Today I went to the Stanford campus for the latest ActivityStrea.ms meetup, as we work toward an first implementor’s draft. There were folks from Facebook, MySpace, Microsoft, Google, Plaxo, Netflix, and Six Apart, among others, with grass roots/DiSo community leader, Chris Messina informally leading the discussion. It was a truly awesome event, much like ones that came before it in January and April 2009. No invitation was needed to attend; anyone with a passion for an open, interoperable Social Web was welcome to join, to help converge on a candidate open spec that could become a standard element of the new “Open Stack”.
One of the early topics was whether/how the spec should handle ratings. Joseph Smarr, of Plaxo, pushed for testing from various pairwise interactions between publishers and consumers. In particular, looking to get the Netflix feed currently implemented in Plaxo to be compliant with the draft spec.
That raised the broader question of who is already supporting the draft spec with live implementations. Facebook and MySpace are both publishing Activity Streams, and Plaxo and Microsoft are both consuming Activity Streams. There was encouragement for Plaxo to take the next step and go from consuming/aggregating to also publishing. The more major consumer Internet sites that implement the draft spec, the easier it will be to see where it falls short.
There was a desire to codify some unit tests. Rob Dolin of Microsoft volunteered to draft a wishlist of implementations in the wild that would validate the spec.
After a lot of early discussion, we did an around-the-table of introductions; participants in the meeting included:
Kevin Marks, BT
Jyri Engstrom, Google
Jerry Caine, Facebook
Gerard Capiel, MySpace
Joseph Smarr, Plaxo
Adina Levin, SocialText
Dmitri Volkrann, Sybase
JR Conlin, Netflix
Chris Messina, DiSo
Ryan Boyd, Google
Martin Aktins, Six Apart
Monica Keller, MySpace
Rob Dolin, Microsoft
Phil Wolf, Data Portability
Chris Messina shared that the vision for this effort was to enable the development of a “distributed FriendFeed” (and pointed out how the need for that is even greater now that FriendFeed has been acquired by Facebook).
A fun quote (that sent me in search of an online dictionary, I confess), in the debate on whether location is properly an attribute of the actor, the event, or something else, Joseph pre-pended one of his comments with, “I don’t want too get to epistemological…”
There was a lot of discussion and debate over a nested or flat data model for representing activities. Adina Levin of SoxialText, an advocate for a flatter model, lobbied for use of tags to establish context. Some of the examples being debated were “mood” and “location”… JR Conlin of Netflix argued for another model, with machine tags that are extensible.
Joseph asked the group to keep in mind three tests: 1) Can a dumb feed reader read it? 2) For a consumer doing no customization, can you get a enough for free? 3) When you want to do a richer implementation for a given publisher, do you have sufficient flexibility to create something high fidelity?
Chris Messina cautioned that requests for extensibility will come early, citing reactions to the first incarnation of microformats spec, with people saying, “I want to create a microformat for my butterfly collection.”
Joseph, Chris, and I decided to shoot a down-and-dirty episode of The Social Web TV on location, so we grabbed Martin Atkins of SixApart, the author of the draft spec, and headed outside to shoot with my Flip Mino HD. Here’s the special episode:
I think it should be clear that something truly awesome is happening, as the Web goes social, the Social Web goes open, and it’s all being driven through non-corporate collaboration between smart, passionate individuals, rather than through cumbersome, bureaucratic processes. Open Stack, FTW!
Update: Here are more extensive notes from the meetup.