Google's Open Social: The missing pieces?

Like many people, I'm having a hard time seeing how the Google OpenSocial announcement isn't missing some essential pieces. While I have inferred the bigger architecture from the OpenSocial API site, and partner sites, I find it disconcerting that the architectural big picture is essentially implied primarily by the Google Gadgets architecture. 

The following picture seems to sum up my understanding of the current state of affairs with respect to what was announced:

The OpenSocial API is a set of Javascript and REST interfaces to provide access a site's social network data/graph. But the blogosphere and marketing buzz has around the concept of a "write once, run anywhere"-style offering. Digging deeper, it seems that the real story is that only when social network sites (OpenSocial "containers")  implement OpenSocial AND the Google Gadgets API/architecture do they provide the application portability dream. I'm not seeing this stated explicitly anywhere, and it's leading to a good deal of confusion (and not a little bit of hostility to boot).

I would like someone to confirm or clarify my understanding. What exactly is Google's vision and promise for the OpenSocial API? As a social app developer, is this really going to solve the majority of my cross-platform (social network sites) issue? As a social network provider, should I plan on implementing the Google Gadget API + OpenSocial?

Google, please help! The messaging here is really really confusing.

Comments

[Julian Bond]

I see this as being 3 pieces.

1) Gadget Writers can write gadgets for Orkut and they'll run on any social network that supports the openSocial specs. Hi5, Plaxo, Ning are already close. However, each SN Container will make it's own rules about what gadgets can run.

2) Social Networks that support the specs don't have to reinvent Facebook's API in order to deliver similar functionality. They can build a relatively rich profile page, relatively quickly and take advantage of a large ecosystem of gadgets by being a Container.

3) Social Networks and other websites can support the Data APIs. This should encourage the sort of ecosystem of Applications (web and desktop) and Mashups that has grown up around Twitter. But instead of being restricted to supporting one site, they should be able to work with any ste that implements the data APIs.

The problem today (7-Nov-07) is that:-
- The Orkut Sandbox is the only testing environment and it's flaky
- There are no examples of the data APIs anywhere.
- The specs for the data APIs are minimal and have some serious holes.
- The specs for the Container side haven't been released. And work on this is currently restricted to Google's NDA partners.

It seems that way too much of the early analysis has focussed on data portability between Social Networks. This is something that is only hinted at by the Data APIs. And at least for a while, the write-once, run-anywhere gadgets will be what is actually built. Except that they will actually be write-once, run-on-Orkut until there are other reliable Containers out there.