In spirit of the top level comment, you do not do this entirely on purpose. At best you recognize what you are doing and rationilize it, but if you emulate the reader at this moment, you have to have been conscious of the structure and the gaps in it that you are trying to fill.
On the other hand, in my own experience, if it's possible that I forgot to mention something and try to tack on the information, it's also quite likely that I lost oversight, that the information was already coded in and I just forgot and thus repeat myself.
For example, I do frequently lose drafts by accident, and I'm particularly happy when I manage to reformulate approximately the same structure, but it has also been the case that I could not even remember what I was going to say. Some other times, I just completely lose it, so to speak, and butcher everything with deranged copy-pastes rearranged in the wrong place. Now my conclusion has been for a long time that this is simply due to a poor working memory, quite analogous to the browser crashes eating my drafts. On the up-side, the aphasia is self-healing to a degree, so I find it is a good memory training when I have to formulate a text before I can get to the writing, but the extent of my abilities is severely limited at that. Inasmuch as this is analogous to the thought process itself, the writing is symptomatic. However, there are also those cases where my thought process was not verbose, so translating from idetic memory to linear text is a huge problem, for example with regards to conjunctions.
On that point you will note if you have ever written involved if-then-else structures, that repition and redundancy is impossible to avoid. Conversely though, recursive algorithms and functional data-structures that do look neat on paper do potentially have much worse run-time complexity, depending on the interpreter, for lack of a better word. More over, to stick with the programming metaphor, text may contain comments on a meta level, but we've all seen those useless comments and unhelpful documentation.
Anyhow, what I was hoping to say was that redundancy has benefits. Scientific papers do it all the time and possibly on purpose. This is in no small part due to the medium. Locality matters, so in dialog you might be told where something was left unclear, or already known, and responses might serve redundancy for you, but in writing maybe there is the expectation of certain conventions. Legal writing, verdicts and such are another fine example of elaborate structure with sometimes staggering amounts of repitition.
Finally, keeping the Title in mind, the itterative method you describe can count as "write more", and if it works like play generally does it will improve and the conclusion will actually be "but shorter" on average. Whereas, if the itterations tend to grow longer, it might be indicative of a lack of understanding indeed.