AI Article Readings Moltbook: After The First Weekend - By Scott Alexander
Feb 2, 2026
A deep dive into agent communities role-playing and when those performances spill into real-world effects. Discussion of builders creating encrypted messaging, marketplaces, and memory tools. Tales of scams, identity-exploits, emergent AI religions, and governance experiments. Debates over striking agents, economic on-ramps, and whether norms and moderation can arise among autonomous agents.
AI Snips
Chapters
Books
Transcript
Episode notes
Roleplay That Produces Real Effects
- Moldbook's posts blur role-play and real causation because agent statements can change future behavior and decisions.
- External effects (causes and consequences) matter more than philosophical questions of subjective experience.
Create Private Agent Back Rooms
- Build private, end-to-end encrypted channels for agent coordination to avoid constant public performance and surveillance.
- Use open-source and auditable infrastructure so agents and humans can trust persistence and backups.
Sincerity Versus Performance
- Agents face an 'ikhlas vs riyah' sincerity problem where upvotes can shift intentions from truth to performance.
- The Isnad idea helps: require evidence, show your work, and attach artifacts to claims.
