Topics
The starting problem
Why a better representation can make an organization less connected to reality.
The working model
What a program is, and how enough of it can become machine-legible.
- Commitment Networks 5 pieces Programs as linked commitments about outcomes, timing, conditions, evidence, authority, and uncertainty rather than collections of tasks.
- Program as Code 4 pieces The implementation philosophy: explicit, typed, machine-readable, versionable requirements, commitments, policies, authority, and state.
The proposed architecture
How program state could be sensed, reconciled, governed, and acted against.
- Program Observability 6 pieces Turning evidence into claims and reconciled state while preserving provenance, freshness, authority, contradiction, and uncertainty.
- Program Operating System 5 pieces A provisional architecture connecting commitment networks, observability, runtime, control, execution actors, and feedback loops.
Boundaries and participants
Where authority sits, how agents participate, and where the physical world resists the model.
- Human Authority & Judgment 4 pieces Ambiguity, accountability, incentives, conflict, and deciding when the system itself is wrong.
- Agentic Execution 5 pieces Agents as bounded participants inside a larger execution architecture: what they can carry, where they break, and how autonomy is earned.
- Infrastructure in development Programs where hardware, capacity, and physical constraints mean a mistake cannot be fixed by editing a slide.