It seems like ML is an interesting counterexample to Colin's point. Industry is now one of the best places to do ML research, but, the history is pretty fascinating. It seems like ML was a product of academia, and quite a fine product too.

I think we as entrepreneur types want to dismiss academia. But there's something to be said for the institution. It's one of the only ways to "infect the world with a thought virus."

Besides, which research problems has Tarsnap solved? At least, solved in a way that can be reused by the world?