published in the World Wide Web Journal (may it rest in peace), now residing on Embedded Markup Considered Harmful.

Written before entitling pieces "Considered Harmful" was in itself considered harmful, Nelson's article argues that inline markup, such as SGML and XML, is problematic. His alternative model comprises three layers: content, structure, and presentation. This is not coincidentally the model used by Xanadu, the hypertext system designed by Nelson and others. Xanadu's hypertext model is a closed world, where links never break, supporting copyright and version-management features. It sounds ideal. It is also widely unimplemented.

When people talk about the genius of Tim Berners-Lee's strategy with the Web -- turning away from the notion that links shouldn't break -- it is Nelson and the hypertext community that play the villains in the piece. Certainly, any brief study of Nelson's home page and the Xanadu page shows an undue amount of alarming UPPER CASE BOLD ARGUMENT, and characterization of his work as a "fight." In Nelson's mind at least, it seems he is locked in combat against the Web and all it represents.

Leaving the somewhat foaming nature of Nelson's web presence behind, it is not at all clear that his W3J article is completely wrong (or right). Hence Rodriguez' question to XML-DEV. The question was particularly timely as we are now all much older and wiser than in 1998, well aware of the failings of both ourselves and XML over the last six years.

XML's official defender Liam Quin said of course it wasn't all true. Eric van der Vlist in return offered the counterpoint that it wasn't all false either.

I think that the issues he is raising are real ones but that either they're not as important as he thought or we've learn[ed] to get used to them and work around them.

Rereading the Nelson article, I find myself in agreement with van der Vlist. The most resonant objection Nelson makes is that embedded markup tends to impose structures that don't fit upon data: "What is not expressible sequentially and hierarchically is deemed to be nonexistent, inconceivable, evil, or mistaken." There's certainly enough empirical evidence of this happening with XML.

However, as van der Vlist observes, we have to some extent worked around this, at least for textual content. He references Patrick Duruseau's Just in Time Trees. I was also reminded of LMNL (Layered Markup Annotation Language), proposed by Jeni Tennison, Gavin Thomas Nicol, and Wendell Piez. As far as other types of content are concerned, in some cases RDF is a strong contender for removing the constraints of the XML structure.

One of the LMNL team, Nicol, offered his perspective on Tim Bray's viewpoint that Nelson was just about wrong on everything:

It might be fairer to say that many of the things Ted Nelson thinks are important tend to fall outside the scope of things XML is very good for. Certainly the question of the technically correct way to do persistent linking, annotations, etc. in XML documents (or the WWW in general, for that matter) is still open to debate, despite HyTime, TEI, Xlink, etc. ...

I personally think the node-centric processing model that almost everyone thinks is the correct processing model for XML is fairly limiting, and doesn't jive with a lot of the processing I have done on XML over the years, just as the notion of validity, and to a lesser degree, well-formedness don't reflect processing reality in all cases either. I guess I don't see the conflict between Ted's model and everyone else's so clearly. ...

Still, I wouldn't have spent as much time as I did in standards groups, and I'm sure, neither would have most people, if we didn't believe there was a (technical) benefit in doing so.

Len Bullard produced a considered response to the truth of Nelson's argument, which summed up well the place of Nelson's essay given the perspective of our current position.

In part, it's true. Nelson set out to design hypermedia that would not fail (no broken links, no copyright problems, tumbler addressing, rich formats, and so on). In part, it's false. XML isn't evil; it smells funny. And XML isn't the whole story anyway. ...

Nelson failed to produce Xanadu. SGML failed to get ubiquitous buy-in. The Web produced a scalable and successful implementation. XML got the buy-in from the programming community that enables web services, layers for the Semantic Web, etc. ...

Either way, the Web won and Nelson lost. Keep in mind, there were designs for spaceships before there was a demonstrable airplane that could lift its own weight and that of a passenger under its own power. Vision is not a working solution.

And what to make of Nelson himself? Bray sums up well: "Ted's place in history is secure because he asked more important questions than just about anybody. I think he usually offered the wrong answers, but questions are more important."

Births, Deaths, and Marriages

Also in XML-Deviant

The More Things Change

Agile XML


Apple Watch

Life After Ajax?

The latest announcements from the XML-DEV mailing list.

Planet XMLhack

Every two or three years I get to announce something of my own to the XML community. Planet XMLhack is an aggregation of weblogs from members of the XML community. It was partly inspired by my work writing this column, as much dialog has transferred from the mailing lists onto weblogs. The weblogs are chosen to have a reasonable technical content, but because this is as much about the community as it is about the tech, expect the usual personal ramblings and digressions as well.

Publicly Subscribed XSL Hands-on Training in Europe – Oct. 18-20, 2004

Ken Holman announces another of his well-respected training events.

Stylus Studio 6 Adds New Support for XSLT 2.0, XPath 2.0, XQuery, XML Schema, Legacy Data Integration, and XML Mapping

Many new features in this commercial XML-development environment.


We blog, therefore we eat ... 78 messages to XML-DEV last week, Ted Nelson rating 400% (hogging Len's limelight) ... patiently pointing out the appropriate forum since 1998 ... when web services go bad, it looks like those who care shut their eyes ... new patent PDF to Mondrian converter ... IGD -- WTF? Ah, HTTP!