Backfeed, AKA Reverse Syndication

Getting Feedback Back

So you've syndicated clones of your data out to Twitter, Facebook, Flickr, Openphoto, Tumblr, etc. But what happens when people on those other networks comment/like/otherwise respond?

Types of Response

• Comments/replies/notes - basic textual commenting
• Favourites/likes - primitive 'positive feedback' type
• Mention - A reference to the content that is perhaps not in direct reply to it
  • E.G. a 'share', 'bookmark'
  • Is this sufficiently different?

APIs

Some APIs have better functionality for this than others:

• Twitter
  • Unreliable, undocumented, unofficial 'related_results' endpoint.
  • Could the search API be used?
    • Only returns recent results
  • How else to get responses from Twitter?
• Facebook
  • Extremely slow API, so caching/async likely to be required to avoid vexation
  • OBJECT_ID/likes connection
  • OBJECT_ID/comments connection
  • Returns extremely unclear failed results for API calls to posts by under-18s
    • This is bloody annoying for us minors. It's more "bothersome" than "protecting" -- barnabywalters
  • Anything else?
  • How best to represent external users?
    • Facebook URI (not preferable)
    • Can we grab their personal URI from their profile?
    • Detect comments from originating user?
• Flickr
  • flickr.photos.comments.getList
  • flickr.photos.getFavorites
• Others?

Debate

Do these responses belong on your indie page, at the original source of the content? There are storage, legal, data ownership, UX, and other questions.

• Yes:
  • Gives people a choice about where to author their replies
  • Enhanced UX (debatable depending on service used)
  • Ease of use, low barrier of entry
  • Meets the majority of users who aren't technical or don't feel strongly where they currently are: in the silos.
  • barnabywalters, snarfed vote yes.

• No:
  • Generic, open systems such as pingback are preferable to backfeed from closed silos (although open response systems are another discussion entirely).
  • Responders didn't choose to publish at the original source. In rare cases, this could cause legal problems.
  • Imposes more overhead.
  • May antagonize silos. They often try to prevent this kind of outward syndication of their data. One example (of many).

Implementations

• singpolyma syndicates posts out to Twitter and runs a cronjob that polls the replies/mentions (authenticated) API endpoint to pull in responses.

• brid.gy lets you register your WordPress blog along with your Facebook, Twitter, and Google+ accounts. It then regulary searches those sites for links to your domain, fetches the comments, and posts them via WordPress's XML-RPC API. GitHub repo here.

• salmon-unofficial does the same thing as brid.gy, except via the Salmon protocol instead of WordPress XML-RPC. It discovers original source sites via LRDD, WebFinger, and XRD, as specified by Salmon, so you don't need to register your domain, just your social media account(s). GitHub repo here.
  Caveat hacker, it's brittle and not well tested for interoperability. However, instead of doubling down on Salmon by trying to improve salmon-unofficial, snarfed plans to put that time and effort into a similar project that implements IndieWeb replies.

Related

• pingback protocol - currently implemented by most major blogging platforms, i.e. Wordpress, and available as a standalone project to use on non-Wordpress sites
• Salmon protocol
• Juvia - a commenting system you can install on your own server similar to Disqus and Intense Debate