Maybe there's a kind of synthesis which is done in long, specific documents which can't occur in LinkedFragments - because in a real document, meaning is actually ground even finer. Subtle context dependency between all the words in the long document mean that multiple meanings can be superpositional across the sentences. The sentences bear more meaning (FractalLoading) because of their tight, collocation with the next and next and next sentences in the document. And this is only achieved by custom selection of every word.
In contrast, attempts at a clunky ReUse, by replacing fine-tuned phrasing with links to generic modules elsewhere in the hypertext, lose meaning. Those links are low-bandwidth, weak interfaces. The fractally loaded, superpositional meaning doesn't cross them.
Maybe modularization only fails to carry sufficient meaning in hierarchies. In proper semi-lattices there is extra meaning from the overlapping, sharing of connections. Meaning is also superpositional on links. So maybe it's a TradeOff between meaning loaded on the subtle large-scale arangement of words, and meaning loaded onto smaller fragments + links.
Reuse and hypertext does not imply the death of long documents, only the relief of economic pressures that push communication through artifacts into excessively costly enterprises – and long, specific documents just happen to be one such expensive tool. By creating a bigger net, the problem of communication might be more natural to solve, although in practice we will no doubt find there is too much confusion about what to do and when. Note that the total volume of a hypertexts (by summing each node) also tends to be much larger than documents, and much harder to read. So, as a counterforce, there is still need for the essay. It's just that really shitty, boring essays are no longer acceptable.
Reuse and hypertext does not imply the death of long documents, only the relief of economic pressures that push communication through artifacts into excessively costly enterprises
At first I thought you were disagreeing with me. Now I think you're agreeing in the sense of saying that economic pressures mean that yes, in general, you get re-usability through making smaller, more autonomous, more generic chunks (modularization, containerization). It's just that you don't always think such "economic pressures" are operant. Is this right?
Note that the total volume of a hypertexts (by summing each node) also tends to be much larger than documents
OK, this I buy. But there's less of an expectation that you read them all the way through. But certainly, there is still a role for good syntheses and summaries. But yeah, not boring gratuitous essays.