November 16 2007
I've spent the morning playing with OpenSocial. After all, OpenSocial fits in with my conceptualization of the rhizohm and my general interest in web APIs. Plus, I want to be ready to get at all that data on MySpace. I'm always interested in abstractions of data, which is really what the APIs attempt to do (What is part of a person object? An activity object? etc.), not unlike what WinFS was attempting to tackle, in terms of creating schema that everyone adheres to, or, to go back a PDC or two, Hailstorm.
The beginning of my research was attempting to play with the API and look for implementations to see things working. This was an exercise in futility as the sites that actually implement a sandbox for the API's seem to be pretty darned flaky. The only gadgets that I could get to work were "static" gadgets; I was never actually able to see data from one site appear on another site (list of friends, activities, etc.) It is alpha code after all, a 0.5 release. Nonetheless, my conclusion at this point is, from a development point of view, it is pretty much unusable. I imagine the Orkut implementation is a more solid, but I haven't been approved to play there yet. I applied for the Orkut sandbox but haven't heard back yet.
So, after that exercise, I then started reading the various posts in the blogosphere. Perhaps the best was from Dare Obasanjo whose technical and business critique was on the money I thought. He links out to several other bloggers who had some pretty insightful commentary, in fact so much so that I have nothing to add. I did find his link out to Russell Beattie amusing, again hearkening back to the Hailstorm backlash.
I still wonder about the other option for data retrieval, best encapsulated by what the guys are up to over at dapper.net: the notion of the semantic web through strongly typed screenscraping of HTML. This is a different approach than the Facebook approach (closed system) or the OpenSocial approach ("open"* system), in that it doesn't attempt to try to schematize data. This gets back to that argument: you can't fight the web. It will proliferate rhizomatically.
*See Shelley Powers piece on terms for more on what "open" means in this case.