Federated video

Jump to navigation Jump to search

There is open-source software for uploading large files and for playing video on the web using HTML5. Just those two components would give us an open-source YouTube that anyone could build (for a small number of users, at least).

I'm proposing adding the following features:

  1. a public API which allows copying of videos (at least by default) as well as all associated meta-information (mainly: title, description, upload date, who uploaded, keywords)
  2. the ability to import videos and their meta-data via this same interface
  3. tracking of other sites using the same (or compatible) software -- that is, each site would have a list of all the other sites (from which each site's admin(s) could block specific sites if needed)
  4. a public API which makes this list available (this could be extremely simple)
  5. the ability to import this list from other sites with the same API
  6. the ability to build a search-index which includes videos and data from all sites in the list, regardless of whether the videos had been imported or not

Some hosts (nodes) won't have sufficient storage space to just import everything from every site, but the cross-site index would allow you to use any individual node to search across all of the nodes.

In that way, you could build up a service with the potential to expand indefinitely and challenge YouTube, while not being subject to corporate priorities or even the priorities of any individual node-owner. If the DHS took down an individual domain, most or all of the videos originating from that site would still exist elsewhere, and be findable with the search engine.

Plus we could start experimenting with a lot of innovations that YouTube hasn't bothered with, such as user reputation management, downloading playlists as a single file... we've been trapped in their static playground for far too long.


  • This was originally proposed here (in a comment).