You very quickly get into the MxN problem: M consumer applications, N publishers. Each of them decides “I’m the best at modelling this problem for my particular domain” and does their data modelling, and then you end up with a whole bunch of individual point connections and some really REALLY fugly glue to get similar consumption.

You really need some people to sit down and hammer out what the content of these messages (no matter how they’re delivered) are going to look, and that’s pretty hard, because every system is slightly different. Some people have started to do a reasonable job at this, but I’d say that’s as big a problem as any of the technical challenges of getting the messages from A to B: getting general agreement on what precisely is supposed to go into them in the first place.

Also, Chris, are you imagining something like Yahoo Pipes writ large?

In this context, ’social’ means sharing updates on ’social objects’ by multicast pubsub filtered by roster; now fashionably called ‘assymetric follow’. This is just messaging – so why not seek to reapply successful integration patterns from enterprise messaging, to social applications.

I think it is useful to then ask:

* what social utilities or shared services would enable the messaging for micro-integration between social applications – eg. relays, firehose services, stream query services …
* what standards are needed? http, xmpp, amqp, app, …
* what microcontent adaptors are needed at the endpoints and for transformations
* what is the pingback model (if any) and/or push pubsub model, for defining updates on web content, and subscriptions to ad hoc microcontent feeds

To address the problem Chris describes in his post above, all this needs to allow for (but not mandate) consumer driven feed creation (“tell me about updates to this microcontent given by this URI that only I care about”).

Once again I agree that OS is a clumsy term. I’m not however that convinced by application execution environment in this context, as it stretches my own interpretation of what an ‘application’ is.

My sense of what this would look like if it ever works properly is some point and click plumbing stuff over the top of some pretty fugly scripting. Clearly the average user won’t be creating their own social applications (= writing fugly scripts), they’ll be specifying how they want things to be joined together and configuring what’s important and what’s not (JP’s ‘graphic equaliser’). For that to happen firstly needs some standards for the interfaces, so that one piece of equivalent functionality can be swapped for another. I doubt that the existing service providers will be that inclined to break down the walls around their gardens, so somebody else will probably have to get involved in wrapping and mashing.

This is all starting to remind me of the enterprise service bus concept (not that I’ve ever really seen it work properly), though clearly stuff like OpenAdaptor was useful and successful.

For those of us drilled into a rigorous CS background every time we see “Operating System” used wrongly (see: Cloud OS, Google OS, Social Network OS) we cringe. :-)

I realised from the outset that the operating system label was likely to cause offense. To be honest I don’t like it myself, and it wasn’t even my label as it came up in a workshop with one of our research partners (I won’t call out the guilty party by name). I used it anyway as I haven’t yet thought of a better label for ’something in the browser’ that lets you do all this user side social network combination stuff.

Perhaps if we can agree to David’s ’snopsys’ (even if it does have the broken term operating system buried in there) then we can move on from the symantics (which are always important in these early days) and on to a richer discussion on how this might be made to work (and who will write it)?

While I appreciate that you’re trying to promote something that would be useful, you’re really not talking about an Operating System. An OS is there to manage hardware resources and potentially provide a primary interface to the user. What you’re talking about is services to be consumed to make doing social applications easier.

http://kirkwylie.blogspot.com/2008/12/snos-quit-angering-andrew-tanenbaum.html

Thanks a lot,

Kirk

Hopefully a vanity search subscription will bring Adriana along shortly. The pointer to her stuff quickly led me to http://themineproject.org/, which looks like a good attempt to by a ’snopsys’, though clearly it’s still early days.

I too arrived here via @jobsworth / JP Rangaswami

And, by the way, it makes a lot of sense. Is Adriana Lukas part of this conversation?