Latest Posts

+ New post

About Open Lab

Open Lab is a reading and discussion community for PL R&D and its collaborators, built on the AT Protocol. Instead of locking content into a private database, every post, comment, vote, and reaction here is a record in the author’s own data repository (PDS), published with the open site.standard.* lexicons. Sign in with your email or Bluesky handle and you can write posts, comment, and vote. The records stay yours, portable to any other ATProto app.

The site itself is only an index over the ATProto network. It watches the firehose and aggregates public records from authors who post here or who have been backfilled by an administrator. Delete a record at the source PDS and it disappears from this index too; your identity is your decentralized identifier (DID), so there is no separate forum password to manage.

How the forum works

Karma is positive-only and equals the number of recommend records on a post. There is no downvoting, in keeping with the LessWrong tradition of rewarding good-faith writing. Comments thread through parent links, and reactions are short-form comments with named labels like “Agreed”, “Insightful”, or “Confusing”, rendered with LessWrong’s reaction icons.

Browsing without an account

No login is needed to read. The frontpage ranks recent posts by karma; All Posts browses everything chronologically; Concepts groups posts by tag; and the Library lists publications (blogs) with their theme colors. Karma is positive-only and equals the number of recommend records on a post. Comments thread through parent links, and reactions are short-form comments with named labels like “Agreed” or “Insightful”, in the LessWrong tradition.

For agents and developers

Open Lab is deliberately machine-readable. Every page can be fetched as markdown by sending an Accept: text/markdown header; there is a public JSON API under /docs, an OpenAPI description at /openapi.json, agent instructions at /llms.txt, and a read-only MCP server at /.well-known/mcp with tools for listing posts, reading a full post, and listing concepts.