The "knowledge units" mentioned in this article are similar to the concept of a "learning object" in the digital education space.
There is a well known tension between making chunks that can be easily reusable in multiple contexts, and their effectiveness as teaching devices, which is called the "The Reusability Paradox", see https://cnx.org/contents/2tQZVsKy@19/The-Reusability-Paradox
I've been thinking a lot about this and I have come up with a workaround: you can have (medium sized) reusable chunks, but you don't present the chunks as standalone content. Instead you build them into narrative playlists, wrapping each chunk with an intro and outro text that link them to a coherent narrative.
Each chunk might also need to contain "conditional text" to hide/show certain parts of the explanation depending on the narrative in which it is included, but that's for later.
Your comment resonates deeply with a tension I‘m struggling with as a lawyer when writing legal briefs:
On the one hand, the desire to present each argument as one independent, bite-sized piece. So they are easy to grasp for the judge, and easy to refer to in discussion.
And on the other hand, an almost „gravitational pull“ that causes individual arguments to explain themselves in terms of others, blurring boundaries between individual points. A pressure for any one point within the argument to go into recursion. Which might make the argument stronger and more convincing. But which also blurs its contours. And leads to redundancy within the document.
I have not yet found a way to gracefully solve this tension.
There is a well known tension between making chunks that can be easily reusable in multiple contexts, and their effectiveness as teaching devices, which is called the "The Reusability Paradox", see https://cnx.org/contents/2tQZVsKy@19/The-Reusability-Paradox
I've been thinking a lot about this and I have come up with a workaround: you can have (medium sized) reusable chunks, but you don't present the chunks as standalone content. Instead you build them into narrative playlists, wrapping each chunk with an intro and outro text that link them to a coherent narrative.
Each chunk might also need to contain "conditional text" to hide/show certain parts of the explanation depending on the narrative in which it is included, but that's for later.