When writing a research paper, article or book. You don’t wanna choose a topic and start writing your opinion or understanding as you go through the process. No, what you wanna do is arrange pre-conceived ideas so that each idea flows through the other as one continuous stream of thought. When these ideas are atomic, it’s very likely that you have to bridge the gaps. That is where you write at the interfaces where ideas meet.
Of course for this style of writing, where one merely arranges ideas and fills in the gaps, you must first have an set of ideas to work with. Now if your plan is to look for those ideas by recalling them i.e to brainstorm, you’ve lost the battle before you’ve even started. The brain is terrible at storing at information and retrieving at will, unless you’re some wizard with eidetic memory.
But to have a library of ideas, you have to write them first. The framework for that is given by the Zettelkasten method. What this translates to in a real life situations is that whenever you learn something - collect ideas, you absorb information and mould it into a building block, so to speak, in your text based knowledge infrastructure.
Now that learning happens through writing, it becomes purpose driven.