John's a good speaker and has done a few "crazy" things.
However I thought this talk was going to be about the wingnut ideas that really will never work because they defy human nature / currently known laws of physics / principles of good software design / basic logic / etc.