The Social Smolnet

don’t hesitate to ask your question on the user mailing-list. Documentation and websites should really be improved.

It might have been an email thread. Or a lobste.rs comment. It was a discussion about yet another attempt at a new decentralized social protocol.

I suppose you are talking about this (edit: ~ploum clarifies this was not the one being talked about in his post, my bad for the assumption and taking up space here). The discussion began with ~susam developing a tool inspired by Kagi Small Web. When the discussion came to the point of developing a browser extension and how it might work while retaining its decentralized nature, I thought it was good time to “standardize” the recommendation lists as used in wander (for now, let's call this wrls). The difference from ordinary recommendation lists (like blogrolls) is allowing recursive references to other wrls like it, each of which then can also reference other such wrls and so on. So a(n external (due to CORS etc.)) tool like wander can be used to traverse this graph of wrls, to discover new websites, maybe even take advantage of network effect that this might entail.

With a schema of wrls in place (which I initially imagined to be something like struct wrl { recommendations: list<link_to<webpage>>, self: link_to<wrl>, wrl_recommendations: list<link_to<wrl>> };), it could be put in XML or JSON and linked from a webpage with some markup that can be detected by tools and extensions supporting wrls, like feed readers detect feeds. Then ~robalex pointed out how people already use OPML for blogrolls, which means to be compatible with them, wrls have to support hierarchical lists. Then I thought that it should also support tags-based classification. Others might find something else to add in this and so on.

While learning about IndieWeb, I learnt about microformats, which I thought might be good enough for our purposes. The wrls can just be stored in plain HTML with a little bit of markup, which the tools can parse to find out the recommendations. Ordinary visitors can also follow the links manually, without installing a tool, and it'd work without JS as well! My comment in that thread contains an example of what it might look like.

---

Sure, we can do without this, as we've been doing, quite well in fact, with existing protocols and standards. The point I suppose is to make it easier for those who want it this way.