Seven Primitives
Chapter 1 of Loop Engineering argues that four traditions converged on the same seven primitives over a 36-month window: coding-agent practitioners, knowledge-work pedagogy, ML reasoning-model training, and autonomous-agency builders. Convergence across incompatible starting axioms is the book's central evidence pattern.
This section will host the full mapping. One sub-page per primitive with concrete examples from each tradition, the academic lineage that produced it, and the failure modes when the primitive is missing.
The seven primitives
- Verifiable stop — every loop iteration ends with a check that passes or fails. No silent successes.
- Maker/checker — the producer and the verifier are different actors.
- Externalized memory — state lives outside the model's working context.
- Materiality gating — effort and autonomy scale with stakes.
- Frame before solve — restate the problem before attempting the solution.
- Visible verification — verification leaves an audit trail.
- Debt accumulation — failed attempts and abandoned approaches are recorded.
Per-primitive sub-pages ship as Tier 2 content. The print book carries the seven-primitives table; this site will carry the extended treatment.