This reminds me of a person who somehow managed an invitation to give a presentation at UC Berkeley on database compression in 1993, with similar levels of snake-oil. I remember one of the funnier moments, a theme of the presentation was how there were more and more large data sets out there, including the Library of Congress. In a burst of enthusiasm, the presenter gushed about how they even had a Stradivarius violin in their collection. I turned to my friend and said, "well, I can imagine how to do data compression on that, but the decompression is going to be hard."

The actual technology was just a variant of Lempel-Ziv, and the claims about lossless compression were just hiding the entropy by not counting a huge dictionary off to the side. Even so, there were claims that this result "extended information theory" and the patterns found could be the basis of learning.

The presenter had climbed in their regional market, and gotten a puff piece in a business magazine, but obviously the grad students at UC Berkeley weren't buying it. The presenter's startup didn't go anywhere, but they have had a successful career, and are now a VP at Gartner.

I guess the lesson is that there's a network that runs on technical expertise and ability to deliver results, and a parallel network that runs on bullshit and the ability to convince people. There's a lot of money and power in the second network, and it intersects with respected, established institutions. Grant seems to be playing that game masterfully, and the publicity from being publicly attacked at a "Black Hat" conference plays into his hands beautifully. Maybe Gartner will be hiring?