Very well put.
I would also add that beyond tailoring one's scientific aims to the whims of study sections, much of "grant writing" is administrative overhead. I exchanged about 200 e-mails with various administrators both at my home institution and NIH in the process of submitting my F31 (which was funded). There were many more e-mails to start the funding process. There were many e-mails when I wanted to use funds to purchase supplies. All detracted from my time doing science.
Just to submit the grant, I had to get on my PI's IRB, and then document that. I had to get animal usage documentation. I had to fill out home institution forms, NIH generic forms, describe the research facilities and equipment and resources of my PI (who was already R01-funded many times over), explain how I would protect human subjects, how I would include women and minorities, explain how many subjects I planned to enroll (broken down by ethnicity and gender), explain whether I would include children, justify the use of vertebrate animals and enumerate my procedures with them, discuss my resource sharing plan, explain respective contributions, explain how I selected my sponsor and institution, brag about my responsible conduct of research training, list my goals for fellowship training and career, write a personal statement and biosketch, edit mentor's biosketch including prior trainees and grants, break down the activities planned under the award, discuss prior research experience, write a recommendation letter from my PI, secure letters of recommendation, including appendiceal letters from collaborators, and write a cover letter.
The grant was over 50 pages, 5 of which were what I consider science. Yes, writing those 5 pages was insightful, but my study section comments were worthless. They boiled down to "great lab, great trainee, project looks promising" or "this trainee has no experience with this technique; it is unlikely he will be able to achieve these objectives". Then there was an arbitrary score, some politicking between my PI and the NIH, and the grant was funded.
Mine was an EASY process by comparison to many others. I have friends who scored very near the cutoff, followed all study section recommendations on their resubmission, and their score had doubled on the next pink slip. They were obviously not funded, so what good did that do them? 6 weeks of work for a couple off-hand comments and a whole lot of disappointment. And these are low-stakes grants, just fellowship applications for people who are fully funded already. It is hard for me to fathom why I would want to go through this process again, and again, and again - to put bread on the table - when my chances of success despite two decades of post-secondary education are dismal.
p.s. This is coming from somebody who loves writing, much more than doing experiments.