Is there a fundamental difference between advice and knowledge management?

Expert systems, on one hand, were positioned as mechanisms to replace experts, or they were positioned as mechanisms to add value to networks. It really depended on the sensibility of the enterprise.

I really think traditional knowledge management suffers from -- let's create a phrase -- "database syndrome." That is, once we've got the knowledge codified, it could be considered advice. And, sadly, we know that's not true, because that's not how people get advice.

For example, I use Google a lot, but four out five times when I use Google, I'm not going to the first two things the search gives up. I'm not really using Google as a search engine but an iterative search engine. Sometimes using Google I'm going to go off on what I would have thought was a tangent but wasn't really a tangent at all -- it's a better notion of what it is I wanted.

I really think we need to start treating advice as much as an interactive conversation than some module or piece of knowledge that gets plugged in to solve my particular problem of the moment.

How close are we to the algorithms that would do that?

The algorithms already exist. Go to Amazon, go to iTunes. What we haven't had is the enterprise commitment. Look at Gmail. They're already using a recommendation engine, because you see these ads on the right-hand side of a message. If Gmail were smart, they should have, "Based on your previous traffic, you should be sending an e-mail to so-and-so." What's the cost of doing it? If you don't want it, you'd ignore it? If I'm an advertiser, I'd love to be adjacent to that. That's a recommendation engine.