[throwawaymaths]

well but the portfolio managers in this case could really do a much better job at picking hits. we wouldn't have 20 years of failing Alzheimer's drugs if pharma identified that the protein hypothesis was junk (which people were talking about in the aughts)

[throwanem]

That's a really interesting point, yeah. Not that I'm really equipped to evaluate the technical arguments, but it's striking on reflection how absent from the discussion the question seems to be of how efficiently on average the funding to trial a given drug and indication is allocated. The governing assumption appears to be that the status quo is acceptably close to optimality.