Re: RFC: tent.io (protocol for social networking)

On 24 Sep 2012, at 12:50, Michiel de Jong wrote:
> 1) On http://tent.io/ there is a section 'what about the federated
> social web', which i think is interesting. They describe the federated
> social web as a "network of networks of servers", in which some
> functions only work if two people are on the same server (instance), a
> slightly smaller set works if two people are on the same software, and
> only a relatively small subset will work of two people are on
> different software.
> 
> I think this is accurate, and identifies an important problem. I do
> not think that tent.io is currently proposing a solution though. But
> let's hope they will add private messaging and some other things in
> the future, and can come closer to a fully multi-feature fedsocweb
> protocol. What is admirable is that they seem to design their protocol
> independently of their server implementation, although unfortunately i
> think it's naive to think that other software projects will now
> suddenly switch to their protocol for that reason.

Maybe they have a bribe up their sleeve! ;-)

More and more I'm seeing the distinctions between various flavours of the social web, federated and decentralised and distributed and P2P, becoming real blurry. Or if they are not blurry then they should. I mean it's all really the same stuff, isn't it? I'd prefer it was then we could all work together.

> In general, i believe in a federated social web where there is one
> name space of contacts, yet multiple languages (OStatus, xmpp,
> ActivityPump, tent.io, diaspora-protocol, StatusNet-protocol, zot,
> etcetera, but also just good old smtp) in which two servers, or a
> client and a server, can communicate with each other.

What scares me about this is the "where there is one name space of contacts". How many namespaces for contacts do we have now? More than one? I ask in earnest as I'm not really sure at which level you are conceptualising the idea of a namespace for contacts. Do we call each website with users a namespace? Or each protocol a namespace? Do we really feel that people are going to align behind one? I really don't think so.

I feel our model has to cater for multiple namespaces for contacts. Can it? Can we create an meta namespaces-for-contacts wrapper? And do it quickly and quietly before someone else creates an incompatible meta namespaces-for-contacts wrapper? And maybe that won't matter actually as a wrapper should be able to wrap another wrapper, right?

> I think we
> should stop looking for "the" protocol, and just be happy that there
> is such a wealth of different specialized protocol being used for
> different tasks. All those protocols together can be used alongside
> each other, and not all servers need support all protocols.

Agreed. Back to that Melvin Carvalho's matrix idea. 

> As long as the namespace in which each account refers to all its
> contacts is generic across all those protocols.

Oh darn!... :-(

Received on Monday, 24 September 2012 15:01:39 UTC