Everything is Miscellaneous
DW: I think it has a big effect. We have thought that to "know" something in the very traditional sense, not simply in the business sense, is to know what sort of thing it is. That’s an idea that goes back thousands of years in our Western culture. If it turns out, in fact, that everything is miscellaneous —that is, that things can be in multiple categories and what you take something as depends on what your interests are at the moment—then knowledge isn’t a simple, stable thing that you know it, and then you’re done thinking about it. Knowledge management at its beginning shared the traditional view of knowledge. It assumed that knowledge is the next step up from information. That’s a very static view, as if there’s "a knowledge" sitting there that you have not yet harvested. The miscellaneous view says that there’s an enormous number of ways that things can be connected and the connections help us to predict, to manage, to understand. But those connections are there as potential, and they are basically never-ending. People are always finding new ways to link things up and new things to say about them. It’s never going to stop. We’re just going to keep adding to the set of potential relationships, and that set of connections and relationships is tremendously valuable because from it comes not a single knowledge ("now I know the way things are"), but rather it provides the ability to respond quickly to situations as they change and to see uses of things that otherwise might escape our view. We get far more use from our information by treating it as miscellaneous than we do by simply trying to find the old-style, stable, single-faceted knowledge in it.
But that means that in order to get maximum value from this ever-growing collection of ideas and information, in a real sense you have to un-manage it. It’s by un-managing it, by allowing everybody who touches it to add to the connections, to add their own links, ideas, reviews, bad ideas, good ideas, insights—that’s how the collection grows. Just as with the Web itself, it cannot scale sufficiently if it’s too centrally managed. So the K has changed, and the M is changing.
HM: What role does the semantic Web play in the miscellaneous, as you see it?
DW: The semantic Web wants to represent knowledge by creating relationships among all the pieces. Within specific domains, there’s tremendous value to that. But it is a reductive process. You’re taking some field of rich human interaction, and you’re finding in it a set of predictable stable relationships. That’s very useful, but it is reductive. When people say they’re going to model all of the legal domain, then I start to think that’s overweening ambition. The legal domain is just too big and too complex and ambiguous. So, insofar as people think that they’re going to master some gigantic domain by simplifying it into a set of relationship triples—you know, a is related to b—it seems to me to be unlikely to be of great value. But when people take smaller confined domains, it’s far more plausible. For example, if they’re talking about documents, every document has an author, every document was written by that author, every document was written on some date—all that sort of information is predictable, and it’s useful to capture it in a way that can be reused and hooked up with other pieces of information. Insofar as the semantic Web is enabling that type of modeling—smaller scale and then hooked together—I think that has a lot of promise.