In the Xanadu world, every new book published would be posted online for purchase, but also transcludable; you could remix the work and the transclusion system would ensure the publisher got paid for the part of the work you used by having the system just pass the payments along across each transcluded work. Effectively, you'd publish your remix in the form of a Git diff or IPS file (Ted used the metaphor of Edit Decision Lists) and your browser would just buy the pages of the book that were copied from and paste them into the remixed work.
This is, of course, not how copyright works in the slightest. First off, most people with valuable works do not want to automatically authorize remixes or modifications of their work even if they do get paid. They want to have prior restraint and exclusive license agreements with specific users of their work for the sake of brand maintenance. There is no legal basis for transclusive remixing of content (see Micro Star v. Form Gen). On the flip-side, if someone were to write, say, a review of a work; the copyright owner is entitled to no control or remuneration whatsoever for any quotations you use in that review. That's called "fair use" and it can only be adjudicated by a judge, not a computer program.
BTW, what I'm talking about already exists, it's called YouTube Content ID. YouTube runs literally every upload they have against a database of copyright samples to effectively generate a list of transclusion backlinks after-the-fact. All the problems I mentioned above are problems with YouTube Content ID, not Xanadu - there are plenty of publishers whose policy is "block", and it is their right to do that; making a fair use of content will trip the filters and you have to argue with YouTube to do what you are legally entitled to do, etc. It is entirely an engineer's view of copyright law, which is just laughably bad.