I can't believe that this paper got the reception it did and no one said, "Congrats, you just re-discovered the reproducibility crisis."

I would suggest that if you subtracted out the amount of R&D spending wasted on irreproducible ideas, much of this analysis would collapse. If they're going to assume that economic progress is driven by new technologies, they have to consider how much is spent following research dead ends. It is possible that the number of irreproducible yet seductive research findings has led to an increase in R&D spending tar pits, but that's entirely decoupled from new idea discovery. Number of new ideas is as distinct from number of researchers employed in R&D as it is from number of commercially-useful ideas.

Could it be that these economists see "more expensive to find" and "harder to find" as the same thing? Given that their model seems to suggest that 100 average researchers could do what 1 brilliant one could, it seems reasonable they would imagine a linear relationship between difficulty and expense, but as you argued above re: TFP, there are hard things that are done without showing up in measures of expenses and also much growing expense incurred to manage solved problems (I'm invoking Moloch here, not inflation).

Richard Klavans and Kevin Boyack have made their careers out of studying new field creation. https://www.scitech-strategies.com/ They found that new fields are created when citation patterns between two adjacent fields become more dense. Seems like any discussion of new field creation should cite this prior work.

They don't seem to have a framework for where they looked. Couldn't the examples just have been cherry-picked?

It's reasonable to imagine some correlation between R&D spending and number of employees in R&D, but not reasonable at all to assume talent is directly proportional to compensation. It could be proportional to the political savvy of the department head or the sociopathy (in the Gervais Principle sense) of the employee, and there's a huge amount of luck involved, not to mention any gender wage discrimination.

You might want to consider starting with this section and making the above sections criticizing the changing definition of idea as shorter paragraphs within this section. I feel like the question you end this section with will occur in the reader's mind early on, so you want to get to your answer to it quickly.

It's true that they don't define "idea" rigorously, and are somewhat inconsistent, but doesn't seem to warrant the amount of attention it gets here, given that they say it's not intended to be rigorous, and this seems to distract from your other good points.