Re: Federation protocols

Hi there,

Dnia piątek, 31 maja 2013 o 01:08:11 Nick Jennings napisał(a):
>  I agree with your concerns, which is why I've started the project
> Sockethub[1] - taking a different approach to the problem.

Thanks for the info, that's an interesting project indeed.

>  Instead of trying to heard everyone into a single protocol (which has so
> far been attempted many times and things just seem to get worse), the idea
> is to implement an open, polyglot service that can "speak" all of the
> different protocols and APIs. Whether it be Friendica. Diaspora, StatusNet,
> Facebook, Twitter, XMPP, or whatever other form of social messaging comes
> into play down the road, I think it's important we start working to avoid
> all of the programming work that gets thrown out the minute a new protocol
> is adopted, rendering old code (old protocol) useless.

> If we continue trying to write solutions that try to implement the
> "kitchen-sink" (UI, UX, app logic, protocol support, server backend,
> storage, scalability, federation) all while trying to remain innovative and
> pertinent, with fast paced feature iteration, choosing a few protocols to
> be "supported" (or inventing yet another new one), then this problem is
> likely to continue.

That is true. That's why I believe instead we should find a workable solution, 
a workable protocol, agree on it, and run with it.

> On the other hand, if we can split our focus into building (A) applications
> that are not married to protocols and implement as little as possible
> outside of their problem domain and (B) a federated infrastructure that
> apps can tie in with to communicate with other websites or services that
> may speak any number of protocols or APIs, then we can start to show some
> real adaptability and flexibility to not only keep up with innovation (the
> private sector is always going to be able to move faster - and that's not a
> bad thing), but also mitigate the network effect by not making it an "all
> or nothing" replacement for sites like Facebook or Twitter.
> 
> I'm interested to know what you (and anyone else) think about this
> approach, I'm still figuring a lot of this out for myself as well, and
> appreciate feedback.

I think your idea is decent, but I believe we need a similar solution in a 
different area.


I watched your presentation (BTW, I was supposed to do a Libre Educational 
Resources workshop at Re:Publica[1]) -- correct me if I'm wrong, this is a 
"client-ish" solution, right?

What I mean by "client-is" is that while it is run on a server, it in fact 
uses client-side APIs of Facebook, Twitter, OStatus, Friendica, Diaspora, etc. 
It allows developers to create *client* software/webapps that can be (to some 
extent) service-agnostic, but *only* client software.


What I believe we really need is interoperable *server* software. We need 
Diaspora, Friendica, OStatus, pump.io, tent.io and any other possible 
federated service/protocol *talking* to *each other*. That's the Goal!

I don't see how SocketHub can help here, frankly. Can it stand between a 
Diaspora instance and a pump.io instance and make them interoperable so that a 
user@pump.io can seamlessly communicate (comment, friend, poke, etc) with 
user@diaspora, *without* setting an account with Diaspora?

That's the big question and that's the really important part. If it can, that 
would be magnificent, and I would start proselytising SocketHub right now!

Making client (web)applications development easier and more service-agnostic 
is great, don't get me wrong. But the real problem is that:
 1. users have to have accounts on all the networks they want to communicate
    through;
 2. all this private data then goes into silos, walled gardens, that are not
    in any way interoperable.

But it gave me an idea for a project -- an abstracted "social network 
protocols" layer (this can be, I believe, grabbed from Friendica source code), 
well implemented and tested with all supported decentralised, federated 
services, that other developers can just grab and build their social network 
of choice upon.

A "server-ish" SocketHub, if you will. ;)

This will, however, be in jeopardy of falling into the standard (pun intended) 
problem, nicely summarized by XKCD:
http://xkcd.com/927/

[1] http://www.re-publica.de/en/sessions/european-libre-education-workshop 

-- 
Pozdrawiam
Michał "rysiek" Woźniak

Fundacja Wolnego i Otwartego Oprogramowania