What do you do to write better proposals, grants, and papers?
Do you think reading books about how to write scientific content is a good way to improve it?
The way to learn to write is, simply, to write.
But then get feedback on your writing and re-write in light of the feedback.
The Software Patterns community has a process called Writer's Workshops that are quite detailed. When you submit a paper to a patterns conference you are assigned a Shepherd who is an experienced pattern writer, usually with knowledge of your field. The shepherd works with you (the "sheep") to improve the paper over three or four iterations of feedback-rewrite.
After shepherding your paper may be accepted to the conference, though not for presentation in the traditional sense. The conference consists of a set of writer's workshops in which a few (8-10) authors each have their papers discussed by the other participants while they listen and take notes. The author has a very small part in the workshop other than to think about what others suggest about how the paper can be improved.
After the workshop the author can ask questions, but never gets to "defend" the work. The idea is that if others misunderstand you then it is your job, not theirs, to fix it.
The paper is then revised one more time and it is this version that makes it in to the proceedings. The whole idea is to improve the print version, not present a version prepared without help.
The patterns community is pretty close knit because of this working together to improve one-another's work.
This process was brought to the software development community by Richard P Gabriel who is both a geek (Lisp et al.) and a poet. The same process is used by poets, in fact and is quite old. RPG has written a book on the process: Writers' Workshops & the Work of Making Things
If you don't have the patterns community behind you, or if you aren't writing patterns, it is relatively easy to set up a local writers workshop and follow the process. You can do this for any sort of writing as long as you have some people with domain knowledge and some writing experience.
There are many ways that one can improve their scientific writing skills. Because humans learn in diverse ways, I do not know if there is one be-all-end-all solution for how to improve one's writing skills.
Two main methods that I have used (and still use) are (1) attending grant writing workshops, and (2) reading other published papers in my field and emulating their overall style. I am someone who learns by seeing and copying.
Some items to note:
A very specific answer: There's an (originally) online course from Stanford called "Writing in the Sciences" that became very popular. While you might not agree with everything she teaches/suggests, I think it is a very good course. It's now all on YouTube.
Two ideas for you:
Many, if not most, universities offer scientific writing courses for graduate students. If you're a graduate student - take such a course. If you're a post-doc or even a tenure-tracker - don't be ashamed; go attend one (not for credit).
Ask your advisor, if you're a grad student, or a colleague you're close to and whose writing skills you appreciate, to help you by performing a language-editing pass on what you're writing. If it's too much to ask or if you haven't someone to ask - try finding someone to do this for pay; it is not uncommon.
By watching this magnificent lecture by Larry McEnerney from University of Chicago!
It is eye opening! He shows how the value of our research and the reader are essential for a writer. He argues that we already have all of the skills needed for writing, but we don't know how to do it because we've been writing only inside school system and academia. We've been writing pieces for the teachers that were paid to read it and care about us, and that is not how the world, and science, works.
This lecture changed my thinking for sure.
Reading books is for sure one way to go. I have read many of them, and my favorites are "Writing science" by Joshua Schimel and "Writing Science in Plain English" by Anne E. Greene.
However, reading books remains a passive way of learning. In my research group, I have therefore decided to create a writing group. In a writing group, we learn together, comment on what we learned and feedback on each other manuscripts. We meet every two to three weeks for two hours.
This writing group has transformed our perception of writing. Group members will discuss openly writing methods or specific writing problems as if it was a statistical or a programming problem. We did not become writing experts, but we improved, and we keep on improving every day (as opposed to taking a lesson or reading a book).
On the Scisnack website you can find some information on writing group including scientific publications. There is also a lot of learning content, including videos, expert advice or book reviews. Also, I have created a small Prezi presentation about writing groups to motivate the new Ph.D. students to join us.
Scientific writing skills are mostly the same as general writing skills. Read good works of fiction, non-fiction, journalism, etc. These are easier to find, and more enjoyable, than good pieces of scientific writing.
Nature recently published a career column titled "Novelist Cormac McCarthy’s tips on how to write a great science paper".