And I think wikis are still quite limited in that bit I quoted: "Englebert is not advocating that we perform artificial acts with documents by superimposing structure on them. Instead, he advocates that we capture the inherent structure in all forms of human expression in order to make them easier for people to navigate through, view in different ways, and hyperlink."
They have hyperlinks, sure, but otherwise most pages in wikis have only those views that can be created through operations on a hierarchical table of contents. The structure is imposed by the tech, there, rather than the expression of knowledge itself. The blocks in Roam - https://thinkstack.club/how-to-use-roam-block-reference-menu... - seems like a major iteration on that to me in terms of moving towards units of knowledge as the fundamental element instead of "pages." (It's also worth noting that Foam doesn't do this - https://github.com/foambubble/foam/issues/294 - though apparently Obsidian does.)
And yes, it's closer to Engelbert's vision. We've been getting closer to that (and the Memex) for a long time, and we'll continue to do so. They are distant visions what a future could look like, while we barely understand knowledge work. I mean, the term itself is already an improvement of clarity on the ideas of the memex.
If you compare knowledge work to Mathematics, I'd guess we're shortly before Elements. We're at best about to understand the basic building blocks. Which I find tremendously exciting - we're still at the very beginning of exploration.
In my head they're akin methods, and so in software that idea of "call this from lots of different places, edit it in a centralized way" concept is pretty well established.
In terms of "a single document flowing with a single voice" you can't necessarily do that from assembling an arbitrary collection of blocks like that, but for "good enough for effective in-corp collaboriation"? Sure!
I think that (and the problem of bait-and-switch) still need an answer before it's usable beyond small groups. Maybe small groups is good enough for now. If it goes in any form like software development, we'll at some point split into two camps. One will say YOLO and just implement shared blocks, trusting in good will, and the other group will noodle on foundational principles to get this right. (Group 1 will make the money, group 2 will have the fun. As always ;)
Foam is simply the next evolution - the shared tool is well-understood, we're now getting an augmented local version. And after that, we'll get something that's far more multi-user friendly, too. It's a pendulum that seems to swing back and forth in our industry.
The thing that I hope Foam & Friends will generate is a much better understanding of knowledge "operations", so we can understand the fundamental structures underneath better. (That's what happened with git & signed DAGs as version log. That's what happened with text & CRDTs as a model of shared edits)