The Architecture of Sociality: Building In Openness

A lot of discussion boiling recently about openness in social applications (like the Bill Of Rights movement manifesto and supporting comments, and Brad Fitzpatrick's Thoughts On The Social Graph.

I offer a thought about deductive openness, by which I mean a path of low resistance for developers of application. Rather than striving for a theoretical openness, based on high-minded principles, I believe that developers will likely taking tactical, well-understood and intuitive paths toward adopting common services.

Centralization

Centralization is bad, because it is closed. However, users can often be the happiest in monopolistic worlds, where everything works together because one corporation has centralized control on design and implementation. This is why examples like IBM and Microsoft have arisen: users opt in to hegemony.

In the social application world, the equivalent would be a single centralized application, which satisfies all needs for all users. Let's call this the Facebook model:

Don't get me wrong: Facebook's management have opened up their service to others, so that many many apps have been built and which run within Facebook. So it is not a closed model: but it is a centralized one.

Users have their identity and access controlled by Facebook, and other companies have to operate within the confines of the Facebook architecture.

There is something appealing about what has been going on with Facebook: a renaissance of thousands of new applications emerging, everyone having fun, but... inherently limited to the range of concepts defined in the core Facebook architecture. Sure, they will extend that over time, but ultimately this is AT&T before the break-up. It doesn't matter if they are benign, they are still dictating the sum and span of what can be done in the core social architecture of their platform.

Largely, today, we are living in a world with thousands of centralized social applications, providing very little interoperability aside from the bleed-over of RSS-based aggregation and reposting. Seems like most apps are caught in a struggle for dominance around becoming the hegemon in some market: trying to be the winning central service. This why we are carrying our Twitter feeds into Jaiku and our Jaiku feeds into Twitter: no one really wants to play together. They want to dominate.

Distributed

The scalability and bottom-up nature of the Web is not based on a centralized model like Facebook, which should be one of the most potent abstract arguments against Facebook hegemony. Obviously, a solution that is more like the Web itself is more likely to succeed, right? Well... maybe not. Getting the Web off the ground required the Arpanet guys to support it before it rolled out commercially. We'll see.

But, the notion of distributed functionality, like that embodied in the Jabber protocol is interesting.

Here we see different servers communicating through some shared protocol. Various users can decide with server (product) to use based on what every decision making is relevant to them, but some mechanism for communication with users logged into other servers is supported. This is how Yahoo and Microsoft now support interoperability in their instant messaging solutions today, although through some custom gateway, and no one else is invited to play.

The Jabber protocol is an open one, on the other hand, and suggests some great lessons to be learned for the social application marketplace. We could begin to develop such a standard, so that new applications could come on line, and incorporate intercommunication with other participating servers (products). I, using social app A could share information about myself, my life and workstreams, with other users known to me even if they were using some other server (product). The protocol would have to support some lowest common denominator of functionality -- basically a shared model of social architecture -- but they would be free to add higher-order and specialized services.

But it is likely far too early in the day for this to happen. It took a long time for the various relational database vendors to get around to a standardized notion of SQL, which was at first just the intersection of the various implementations with a lot of weasel-worded extensibility supported. But decades later, SQL really has become a largely standardized basis for application development. A similar process could take place in the social application space. And it is likely to emerge through one simple path.

Interlinked

I think the path to a generalized shared architecture foundation for social applications will emerge through the adoption of smaller, more well-defined common services that will link independent applications together.

Consider RSS as a simple example. Or identity as managed by OpenID.

As a better, more compelling, and as yet unrealized example consider the various approaches to streaming that are emerging in the multiplicity of flow apps. One near-term benefit for developers would be a common mechanism for interconnecting streams across applications, so that users can have accounts in multiple applications and combine their multiple streams in a useful way, and so that information created in one context can be displayed, and perhaps even manipulated in others.

I am betting on this scenario: small networks of applications sharing common services will emerge, and the collaborative benefits of interoperability within a group of interlinked applications will lead to that network becoming more competitive than collections of unintegrated applications. The best models for common services will rapidly emerge, and will become de facto standards. Over time, dozens of these standards might emerge, and finally, an open ecosystem for social applications will exist, without the intervention of a centralized authority or a monopolistically dominant player.

We don't need a Microsoft or Google to build everything and make all the decisions about architecture behind the scenes. We don't need Facebook to benevolently allow us to fool around in a sandbox of their devising. We simply need to accept the need to push toward openness based on short-term, achievable, focussed shared services. Small collections of collaborating companies can do this: it doesn't need to be a global movement. The market will pick the winners through experimentation and Darwinian selection.

The Web 2.0 principle of tightly focused apps that do a small number of things well is great, and works as far as it goes. But when people want to have the results of this collection of small things bleed over into that collection of small things, something more is needed. An architecture for sociality will emerge, based on dozens of focused bridges built between collaboration applications.

The general movement toward abstract openness is great: it provides us a vision of where we want to arrive. But concrete openness will be achieved through small, near-term and tactical approaches which over time can snowball into something much larger.

As you might imagine, I am involved in something along these lines, and I will be providing more details over the coming weeks and months.

Comments

[Colin Carmichael]

With FB's moves to open itself up, it could be that FB itself becomes the common service (dare I say Platform?) upon which other social functionality is built outside the garden?

[Anthony Kuhn]

Stowe:

I always enjoy your insights and opinions. Thanks for cheerleading the need for open platforms in the brave, new world of Web 2.0. I think interoperability will become key to web business success in the not-so-distant future. My target audience enjoys this type of post and topic, so I cross-posted on your piece, along with a few of my own comments, at http://blog.innovators-network.org
The Innovators Network is a non-profit dedicated to bringing technology to startups, small businesses, non-profits, venture capitalists and intellectual property experts. Please visit us and help grown our community!

Best wishes for continued success,

Anthony Kuhn
Innovators Network

[Dennis McDonald]

Thanks for such a clearly thought out piece. It helps me understand why I think that the days of the generalized social networks (e.g., Facebook, Linkedin) might be numbered.

Thats is, IF the interoperability standards and protocols become applianced and available as plug-ins allowing individual server and website operators to practice and advertise their adherence to a particular relationship model, and therefore available to "hook up" with others who also adhere.

That would seem to allow individuals to bypass the need for networks (such as Facebook and Linkedin) to define and operate their own infrastructures; as a result, wouldn't such networks have to look elsewhere to attract eyeballs?