> A language for the markup of hypertext documents (no, HTML doesn't do that)
What is missing? Specially when accounting for anchor links being able to target protocols other than HTTPs.
> A strategy for recording an annotative trail
Could you go into descriptive detail of what you actually mean here?
The decision to embed markup in documents was horrible in its effects. It gives copyright holders extortive power over criticism, for example. You have to copy a document to mark it up, which didn't need to be the case.
As for annotative trail... imagine being able to lay out a set of pages of paper on a larger white sheet... and then make notes between two things showing how they are logically associated, in context. HTML links are blind, unidirectional, fragile, and can't go to an arbitrary position in a document.
If the whole point is to skirt around broken legislation/executive power not enforcing fair use as it should you've already lost on the fact alone that courts can interpret your layering as infringement; courts around the world have previously rule linking to be infringement as-is. But if you believe that'll actually solve things you could easily build an extension that layers HTML patches on top of another base HTML via XPath "coordinates". Of course I'm sure you've realized that if your base HTML is hosted by a hostile entity, good luck keeping it up to date with their modifications, and that's not limited to HTML as it would be a an issue regardless of whatever system you come up with and expect "the bad guys" to adopt. You either have copies, which means hosting copyrighted material, or you have references, which can go out of date.
> HTML links are blind, unidirectional, fragile, and can't go to an arbitrary position in a document.
Blind/unidirectional is fixed at the client level (e.g. the back button, history, you could do something like what this extension did if you want trees: https://www.ghacks.net/2009/08/09/display-firefox-browsing-h...). It doesn't make the slightest bit of technical sense to track backlinks on every hosted document because the incentives don't align and you'd be wasting a whole lot of space. You could have a separate service store this in a graph much like a web crawler. Don't need to reinvent HTML for that.
Fragile, you fix with copies. You can't force people to host an immutable record forever and yes this is abusable by copyright trolls, again, can't fix the law with tech.
As for linking to an arbitrary direction, you can href to a an ID and that's good enough a lot of the time. If you want any more than that you don't have to re-invent HTML: you just have to extend it with support for XPath targets. And that's something you can easily do with a browser extension all while keeping it fully backwards compatible with the web.
