William Loughborough asked me if I'm negative about folksonomies, he having been apparently enthusiastic about them. The context is that he'd compared them to making an index in a book, and was worrying that people didn't see the accessibility advantages of them. He'd also prepared, in the same document, a wonderful summary of Bucky Fuller's point about sunlight, gravity, and the accumulation of lore. This was my reply to the folksonomies question:

* * *

Very negative. I haven't written this up yet, so I can't yet give it the elegant expression that seems to just roll out of your brain (I looked up “elegance” the other day on Wikipedia whilst I was reading about æsthetics, and the definition on there made a previously very generic and boring word seem to me to be very exact and useful!).

The general idea is that, sure, books come with indexes and we have to manually compile the indexes from the texts. But indexes are a limitation of paper, and we don't have that limitation anymore. Now, if I want to find something in a work, say a Project Gutenberg book online, I don't scroll down to its index. I do a text search in the browser.

How do I search for things on the web? Google has a bar on its homepage, [...], and it has a cursor in it. I type text in and I press enter, and within two seconds it's searched ten billion documents of human knowledge, ranked the results, and given me a presentation of the most pertinent ones.

Does it ever use keywords? Never!

I'm not saying that we should never annotate things, or explain things. Just that to reduce things to keywords is usually not the right thing to do. Sometimes it is. One memorable place where keywords are a brilliant solution is on the technology site Slashdot. There they post news stories that people can comment on, and the comments are always 90% people bickering and moaning, and 10% hilarious comments with occasional bits of insight. There's a peer-ranking system of comments, but it's never good enough.

So what they did was they started up a peer-ranked keywords idea. And some of the jokes that people have managed to squeeze into these keywords far outweigh the time:value ratio of reading the threads. For example, when a news story is some question like "Is Microsoft Better Than Apple?", you'll invariably get, underneath, "Keywords: yes, no". It's hilarious.

[The sentence that I wrote here has been elided as it may be perceived by the common minded as a slur against two geniuses who have done a tremendous amount for humanity; it was a sweeping summary of my complex position of the negative aspects of the capabilities of Ted Nelson and Tim Berners-Lee that, in overcoming, they have created such outstanding technologies.] Nelson is much worse off than TimBL so he's much more insightful, but somehow TimBL hit on that right level of complexity/simplicity that made the Web, which has changed all our lives. But like Newton, he was in the right place at the right time: someone else might have made something better than TimBL two years later, or perhaps something worse ten years later. It doesn't matter.

Once I was talking to someone and I had this remarkable vision of model cadences. You know how the map is not the territory, but of course we need maps too. It's not that maps aren't valuable, it's that we mustn't conflate the two! Map != Territory.

And I realised, that's what happens when people make maps. People make a really simple map, and someone comes along and says, that abstraction is wrong. And then someone adds more complexity to it so that it models the territory better, but someone is always excluded. So eventually, you get Lewis Carroll's thing...

“We actually made a map of the country on the scale of a mile to the mile! It has never been spread out, yet, the farmers objected; they said it would cover the whole country and shut out the sunlight! So we now use the country itself, as its own map, and I assure you it does nearly as well.”

Perfect. So we go back to having that really craptacular model again, and then it goes back and forth like a hendiadys. Model cadence. Start simple, get complex, get simple, get complex... Newton's model was simple, Einstein's model was complex. Simple, complex, simple, complex...

Jacob Bronowski... I've been reading a book he wrote about magic and science. He plots the period in which magic became science, and how it became science, with some fascinating examples. But one of the things that he talks about later on is that science isn't a territory, it's a map! And that surprised me a little bit until I got it. He's right: we're not discovering truth, exactly. The truth is always “true”; the universe is always in whatever state it's in. What we're doing is producing models of the world so that we can understand it better, put it on our pile of accumulated lore, and leverage it to save babies.

That's what we do now, but before the notion of science, it was hard to think of that. Though we did have a pile of lore, it wasn't accelerating as fast, because one of the primary things we didn't understand is how to get around that idea of superstition. We thought “oh, we have to bend nature against its will to have it work for us.” I'm just repeating Bronowski; Bronowski is interesting. Kuhn is interesting too, he talks about similar things, and there are other guys.

Anyway, that's why I don't like keywords mainly!