In a review of Niklas Luhmann's Theories of Distinction: Redescribing the Descriptions of Modernity. Edited, with an introduction by William Rasch (Stanford University Press, 2002) in the Canadian Journal of Sociology Online (November - December 2002 [Luhmann.pdf]) Marion Bute finds that "Luhmann is not for those who prefer an extended, linear, rational argument. Increasingly, his later work became (some might say degenerated into) an almost schizophrenic thought (albeit not word) salad." And then asks the rhetorical question: "How could it be otherwise, given his famous card-shuffling method of working?"
I would agree that (i) "Luhmann is not for those who prefer an extended, linear, rational argument." I might also agree that his non-linear and associative style "degenerated" in his later work into something like a "thought salad." But I would (ii) reject the description of his writing as "card-shuffling." Rather than simply "shuffling" his index cards before writing, he "selected" cards that seemed relevant to him and followed the leads he "found" in the notes he carefully (and in a highly controlled way) had taken during several decades, trying to "discover" connections between them. See Luhmann's Zettelkasten. This is something that he did during his entire life. I would also reject (iii) the suggestion of determinism in the claim. While the tools we use in note-taking and writing may indeed have some influence on our work, they certainly do not have to predetermine its outcome in every detail. In any case, before I would accept such a claim, I would like to see some proof.
Indeed, if it is true that Luhmann's writing did indeed degenerate or change in his later work, then this is sufficient proof that things "could be otherwise" simply because they were so before. The perceived change probably was not so much due to his method of note-taking, as it was due to a change in how he processed these notes in composing his articles and books.
That being said, I should perhaps also point out that I believe that all writing benefits from "extended, linear, rational argument." This should be the distinguishing mark of the finished product, while it need not and probably cannot be a characteristic of the process of discovery. (See also Wittgenstein on Note-taking.)
When Bute finds that "in reading Luhmann, before long one inevitably experiences a “clang” as a thought resonates with something pre-existing but only half-formed, evoking a “now that’s something worth thinking about”, and you put the book down gently for awhile lost in thought," she describes an effect good note-taking can produce in the process of discovery—even if there is no inevitability here either.
1. None of this means that Luhmann's method is without its problems. See also Critique of Zettelkästen.