Guys, I think you're spot on with critiques of the product and the presentation. But the interesting thing about what reasonwell is trying to do here is its scope. He's trying to map human thought. I for one want to try to help the guy instead of saying why it won't work.

Models of natural argument are a really fascinating area of AI. Check out the Argumentation lab at University of Dundee for some of this work: http://www.arg.dundee.ac.uk/?page_id=96

The tradeoff in knowledge representation is specificity vs usability. How do you structure a machine-readable argument that represents what you think in an acceptable way but that isn't onerous to create? What is the schema of an argument? Are they all the same?

There's some basic research to be done here that's empirical, not theoretical. I hope that Reasonwell narrows down to facilitating a credible use case rather than presenting yet another grand platform. Here's my line of thought for the guy: Who s currently making structured arguments on the web? The Dundee guys have found plenty of natural language argument (it's everywhere, just like in this post) but haven't been able to get many to actually mark up their thoughts. How could you get people to do this? Perhaps an Intrade like platform for predictions, where people could get extra points for being right for the right reasons. It's a big problem and you'll have to tack into the wind I think.