> AFAIK this phase was never completed, and IMO this is a major reason why the web is so centralized today, why users have no control of their data, and why it'

> AFAIK this phase was never completed, and IMO this is a major reason why the web is so centralized today, why users have no control of their data, and why it's primarily aimed towards consumers. ¶ Had publishing content been as easy as consuming it was from the start, there would've been more tooling built around this concept The original implementation did support publishing content. The problem is that the Web was prototyped on a NeXTCube, and nobody could run it because nobody else had NeXT machines. Most people's exposure to the Web was through the linemode browser or Mosaic, and neither one implemented the full range of functionality that WorldWideWeb.app did (including content authoring). Andreessen reportedly did want a way to do annotations in the browser but it didn't happen precisely because the only way he could think to do it was by coordinating everything with a centralized service. As for linking specifically, early versions of HTTP had a LINK method, but it was underspecified and eventually removed in later revisions of the standard. If you operate the server, though, you can make it support whatever methods you want, so there's no reason people couldn't take matters into their own hands and bring it back so long as they're able to figure out how it should actually work.