I loved watching Hyperland. This documentary written by Douglas Adams for BBC Two in 1990 is an amazing insight into the dream for what Hypertext might bring with it. Full of non-linear adventures, segues and "Micons".
In some ways the web we got is this reality - hypertext allows you to link out to virtually any information - but what I don't think we've arrived at is an understanding of the concept of non-linear.
What hypertext presented was a concept of non-linearity trapped inside a linear one. The implementation, somewhat limited by peoples ability to grasp new concepts, remained tied to the linear nature of print. Things had an order to them, a sequence, and so as the web progressed those concepts became re-engrained. To the point where we seem to have regressed back to linearity and sequence.
Hypertext is dead.
Text As Polysynchronous
But what if the concept of what is non-linear related to our concept of time rather than the object?
What if we read in a linear and sequential manner but hypertext was an overlay of time?
Could hypertext instead be away of seeing and interacting around the text rather than being of the text itself?
Could hypertext be rethought as Polysynchronous Text?
A colleague of mine did some interesting research into developing the concept of Polysynchronous Learning from which I've drawn this concept of non-linear as a relation to time rather than object. I've also drawn on the nature of the fedwiki platform and global participation. See Prayer Wheel
Re-imagining hypertext as polysynchronous provides an interesting platform for the imagination. A text platform where the narrative is linear but the surrounding interaction - reading, commenting, note taking, expanding, defining - is both part of the object and yet separate has some alluring possibilities. Perhaps this is kind of functionality you develop in A SFW "Book".
Related: Walking The Lines, The Web That Was
This is actually the original conception of hypertext -- via Bush, Nelson, and Engelbart. Not really that the narrative was linear. But the hyperrelations were meant to be tracked across time, as a sort of conversation. Both Xanadu and NLS kept links outside the text as overlaid annotations, and both insisted that objects kept a history which could be replayed. The fact we do not see these relations Nelson would blame on the simplifications of the vision made by people (CD developers, Brown's HES project, Mosaic) who separated publishers from conversers from readers and thought of hypertext as a "publication". -- Mike Caulfield
I had an inkling that early concepts and vision would have been more conceptually broad. Watching Hyperland was a very similar experience to that from viewing The Future of Programming 
by Bret Victor - that early conceptualisations of the future were never really met. That decisions were made that compromised and changed the path we went down. HTML is less "hypertext" and more document structure. It's fine when information is static but we're getting close to its limitations. Access to digital Information is not the problem it once was. We need to move beyond that. What I'd suggest is the next big thing is building out better digitally mediated interactions. Ones that go beyond simple back-and-forth broadcast of informational statements. Fedwiki is exciting because its the first thing I've see that does this. I think part of this is that architecturally its fundamentally different - that it's not building on the past - but building out from them. Mutating and evolving the form as well as the possibilities. -- Tim Klapdor