Genuine question, what work has Paul Graham done that can be classified as 'great'?
I think all the parts were invented by other people, but he got the right mix to make it work.
It's also like the wheel. Before it was a huge problem and no one know how to solve it. Now it's invisible.
Note that all modern versions use enhanced variants with a lot or trick to make it better.
So yes Bayesian spam filtering was impressive for its time. I remember it being done locally in mail clients.
It’s also one of the early popular deployments of a form of “AI” but that was still during the hangover from the first AI bubble when you had to call it “machine learning.”
Whatever you feel about it, it would be inaccurate to say that he has not found success. Clearly he has. So it is up to each person to decide whether or not any products of that success, including his writings, matches their own definition for 'great'.
You are somewhat smart in one specific domain; you are a "Genius".
You are successful in something; you are the "Greatest".
You fail in something; you are a "Loser".
You make one single mistake; you are "Dumb".
It is almost as if we have forgotten how we got here viz. lots of trial-and-error, sheer persistence, extreme specialization, building theories, sharing knowledge, building teams, focusing on goals, good dose of luck, etc.