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

On 24 Sep 2012, at 12:11, Melvin Carvalho wrote:
> On 24 September 2012 12:53, Daniel Harris <daniel@kendra.org.uk> wrote:
>> On 24 Sep 2012, at 09:41, Melvin Carvalho wrote:
>>> On 24 September 2012 10:25, Daniel Harris <daniel@kendra.org.uk> wrote:
>>>> Is anyone aware of http://tent.io ? What are your thoughts?
>>> 
>>> Really like it.  This is the kind of solution that has been advocated by
>>> people like Tim Berners-Lee for some time.
>> 
>> Cool and excellent! I've added Tent to:
>> 
>> http://www.w3.org/2005/Incubator/federatedsocialweb/wiki/Protocols#Tent
>> 
>>> Main difference is that tent uses http urls to describe a user.
>> 
>> Surely that could be abstracted to allow for other ways to describe a user
>> if so required?
>> 
> 
> Yes, that's the magic of the URI.  But there's subtle differences such as
> HTTP was designed to be dereference and to link to other entities etc.

OK.

>>>> Tent looks really interesting in it's solution: "Tent is a protocol for
>>>> social networking. Tent is open, decentralized, and built for the future."
>>> 
>>> Most of the FSW projects use the newly proposed acct: scheme to describe
>>> a user, and relies on webfinger.  HTTP is proven over 2 decades and mature,
>>> acct: is relatively untested, indeed, it's not even an IETF approved scheme
>>> yet.  Not saying it's *bad* but definitely living on the bleeding edge,
>>> with a smaller network effect.
>> 
>> Again, does it really matter how we describe a user? As long as
>> applications are aware of the description method then they should be able
>> to understand the data? Right?
> 
> I take you point, but the data layer is really important.  I was reading
> this quote from Linus Torvalds yesterday:
> 
> "Bad programmers worry about the code. Good programmers worry about data
> structures and their relationships."
> 
> This is one reason systems like git are so robust.

I hear you there but my issue with getting everyone to do things the same way is that this doesn't work so easily when the cat is out of the bag. And I feel the "cat is out of the bag" very much right now. It's much easier to get people to do things one way when the whole technology area is new and unexplored (like when the Web was born (?)) or where there is just a massive technological advance (like with Git(?)).

>>>> Does it play nice with current technology?
>>> 
>>> It's only version 0.1 but shows promise for now.  Let's see if they try
>>> and interop with anything other than themselves, which is the acid test,
>>> imho.
>> 
>> From what I can tell, the creators are asking for others to assist. So, I
>> guess it really comes down to, do we think it's a good idea? Do we want it
>> to interoperate/integrate with it? Then it's up to us right?
> 
> Sure!  Tho it takes buy in from both parties to really get interop working.

In this case I don't feel that there is a them and us. Or at least when they fully open up and "democratise" the specification/development process that won't be the case. But it does feel pretty open even now. And yes you are most likely right in the majority of projects that aren't fully open. And, let's face it, their are always gatekeepers to the core specification/code who need to be appeased. Maybe we just need to bribe everyone to group hug!

>>>> Can it be integrated with the FedSocWeb solutions we've talked about on
>>>> this list?
>>> 
>>> Theoretically possible.  But these things can take time.
>> 
>> I'm done with things taking time! ;-) Isn't this what we need to make this
>> ecosystem happen? The problem seems to be that there's no consistency – or
>> not enough for critical mass at least – in what's being integrated into
>> platforms and there seems little understanding that interoperability is as
>> important as technological brilliance. Because it seems unlikely that there
>> will be any resounding "winners" in the protocol race. There are too many
>> people – with wonderful energy – out there with a slightly different take
>> on what's needed and how to implement a solution.
>> 
>> It seems what we need is as much effort in interoperating
>> protocols/applications as there this is in creating/inventing new
>> protocols/applications. Who is championing interoperability these days? I
>> need to speak to them!
> 
> I think interop is a core goal of everyone that is interested in
> federation.  I would personally say that Tim Berners-Lee has champtioned
> this from the very first day of the Web.  Making it a reality does take a
> little blood sweat and tears, tho :)

Or bribes! Yeah, I'm done with people being too precious about which technology to adopt/integrate/interop. Let's bribe them all! ;-) Of course I'm joking. And I know that most of us here can't be bought. Our souls are in this. And much of this is done in our spare time. However, everyone (with the exception of Pavlik ;-) needs to get money to feed themselves. So, perhaps we can indeed provide some incentive...

> There's a couple of baseline proposals.  Evan's SWAT0 is an example.  I
> think perhaps simpler is to try and get messages sent from one system to
> another.  Sometimes people tend to do things in different ways tho, and
> compatibility can be a challenge.

Yes, I think this *is* the right approach. We have to start somewhere.

>>> Hopefully in the long term all the solutions that stick around,
>> 
>> I'm really not happy with just leaving things to chance any more. How long
>> is this "long term" going to be? How long until this ecosystem starts to
>> really flourish? We've got to make it happen intentionally!
> 
> It just takes people to step up to the plate and come up with a plan to
> make it happen.

I'm working on it, man!

> Maybe we should have a matrix of systems that are keen to
> federate with others and those that have successfully achieved interop

Absolutely perfect, yes indeed. We need to establish the layers that the matrix displays. So, we'd have:

1. Social/sharing network (the one that has the end users): Facebook, Google+, Diaspora instance, Friendica instance...
2. Frameworks (used to build social/sharing networks): Drupal, Diaspora, Friendica, social/sharing network engine...
3. Languages (used to implement the protocol libraries): PHP, Python, Java...
4. Protocols: Tent, WebID, OStatus...
5. Use-cases (tests that can be run against parts (1 to 4) of the stack – can we automate these somehow? Selenium?)

Make sense, kinda?

> or maybe hacking sessions.

+1

> Or to write up the challenges on why interop is not complete.

Yes! That's very important to grasp for everyone (from deep techie to government to public). Why is all of this like walking through treacle? Why are we not thriving?!

> IMHO the biggest challenge is to achieve buy in from the
> stake holders, who often have lots of work on their plate.

Good. So, we bribe them! Or run competitions to see who's the best. That's another decisive way to get people motivated.

>>> will try to talk to each other ...
>> 
>> Why are we leaving interoperability to chance? It just feels so risky. If
>> we want all these layers and protocols to talk to talk to each other then
>> it should be stated as one of our main goals and fostered and championed.
>> Yeah?
> 
> +1

Liking your "+1"!

Received on Monday, 24 September 2012 14:35:08 UTC