All joking aside, imagine if pg did try and build a real expert system combined with a chat-bot style interface, that could represent his expertise. He could probably sell access to something like that for a few bucks.
I can just see pgbot turning down Google because its founders were too weakly committed - they wanted to sell the whole thing for <$1M and go back to their grad studies.
It was their 2nd invention---highly relevant advertisement that made them an explosively profitable company.
I'd posit that people like the investors mentioned above have a highly intuitive "pattern matching" system for finding promising startups. The lack of any one quality won't disqualify a startup; rather, they can unconsciously evaluate the whole startup as a package and decide whether it's likely to be a winner.
Hence, the win of probabilistic AI systems over rule-based ones.
(And that might be a lesson to those guestimating future valuations.)