As I understand it Nelson's vision is still not grasped today.
His original hyperlink was bi-directional.
Pause and think about that for a moment.
In practice, at the time, the target in a client-server implementation
had no way to store "return" links pointing to it. The WWW could not
rise to Nelson's ambition. Maybe that's changed today, and we could
certainly rewrite servers or create a more peer-oriented way of going
about things.
The upshot is a totally changed "Web", in which search is
fundamentally different from a centralised index a-la Google.
Not sure I understand. Can you explain? How would a bidirectional link work? What would be an example? And how would that free us from the need for centralized search engines, for example? I ask not because I doubt, but because I have a hunch that you may be right. Thanks!
I suppose there's three things to talk about. Ted Nelson. The idea.
Possible implementation.
First, Ted.
He's clearly a genius, also the type who has a lot of magical
thinking. Which I mean that in a mostly good way.
Then there's the theory. Graph theory. The WWW (basic concept) is an
ephemeral directed graph under asymmetrical navigation, such that
sources are initiated by clients (requests) to destinations on
listening servers (respondents). It's a stateless static document
model, and within a document the outdegree is the (known) number of
links in a document, and the indegree is the (unknown) number of all
other links in documents out there that point to it. Links are just
text strings (URLS).
As it stands the only way to know the indegree is to spider all
documents exhaustively and count. Larry Page's "rank" did this, giving
a rough metric of how "important" a document was by its popularity
(indegree), which launched Google.
As I understood it, Ted wanted links that were containers of other
links that referred to them. The server would somehow need to watch
for these coming in, and remember them. The upshot is a much richer
information space.
Implementation. Can-o-worms. What happens if a million pages refer to
yours? How to prune dead return links? How to trust servers to be
honest? It's completely different privacy/visibility universe.
So I think it's just a theoretical pipe dream.
But I hope I may be proved wrong by really smart people. I think the
key is to move away from client-server to something a bit more
flattened, perhaps with IPFS or suchlike underneath. Who knows.
As I understand it Nelson's vision is still not grasped today.
His original hyperlink was bi-directional.
Pause and think about that for a moment.
In practice, at the time, the target in a client-server implementation had no way to store "return" links pointing to it. The WWW could not rise to Nelson's ambition. Maybe that's changed today, and we could certainly rewrite servers or create a more peer-oriented way of going about things.
The upshot is a totally changed "Web", in which search is fundamentally different from a centralised index a-la Google.