The answer is clear if you look at his most recent failure. He came up with a device for doing thermoelectric conversion(waste heat to electricity) at a much better efficiency thanwas available. Even though it was verified by outsiders, it turned out to beimpossible to scale up to industrially useful size... theefficiency sank to useless levels when he attempted to make larger versions. So he learned the hard way that you can have a demotechnology that can mysteriously stop working when you attempt to scale it to industrial levels.
