as the author points out, the incentives are always in favor of doing a bad job with regard to efficacy and trial design. a company that can produce a really great and well-proven drug is not going to make more money than a company that produces a shitty drug that is also approved for the same condition.
in fact, with the right marketing efforts, the execs would even see similar sales figures for both of these hypothetical drugs. hence why more is spent on marketing than R&D.
Testing for rare long-term effects requires a huge sample size, and billion dollar clinical trial budgets.
Teams do decide, early on, to try to discover drugs for serious diseases or common diseases, depending on how much money they think they can raise. Usually only proven teams can raise the huge amounts, so most pharma startups target serious niche diseases hoping for approval under the orphan drug program.
Yep, there is only a weak mapping between FDA approval and actual drug usefulness.