Re: Federation protocols

On 31 May 2013 11:50, Michał 'rysiek' Woźniak <rysiek@fwioo.pl> wrote:

> Dnia piątek, 31 maja 2013 o 06:59:52 Melvin Carvalho napisał(a):
> > On 30 May 2013 20:26, Michał 'rysiek' Woźniak <rysiek@fwioo.pl> wrote:
> > > Hi there,
> > >
> > > I'm #NewHere, to use a popular cliche on federated social networks. I
> am
> > > an active user of Diaspora, Friendica and StatusNet (soon to be
> > > converted to pump.io).
> > >
> > > I am also a stern proponent of free and open federation protocols and
> > > networks.
> > >
> > > For a while now I have seen Friendica as a great project, allowing the
> > > different federated social networks (Diaspora and OStatus-compatible)
> to
> > > be able to communicate and for a single, huge federated network.
> > >
> > > I am however baffled by the different approaches and protocols being
> used
> > > in
> > > distributed social network projects. With the introduction of Red,
> > > pump.io ,
> > > tent.io and other projects not exactly compatible with protocols
> already
> > > utilised, I feel we are not heading in the right direction.
> > >
> > > What I feel we need is a single, extensible, well-defined protocol, or
> > > suite
> > > of protocols, that we can build a single, compatible, interoperable
> > > federated
> > > social network upon.
> > >
> > > Right now we have OStatus, Diaspora's protocol, DFRN (used by
> Friendica)
> > > and
> > > the protocols that are used by Red, tent.io and pump.io, that I am not
> > > even
> > > sure are properly defined anywhere.
> > >
> > > If we do not get together and devise a single, workable protocol for
> all
> > > such
> > > services to use, the Network Effect will always work against us,
> instead
> > > of working for us:
> > > http://rys.io/en/88
> > >
> > > So my questions are:
> > >  - is this the right list to start this discussion?
> > >  - is there any work done in this regard?
> > >  - if some, where are we on that road?
> >
> > The web was designed to be social from day 1.  There are standards for
> this
> > kind of thing, but they are highly underused, with perhaps, the exception
> > of facebook.
>
> Are you talking about how Facebook uses XMPP? Otherwise, I don't see the
> "open
> social interoperable standard" in Facebook (although, granted, I'm not a
> user
> there).
>

There are many things about facebook that are not ideal, such as privacy
issues and centralization, but it is a market leader and some of the
technology is worth examining, imho

There is the xmpp, but I'm more referring to how facebook uses web
standards to federate.  Facebook federation is found on over 10% of all
websites, so they must be doing something scalable.  The techniques are to
leverage HTTP via the open graph protocol

http://ogp.me/

Notice that this is a protocol anyone can use, that is independent of
facebook centralization.  It's also one of the few that follows web
standards quite well.


>
> > There is a tendency to want to create, rather than, reuse.  However the
> new
> > 'protocols' tend to scale at most to themselves, and it's relatively rare
> > that heterogeneous systems can communicate.
>
> That is true, but that is *precisely* why we have working groups like this
> one! Think HTML, XMPP, SMTP. Somehow it was possible to get these defined,
> documented and rolled out across the web.
>

+1 im not against other protocols, but getting a good solution to html +
http into wider deployment alone will be a big win, for the web


>
> > This group made a big bet on OStatus about 2-3 years ago, and arguably it
> > has not exceeded expectations.
>
> True, but:
>  - it gained some traction;
>  - it has important flaws (privacy-wise) that have been opointed out.
>

I think OStatus was a reasonable thing to bet on, and agree it had some
traction.  However we've learnt some lessons in the last few years, and
linked data has been steadily rising.  Consider the following chart:

http://www.google.com/trends/explore?q=ostatus#q=ostatus%2C%20%22linked%20data%22&cmpt=q

Some of the founders of OStatus projects have moved on to other things, and
I think it's time to give linked data a chance.


>
> Let's fix it. Let's choose a protocol that does not have these flaws and
> put
> our weight behind it.
>

I think James Snell has made a fantasitc update to activity streams
(codename activity streams 2.0) which OStatus is based on, to include solid
linked data principles:

http://www.chmod777self.com/2013/05/time-for-updated-activity-streams.html

There's even talk of this going through the IETF, which I expect would not
take long


>
> > There seems to be an effort to steer things back to standards and best
> > practices, from a high level perspective.  I'm optimistic that this new
> > approach will lead to interop, for those that get on board ...
>
> I do hope so. Otherwise we have no chance to get people out of walled
> gardens.
>

+1 :)


>
> --
> Pozdrawiam
> Michał "rysiek" Woźniak
>
> Fundacja Wolnego i Otwartego Oprogramowania
>