27 April 2008

John Wilbanks on the Knowledge Web

Here's a nice meditation from Science Commons' John Wilbanks on openness, access and innovation, which includes the following thoughts on the "knowledge web":

Just to be clear, here’s what I mean by a knowledge web: it’s when today’s web has enough power to work as well for science as it currently works for culture. That means databases are integrated as easily as web documents, and it means that powerful search engines let scientists ask complex research questions and have some comfort that they’re seeing all the relevant public information in the answers. A knowledge web is when journal articles have hyperlinks inside them, not just citations, letting systems like Google do their job properly.

A knowledge web is predicated on access, and not control, of knowledge. There will never be a competition to provide the best single-point query to the full-text of journals without access- unless the journals all merge down into one company. That’s the only way a controlled system covers the whole world, through monopoly. There will never be a knowledge web where the entire backfile is hyperlinked to databases for relevance based indexing without access. Scientists won’t get to use the newest and best technologies until those companies that control knowledge decide to adopt those technologies. Control is the enemy of testing the newest technologies, of building one’s own system to suit one’s own needs. We have to have access to build a knowledge web, at least if we hope to replicate the success of the regular Web and the Internet.


Anonymous said...

I can agree with this general principle of Wilbanks: "Just to be clear, here's what I mean by a knowledge web: it's when today's web has enough power to work as well for science as it currently works for culture." This is what I call making the Web work for science.

But most of the article is about a technical concept of the "knowledge web." It argues that open access is desirable because of the Semantic Web and related ideas about building vast computer based knowledge representation systems. Such systems are not part of the way the Web works for culture, nor are they likely to work for science. Findability is what we need, not computer generated structure.

Linking everything to everything is an old idea, older than the Web. Ted Nelson was a leading proponent in the 1980's. But the underlying concept of knowledge is incorrect. The same is true for the Semantic Web. This is basically the dream of computer people, the dream of machine understanding. It lacks a scientific foundation in the science of human knowledge.

Simply put, we do not understand enough about human knowledge and understanding to build the machine understanding systems these people are talking about. This is the perennial problem of artificial intelligence, the fact that we do not understand human intelligence.

Findability and full text search are what make open access valuable, not the potential for a vast computer based knowledge web.

David Wojick

Glyn Moody said...

I don't think they're mutually exclusive. We already have search that works moderately well, and will doubtless work better. We don't have the knowledge web, not least because things aren't open, as you say. I think much of the semantic web stuff will just happen as things open up. Then we can find out if it's *really* useful....