There has been a lot going on in protocol design land around non-public data and community groups this year. The permissioned data proposal was published back in July, an atmospheric groups proposal forwent out in August, and the atproto spaces alpha PDS has been running successfully for a couple weeks. With northern hemisphere summer coming to a close, and long carefree sunny days behind us, folks might reasonably wonder: just when exactly will all this be ready to launch for real apps and millions of real human accounts?
Speculating about timelines is always a bit dicey, especially in an ecosystem with multiple independent organizations. But we (the Bluesky protocol team) think it would be helpful to give our best guess of what will unfold over the coming months. This covers both the permissioned data spaces functionality in atproto itself, as well as the interoperable community group system that will layer on top.
September: this month we should be kicking the tires of the current design and surfacing any major issues. This is a period of “de-risking” and feeling out the shape and limitations of the overall design. This is the last window to debate and come to rough consensus on that architecture. We should prepare and discuss medium-size changes and “amendments” to the proposals. We can make clarifying edits and small changes to the proposal text, don’t intend to churn too much.
October: based on what we learn, we should do an iteration on the proposals, resulting in something closer to a draft specification. This iteration could roll up multiple breaking changes in a bigger batch. In this month we should do a round of polish and hardening, turning up ambiguities, abuse vectors, etc. Also work on developer resources and reference implementations.
November: coordinate around an informal launch date when the protocol functionality becomes “generally available”: reviewed specifications are published, stable SDKs are released, reference PDS update rolls out, etc.
December: respond to any early operational issues, and polish developer resources and tooling.
I would ask that application developers and service providers not publicly launch production features that rely on spaces data, services, or lexicons until the protocol is “general availability”. Folks can absolutely start designing, prototyping, and experimenting. And I think it is safe to publicly share plans, demos, and do user testing as things start stabilizing in October. But in the interest of interoperability I think it would be best if folks don’t try to “jump the gun”. I think that the easiest way to collectively coordinate that is for the major PDS operators to agree on a “general availability” date.
What are your thoughts, questions, observations?