Computable Mind: A Field Theory of Capability by Rogério Figurelli
English | October 14, 2025 | ISBN: N/A | ASIN: B0FW5NPQ6S | 153 pages | EPUB | 1.26 Mb
Too many "intelligent" systems persuade before they prove. They shine in demos, drift in production, and leave no receipt when it matters.
This book offers a different stance: treat capability as a field - distributed over space and time - with gradients that guide movement, curvature that signals risk, and invariants you refuse to trade. Tie every claim to a receipt in the same unit as the commitment, and let small ON/OFF gates decide change without theater.
Built on the Science of Intention, Large Language Fields (LLFs), and Field-Driven Design (FDD), the book turns governance into conservation. It introduces a minimal field vocabulary (invariants, gradients, local curvature, κ-gates), operational patterns that favor decisions over dashboards, and standards for receipts that make claims replayable by third parties. You'll learn a lightweight calculus of change cost, evaluation that teaches rather than dazzles, and healthy couplings between LLF and FDD so agents cannot rewrite values behind the scenes. The rhythm is prosaic: problem → claim → receipt → decision → memory - small moves you can ship next week.
You will learn how to:
* Name invariants with explicit units, tolerances, and ON/OFF gates.
* Attach one-page receipts that test claims in their native unit.
* Map gradients and curvature to predict stress before incidents hit.
* Compare fairly on equal budgets and prefer the smallest sufficient models.
* Run quiet SLAs: gates, evidence ledgers, and routine rollbacks.
* Govern retrieval, privacy, and consent as policy without freezing progress.
Who it's for: engineering and product leaders, operators, ML/AI teams, risk and compliance, and anyone measured by outcomes - not volume.
The promise is modest and durable: keep what works - ingenuity, taste, iteration - and bind it to receipts at the same scale as the commitment.
With LLF, you stabilize intention; with FDD, you conserve outcome; with a field view you can measure, steer, and preserve capability - even as the world moves.
Links are Interchangeable - Single Extraction
