⚠️ This is a draft, maybe not yet worthy of your eyes.
The scope and background of this vision: This is a combination of James Burke's visions as interpreted and reframed by me and visions for computing and hypertext / hypermedia that never truly was realized (much or which surely inspired Burke's original vision).
CONSIDER: An infinite canvas version of this article (can we build it on Google Docs / Drive?)
I've recently thought about consuming versus creating Knowledge Webs (see What Should the Knowledge Web Be?)... The only clear situations I've found where people consume them happily is the Knowledge Trails on the Web as curated by James Burke in Connections (here everything is set, you can but pause, stop or skip forward the video) ...and there are Wikipedia rabbit holes where the content is set (though it is text so you get to choose what & how much & how you read) but you choose the trail and at the end of it there is no Knowledge Tail to look back at, no artifact (though this could easily be built).
Anyways, I'm leaning towards creation, but making it as easy as possible, and probably making it more into discovery & scrapbooking / remixing than writing a book. The most elegant system I envision fills the whole spectrum, use as:
An augmented Wikipedia; no information is modified. Delivered as a browser extension / Userscript primarily through a (web) app should probably live side by side (& maybe you go there to see your previous trails etc.
Tools: Timeline and globe (in solar system) will be useful for just the current article!
[Image: Superimpose timeline + globe from my 2023 prototype on wikipedia?]
Trails: Your (Wikipedia) Knowledge Journeys / Trails
[ Image: Maybe a really old illustration of this, like from the Memex ]
Connections are highlighted, easier to find, labeled and displayed (be default) in list just like in old K-Web prototypes
[ Image: Screenshot of old XML / The Brain article's Connections list ]
Users can choose to personalize the rendering, ex,
prefer simplified english (when available)
avoid walls of text: Hide all but the introduction behind "> Read more..."
Marginalia (see Computers for Cynics by Ted Nelson on why this never became a thing in computing, but should have).
Comment on the whole article, a section, paragraph, or word (move it if you want; default is perhaps in the margin next to the subject block)
Draw, sketch, scribble, shapes, whatever... like tldraw, excalidraw or MS Paint on top...
Co-presence
See others visiting the page... see their cursors (and eyes if they have eyetracking)
"Multiplayer" (a different topic?) share your trail, marginalia etc with others as viewers or editors.
Remixing, modification & multiple sources. Delivered as separate (web) application (seamlessly sharing data with the browser extension)
This is where Wikipedia just becomes another source...
Here a node may default to show Wikipedia content but you can also create your own or have other sources (ex Old K-Web data & nodes generated from Burke's work)
High resolution reference system:
Use a section, paragraph, sentence or word from any web source, the text will be pasted into your node as plain text and then meta data will have information on exactly where it came from as well as have a local archived copy of the source / page.
History, trails & personalisation
Web pages used to color visited links differently, let's take that and make that more powerful: Have information on when you visited a page and if you actually read it (staying a second should show it as visited but not read (though maybe you read it earlier outside the system so maybe also make possible to "Mark as read"?)
Trails are always stored automatically
Trails can be curated on top of this...
Viewing several trails as a Graph makes a (Knowledge) Web...
On the user experience: Bring the experience to where the people are, try not to disrupt flow / existing user interaction, try to add to them; augment them (at first anyway):
Browser extension (or Userscript): On Wikipedia
Separate (web) application, use same database as browser extension.
Isn't all this useless now that we have AI / LLMs?
Alternatively: It's clear that it will be useless in 0.5-2 years so why bother?
TODO Think...
-[] ...about all the above and ask "Can an LLM do / replace this?" ...and "How can AI improve this?"
-[] ...about what is now possible that we couldn't even dream about before LLMs
-[] ...and how to imagine what is beyond self evident; beyond the adjacent possible (and also what now showed up as possible)