29: State of the Mint and Mind w/ Calle — No Solutions

"You cannot look at something without touching it."

Calle & Gigi take a walk.

Calle and Gigi start with Cashu mint internals and end in embodied cognition. Along the way they unpack sparse Merkle trees, exclusion proofs, unilateral exit, trusted execution environments, recovery paths, the move to BLS signatures, agentic coding workflows, specialist versus generalist models, world models, and why Bitcoin, Nostr, and local AI still feel like reasons for optimism.

In this dialogue:

how mint state can be compressed into a sparse Merkle tree and why exclusion proofs matter for ecash
how unilateral exit could reduce both the slow rug pull and the fast rug pull for Cashu mints
why trusted execution environments change the trust model, and why on-chain-first mint backing is the easiest place to start
how timelocks and third-party recovery paths might soften mint downtime
why Cashu is moving away from its current signature setup toward BLS signatures and what that unlocks
how Calle actually works with coding agents, from OpenRouter and OpenCode to fresh contexts and ruthless resets
why context windows are scarce, token maxing is wasteful, and one agent per project beats one giant assistant
why cognitive science still matters when talking about specialists, generalists, and what frontier labs are betting on
why world models differ from next-token predictors and why Vervaeke on AI and embodiment keeps coming back
why Thomas Nagel and The Blind Spot show up once the conversation turns to observation, physics, and experience
why high-agency builders should feel more hopeful than scared right now
why Bitcoin, Nostr, and improving local models still look like the right ingredients for a freer future

Related dialogues:

People mentioned:

Projects & tech mentioned:

Readings mentioned:

Recorded at 952,654.