11. DoWhatDo

SLAYTON: Basically I like collaborative projects like DoWhatDo. I came across this idea, a model of conversational and learning systems theory developed by Paul Pangaro and Gordon Pask. One thing the theory describes is how conversation evolves over time. For example, what we're doing right now -- we don't know what we're going to say next...

CL: But we're trying to keep it going...
SLAYTON: DoWhatDo was a way of describing that relationship in a software model. And the reason that it was being explored was to develop computers that could converse, so that they could be given a general range of topics to discuss and over a period of time they would illuminate some interesting ideas.

Say you set up one computer that knows everything about engines, and one that knows everything about speed, and add another that knows everything about leather. Then you get these computers talking to each other in this conversational systems theory model -- they would never end up talking about the Gulf War crisis! They would most likely, at some point, end up talking about motorcycles, even though no individual machine has any knowledge about motorcycles. So given a set of topics and this conversational structure they will come up with ideas that no individual machine has. Which is pretty much the way it works with humans.