The digital context revolution

by Bruce Taylor

February 1, 2005 —

Dr. David Weinberger is co-author of The Cluetrain Manifesto and author of Small Pieces Loosely Joined. His most recent research has been on the changing nature of how information is organized for search and retrieval. It's the field of taxonomy and classification indexing and information tagging, and it's undergoing nothing short of a revolution.

Bruce Taylor spoke with Dr. Weinberger about the social, historical, and intellectual impact of how digital content is tagged. Following is an edited transcript of that conversation.

Bruce Taylor: Going back over your career over the last decade or so, you've had an enormous influence in the whole area of intelligent search and retrieval, indexing, taxonomies, tagging, meta tagging. It's a very, very arcane and incredibly important part of the world we're in today digitally. Where are you going with that now?

David Weinberger: For the past year I've been writing not a book, but a book proposal exactly on this topic. The book proposal is now 25,000 words long. So I may go straight from proposal to book. You never know. But the topic that really interests me these days is the way that we organize. We have well-developed ways of organizing and categorizing and thinking about stuff in the real world, very basic principles that have been with us for thousands of years. Now we have this digital way of organizing digital stuff, and the principles of organization are quite different.

They violate all sorts of the rules in the real world. In the real world you can only put a book on one spot on one shelf, whereas if you're doing this digitally, you'd put it into as many different categories as you could. That's a pretty basic rule that gets violated. And furthermore, the rules by which we organize, the principles by which we organize the real world and have done so for 2500 years are really basic and important. They determine things like what constitutes a discipline, what constitutes a topic; in some senses what constitutes meaning. And when you change that stuff digitally, it should have some effect. So that's what I've really been interested in.

In the past few months and with the pace picking up, there's been a huge amount of development in the realm of tagging online and digitally. The differences between the sort of real-world way of organizing stuff, which in its finest incarnation, you have somebody who builds a universal taxonomy, a set of classifications into which everything can fit, you know, like the Dewey Decimal System. Every book can find a place in this preset set of categories arranged into a hierarchical tree.

The absolute opposite of that is what's happening on the Web now as more and more sites allow people to tag content, whether it's bookmarks at del.icio.us or it's photos at flickr. So you post your photo and you put in a word or two that constitutes a tag. And those tags are then made public, so anybody can find all the photos that have been tagged as Grand Canyon or as humorous or whatever. So that's a bottom-up taxonomy, a bottom-up set of classifications. There's not a preset set.

It's useful, and it says something profound about the change in knowledge that we're seeing from being a top-down set of categories that preexist to us making stuff up, clustering according to interest. Not by some God-given one single plan, but we cluster fluidly and flexibly by interest. And the owner of the information no longer is the owner of the organization of that information. We're really taking that upon ourselves now individually, and most importantly, socially. That is a huge difference of what it constitutes to know something, to categorize it.

Taylor:It would seem to me that that would have very large implications. What are some of the high points?

Weinberger: Well, for one thing, I think it will have important implications on what constitutes expertise, what it means to be an expert. I mean, that stuff has already been changing rapidly, thanks to the Web. It will make institutions that insist on owning the organization of their information feel hidebound and traditional and difficult to work with.

And maybe most important, or at least very, very important, I think, is the notion of what happens to disciplines, academic disciplines. What happens to the notion that there is a set of topics in the world, and we can organize them ahead of time into a tree of importance and categorization. Well, any tag that anybody has made up now is a topic, and you make up tags for a wide variety of purposes. They're dependent upon what you're trying to do. And so the notion that the world consists of topics that we can establish and maintain, that idea is being eroded by tags. That's a very different way of thinking about your world, I think.

Taylor: Certainly the academics are going to ponder this; certainly the academic and research librarians are deeply concerned with this. What is it going to matter to us on our desktops?

Weinberger: Well, we are going to be able to find and track all sorts of information at a pretty granular level. You know, the Internet has already been this gigantic recommendation engine, but by being able to organize and cluster by tag, which are expressions of human interest, that recommendation engine is getting more organized and self-organized, which is the important part.

Taylor: And intelligent?

Weinberger: And ultimately, yes, exactly, intelligent, because how you tag things is such a rich indicator of how you think and what matters to you. The other thing that I think is almost an immediate fallout of this tagging revolution -- because that's what it is -- is that social groups will form around these tags as well. And there are huge questions about how this is going to work out, even whether it is a good thing or a bad thing. It doesn't matter. It's going to happen.

So groups that tag the same stuff in the same way are deeply aligned, and not just interested in the same stuff, which is the most important way in which we socialize and form groups. They're interested in the same stuff and thinking about it in the same way. And so social groups now will, I think, inevitably form rapidly around the way they tag stuff.

Taylor: Is it how they are going to tag stuff or is it how it's going to be automatically tagged?

Weinberger: Yes, but I think that the way that the stuff is going to be automatically tagged -- and I'm guessing here, but it's from talking with some people, so it's not entirely a guess -- but it seems like the most fruitful way of automatically tagging stuff is not going to be through machines analyzing content and trying to figure out what things are about, and perhaps not even around looking for particular word usage patterns. With tags, you have this unique situation in which the machine knows that this particular photograph or this particular URL has been tagged by some large number of people in a variety of ways, and so it can cluster based around the similarity of tag sets. And in doing that it is, in fact, surfacing a potential social group that is, as I say, not only interested in the same things, the same pages, but are tagging them, that is thinking about them, in the same way.

Taylor: Can you give me an example today where that shows up that we wouldn't otherwise be aware of?

Weinberger: At flickr, which is a great site for posting and sharing photos, how you tag your photo says a lot about you. So if it's a photo within current events, obviously your values may well show up. Likewise, if it's photos of cats, you may be tagging them because you love cats or because you like the way they taste. Those feeds can then be read by or reviewed by anybody and are potential ground for social groups to emerge, which can happen rigorously or totally accidentally if somebody asks a question. It can begin very simply that way. I think it's almost inevitable that social groups will emerge around these shared clusters of meaning because that's what tags are.

Taylor: If you had one source that you would send us to today to smarten us up, where would you send us?

Weinberger: One of the ways to do this is to go to del.icio.us, which is a site where you can post your bookmarks and share them with others, and subscribe. Poke around there at the tags that are interesting to you. For example, I subscribe to the tag taxonomy, because that's what I've been researching, and so every day I get a stream of "here are other pages, here are pages that other people have tagged as being relevant to taxonomy." So that's a pretty loose-limbed way of exploring a topic, but it's a great way because you're being steered by the wisdom of others.

Taylor: David Weinberger, as always you take us on a fascinating journey. Thank you.

ITworld.com Voices