For example, not many people used Tim Berner-Lee's original web browser but it did help define the standards that then went on to be picked up by Mosaic which was the first really popular web browser. However, by the point that Mosaic came out most of the features of the architecture of the web were in place and remain recognizable to this day.

Surely the goal is to provide common lower levels of a "social" protocol stack and allow as many and different user interface applications to layer on that common core?

That's one of the goals, but it's not the only goal. I think what you're bringing up is what makes social networking somewhat new terrain for open source. So far, open source has been able to dominate the server space because of superior tech implementations for an insanely low price point.

For social applications, that doesn't matter as much when you're trying to draw users. You need a usable UI, you need to learn all the lessons that the proprietary systems have learned over the years. It's not good enough to just have a better technology, you have to present it as well or better than the competition.

I think it's a new trajectory, because there's less incentive for walled gardens to open up than there was for Netscape to use http. So I think this is a necessary trail for open source to blaze through.

So in a lot of ways, there's very much a chicken-or-the-egg dilemma: Usable protocol stack that others can build on, or sizable enough user base that developers want to build on it. And the closest I've seen to open source dealing with these questions was actually, the gnu/linux desktop projects.

It would be interesting if there was an equivalent for social applications... a common server component, an application library and maybe a reference application but with plenty of scope for people to innovate in the UI without the need to continually reinvent the core communication components if they don't want to.