Singing the Bite Me Song

« May 2002 | Main | July 2002 »

June 28, 2002

XML Discussion with David Weinberger

The Semantic Argument Web:
What really scares me.
By David Weinberger

June 14, 2002 — Tim Berners-Lee (TBL) is frustrated by his creation. It's turned out to be a good way for humans to post and read stuff. But we're overwhelmed by the humanness of it, its randomness, its capriciousness. We need help from our machines to make sense of what's out there, to get answers to our questions. And the way to do that -- or so says TBL -- is to put more smarts into the data.

It's interesting that TBL rules out the other obvious approach: make the machines smarter. But in his major work on the topic of the Semantic Web, he writes:

Leaving aside the artificial intelligence problem of training machines to behave like people, the Semantic Web approach instead develops languages for expressing information in a machine processable form.

Note the "instead" in that sentence. This doesn't mean that TBL thinks AI is pointless or evil. But it does clearly mean that the Semantic Web doesn't take the AI route. [...]

I'm being lazy and will edit this later when I have time. These are some comments we traded on David's comment board, and I wanted to save the ideas so I could think more about them later.

Here it is:

06-26-2002 08:32 PM ET (US)
Miasma 50
Edited by author 06-26-2002 08:33 PM

I'm posting here, but don't see any reference to the Semantic Web story.

All I wanted to add (and I'm not as informed on this topic as I'd like, but getting closer) is that SGML and these central control content management impulses are evil, but there is no need to throw the baby out with the bathwater.

XML and all the stuff coming out with it is the open-ended side. Anybody can make a DTD. That should mean something, right?
Big corporations have their own in-house DTDs. Prosyletizers develop a DTD and preach it sistah. Stuff is thrown into the mix, RSS, SMIL. A DTD for mathematicians. For chemists. The specialization needs of these are obvious.

Meta data can make content smart. I believe in this. But I'd rather see a proliferation of DTDs than a consolidation of them.
Let's get all social science on this stuff. A DTD is essentially the keys to the Kingdom of Heaven. What is blessed is blessed, and what is damned is damned. That is, damned with invisiblity.

So yeah, I want to see a feminist DTD. A socialist DTD. I want to see DTDs that make visible things that are marginalized or ignored. I want those things to live in meta data. You may say I'm a dreamer, but I'm not the only one...


David Weinberger 51
06-26-2002 08:41 PM ET (US)
Miasma (/m50) -

We only have the one QuickTopic for all the columns, which is why you don't see one specific to the Semantic Web article.

Yes, you're dreaming the dream. And I like the way you put it. But you're also - IMO - drinking the KoolAid. Also inhaling. I agree with you that it'd be great if consistent and coherent metadata attributes were used. The benefits would be spectacular. But I still don't believe that it's achievable. It would require us humans to do something that we apparently don't want to do: agree on schema and then *use* them. And fill in metadata. My experience with SGML leads me to think that you can't get to the dream that way. And I don't know of any other way. I'd love to be wrong.

The idea of a feminist DTD is certainly interesting but I don't understand it. DTDs apply to types of documents - repair manuals, thank you notes, glossaries, coloring books. Since a feminist document can take any form (can't it?), how would you do a feminist DTD? Or is the idea that DTDs themselves are somehow sexist so we need a feminist syntax? I'd love to hear about it...

Miasma 52
06-27-2002 07:31 PM ET (US)
Edited by author 06-27-2002 07:34 PM

Hi David, thank you for your response. You are welcome to inhale with me (miasma being what it is--).

Regarding the idea of a feminist DTD, I been working on this thought for several years now, still stewing on it a bit.

The idea isn't that DTDs are sexist, exactly, but rather, that language is patriarchial, reflecting the dominant patriarchial culture. Language constructs us as much as we construct it. So reflecting on this in terms of building interesting DTDs is sort of fascinating.

Think about what is acceptable to talk about in our culture. Now think about what is beyond the pale. Now, stretch further, and try to think of things we don't have names for.

Case in point. In the land of command line interfaces, the concept of GUI did not exist. It was beyond the pale. It was invisible. Did a mouse tool create a GUI? I don't think so. Language, even visual language, created the GUI just as surely as hieroglyphics morphed into syllabic alphabets.

Sherry Turkle talks about the culture shift that comes about when moving from a command line interface to a GUI--when she talks about precise cultures of calculation vs cultures of simulation (_Life on the Screen_).

Could a culture of guess and by gosh point and click, look and explore have existed in the compile and calculate world of command line?

(If this analogy fails, then think of Robert Pirsig's _Zen and the Art of Motorcycle Maintenance_, where he talks about the two different motorcycle cultures, those seduced by the romance of the machine and its power on the open road, and those who feel the need to lovingly understand & personally maintain the inner workings of engine)

Are such worlds mutually exclusive? No. Neither would a feminist DTD be. Several different DTDs can be referenced in the same documents, for instance.

Do all documents live in boxes taxonomically described, like a biology textbook gone mad? The taxonomy and classification approach to biology is giving way to a more systems-oriented approach--less influenced by a mono-causal scientific method related more to physics. Turns out mono-causality doesn't describe physics too well either.

Suppose you had a Newtonian Physics DTD. Would relativity be described by it? Would quarks and string theory? See where I am going? What is blessed is blessed and what is damned is damned, as in damned near invisible, ostracization through language. To watch it snow in a culture that has no words for snow...

To talk about mathematics with a DTD that can't handle equation symbols...

And a feminist DTD, it would do the same thing. It would identify things not described in the current languages, and make meta data tags for them.

What do you think?


David Weinberger 53
06-27-2002 06:45 PM ET (US)

I understand the shift away from a patriarchal paradigm well enough, but I still don't see what a feminist DTD would be like. The DTD a publisher uses for a book about feminism and for a book defending traditional sex roles would probably be the same, with elements such as "title," "illustration," etc. Is there something patriarchal about such tags? Or are you envisioning capturing metadata attributes that are somehow non-patriarchal? Do you have an example in mind?

Miasma 54
Ah, I see where our wires cross.

You still pretty much envision meta data, or smart data (available for future feeds or manipulation or crawlers or AI agents) as a formatting tool.

I'm not talking about formatting at all. Semantics. That was what I thought you were getting at with the Semantic Web. Syntactical parsing. Computational linguistics. AI. CONTENT meta data, or really smart data. It doesn't tell you shit about how to format something--rather it works as an advanced ID tag for future database crawls.

See, my dissertation advisor had a nascent idea like this in the early days of the web, before any of us at Rensselaer were thinking about XML. He wanted to make an experimental page where every single word was a hyperlink. (as you can see, we come from the land of hypertext theory).

I thought the idea was nuts at the time. If every word and item were a link, what would be the point of taking any of them? Of course it didn't make sense to me. Links were dumb then. They had to be created manually and they could not convey the sense of the content or potentially open up MULTIPLE destinations from a single link. And worse, they were not dialogic--yet. The technology just wasn't there. (dialogic, meaning, like the interstitials that Radio Userland blog software sets up in the templates, as well as in the commenting opportunities attached to each item, like on Slashdot and to some extent here)

Now if we start looking at what is happening with blogs, XML, feeds, and other nifty built-in things, we can start to see the potential for smart XML tags designed to connect to databases. And better yet (and this is my main hobby horse) self-organizing sites and self-organizing hypertextual systems (where links generate and organize themselves based on smart tags, agents, or other connectors based on frequency, popularity, or some other criteria).

Why do I get so excited about this? It is the true potential and revolution of the Web. The difference between crocheting an afghan and using a loom to weave something with a dense thread count.

AND that is why the language issues, what is visible, what is invisible, are a CRUCIAL part of the discussion, at this building stage of the process.

(also makes it really clear how off-track the whole dot-com widget-selling biz was in understanding the true potential of the Web. It was like using a restaurant-level food processor to cut one carrot at a time.)

What would a feminist tag set look like? It would have markers for things that are normally invisible. Like if a post or item were written by a homeless person, there could be a marker for that. Later crawls could get a bead on the degree of homeless participation in web culture. Little table tents at public access terminals could tell people how to stick that marker in. There could be a content marker for violence against women items. A marker for content items on gay-bashing.

What other things are invisible? How about a content marker whenever someone puts up content, anonymously or otherwise, about being abused by a priest as a child? No, I'm not talking about true confessions. I'm talking about the politics of deep structure interfaces, and what things they choose to identify. See? An XML tag for every time someone discusses how they have been exposed to a person yelling a racial epithet at them. The possibilities go on and on. They are as content-based as identifying a particular kind of chemistry formula as compared to another kind of chemistry formula.


June 28, 2002 at 12:51 AM in Best Essays, Interactivity | Permalink | Comments (0) | TrackBack

June 13, 2002

Feds Stockpile Anti - Radiation Pills

Now THIS is the sort of thing I've been expecting. I'm still figuring out Radio Userland, so sorry no indent. I'm loving the XML stuff, but learning to make best use of the code, macros and stuff.


Filed at 2:52 p.m. ET

WASHINGTON (AP) -- Federal agencies in Washington ordered 350,000 potassium iodide pills this week from a North Carolina company to protect people from cancer caused by radioactive iodine, which can be released in nuclear explosions.

They were passing these pills out at places near reactors the week before Ashcroft's Moscow announcement, but I got some PR thoughts on that, so more to follow

June 13, 2002 at 12:17 AM in News to Note | Permalink | Comments (0)