I don't necessarily disagree. Instead, I would modify what I said originally. There are structural factors that make taking large risks less profitable to the researcher than taking small incremental ones. The small incremental steps are important, but the large risks are also important. I think that the portfolio of research that results from "professional academia" tends to underweight risky projects. I am not claiming that it shift towards a total random search; I only claim that we need more risky investment compared to what we have now.