Regarding the main point of the article. I do agree that it's very important to not get stuck in local maxima. However, I think the author heavily underestimates/overestimates the chances of the following:
1. The author underestimates the rate of competing ideas coming out of 'learned' scientists.
2. The author overestimates the possibility of an 'internet dilettante' without good knowledge of prior art coming up with something useful.
To me it's a bit like me playing chess with a Grandmaster. Sure, I know how the chess pieces move, I know basic tactics and strategy. But I have never studied a chess theory book, never learned any openings. What's the chance of me coming up with a chess theory novelty that provides a better model of the game and lets me win the Grandmaster? It's not none. But due to my lack of knowledge of prior art the chance is pretty thin.
1. The author underestimates the rate of competing ideas coming out of 'learned' scientists.
2. The author overestimates the possibility of an 'internet dilettante' without good knowledge of prior art coming up with something useful.
To me it's a bit like me playing chess with a Grandmaster. Sure, I know how the chess pieces move, I know basic tactics and strategy. But I have never studied a chess theory book, never learned any openings. What's the chance of me coming up with a chess theory novelty that provides a better model of the game and lets me win the Grandmaster? It's not none. But due to my lack of knowledge of prior art the chance is pretty thin.