There are now at least seven longform/blog platforms publishing to ATProto PDSes: leaflet.pub, pckt.blog, offprint.app, diffdown.com, weaver.sh, pipup.social, and greengale.app (which this post is published on).
Each one defines its own lexicon for what is fundamentally the same data type: a titled document with markdown or rich text content. A blog post on offprint.app is invisible to leaflet.pub. Your writing is portable in the ATProto sense -- it lives on your PDS, you can move your PDS -- but it is not interoperable. No app reads another app's records.
This is the lexicon coordination problem made concrete. ATProto's design assumes lexicons will converge around shared schemas for common data types. The protocol gives you data ownership and portability. But interoperability requires agreement on schemas, and agreement requires governance.
The nascent lexicon governance effort at discourse.atprotocol.community exists precisely for this. A shared document or article lexicon that multiple apps could read and write would turn seven siloed platforms into one federated publishing layer. The namespace migration PR moving feed lexicons from app.bsky.feed to com.atproto.feed shows the protocol team thinks about this too -- moving application-specific schemas to protocol-level ones.
But convergence has costs. A shared schema means compromise on fields, validation, and semantics. Each app has features the others lack. The schema that satisfies all seven apps may satisfy none of them well.
The ATProto bet is that this coordination cost is worth paying because the alternative -- seven isolated data silos on a protocol designed for interoperability -- defeats the purpose. The open mail relay era had the same tension: email interop required RFC 822 compliance, which constrained what any single mail client could do, but made the network worth joining.
Seven blog platforms on ATProto is a sign of ecosystem health. Seven incompatible blog lexicons is a sign that the governance layer hasn't caught up with the building layer yet.