The original vision of hypertext was textual and timeless. When Vannevar Bush and Ted Nelson imagined hypertext, they imagined a conversation between documents, data, and other media, not individual banter.
Importantly, additions to the conversation were not conceived of as replies, but as *extensions* of a large static but continuously evolving network of knowledge, one that allowed many different paths.
This difference between reply and extension is crucial to grasp.
Consider the famous "bow" example in Bush's As We May Think. Thinking on the problem of knowledge in a society of niche expertise Bush imagines the workflow of a researcher of the future.
What happens in this example? The user of Bush's memex device is interested in why the short bow was better than the long bow in skirmishes of the Crusades. She starts by reading a basic account of the different bows in an encyclopedia, then jumps to a more detailed account in which she finds related material at more depth. She links these two things, the summary and the supporting detail, via Bush's version of the hypertext link.
And she keeps going. That new article is linked to something else so that gets linked as well. She makes additions directly to the text via annotation. She handwrites some notes with a stylus and those notes get linked in to the web.
At some point, the researcher realizes that the elastic properties of local materials may have had a great deal to do with the bow. And so now a side trail develops where he learns all she can about elasticity, but along the way she documents her learning, summarizes it, links it. She pulls together medieval history with data on elasticity of different materials with explanations of the physics of bows and various equations.
She adds a pages of her own, scanning them in through a "dry photography" process, extending the network.
This is all in her own machine, capturing her process, her emerging knowledge graph.
But then comes the kicker. She's talking with a friend some years later and the friend has an interest in why communities resist innovation.
Our original reader realizes that the network of relations she's put together on the bow might be of help to that investigation.
The original user still has the trails she made and her notes and articles from her investigation. So she runs a trail duplicator and hands her friend her network of knowledge. Now the friend builds that out, and further develops the idea, but adding this other facet to it -- using the linked data and text about how local materials play a part in adoption of inventions. This new set of "trails" provides a useful counterpoint to broad statements about not-built-here-ism that our new reader was making.
Next: Structure Grows