Ted Nelson invented the concept of hypertext, an achievement that has earned him many accolades including (in 2001) the French honour of "Officier des Arts et Lettres". Arts and letters not, interestingly enough, computer science, where the impact of Ted's particular vision has been far less pronounced than he might have wished as a young man.
In Literary Machines he tried to pinpoint the meaning of his brainchild thus: "By 'hypertext' I mean nonsequential writing - text that branches and allows choice to the reader, best read at an interactive screen."
Interestingly enough, this definition positions hypertext as end product of laying down of non-sequential, branched associations, as opposed to the internal mechanism of a universal system of reference. Yet he has devoted most of his life to just such a system - Xanadu.
At the heart of the practical side of Xanadu was Ted's notion of transclusion, whereby author-creators could 'quote' each other using shortcuts that didn't actually generate a new copy of the quoted material. Transclusion was to have driven a system of royalty micropayments, the units of which Ted called ernies.
According to Wolf, the project had some serious technical limitations - everything in the library had to somehow be simultaneously accessible which implied vast memory requirements and almost unpredented computing power.
Perhaps there are other more fundamental problems with the approach? For a start it seems to belong to the un-wholey 'atomist' alliance (genes, particles etc.) which describes any system as a compilation of discreet units. As I pointed out in my previous post, high-order human knowledge is laregely symbolic - and symbols (unlike mere correlations) are hard to learn one at a time...