LINEAGE
Where this fits
Every part of this machine has ancestors in established traditions. This page names them, and points to where each idea does its work in the system.
Framing
This lineage was not a research path. The system took shape from practical work in decision systems, and the traditions below were found afterward, as an attempt to understand what had been built and where it sits among ideas that came before it.
Decision doctrine
Military decision-making is a discipline of structured judgment under uncertainty. Commander's critical information requirements declare, before contact, what information would change the decision: the question stated in advance of the evidence. The decision support matrix tracks typed conditions against declared criteria as an operation unfolds. The after action review turns what happened into structured institutional memory, so the organization learns from a record rather than a recollection. ADS shares this discipline's shape: intent declared first, conditions typed, criteria set before evidence, every act kept as a record to learn from. It runs through the whole machine.
WHERE IT LIVES - The whole machine, on the architecture page
Neural and symbolic
Neural systems, the language models, reason in natural language but keep nothing checkable. Symbolic systems, the databases and ontologies, are checkable but cannot generate. Combined at the data level, what survives is data about reasoning: the reasoning itself happened inside the model and is gone. SDI bridges the two with a language, then builds a computer around the language. The home page makes this argument in full; the grammar section of the architecture page shows the bridge working.
WHERE IT LIVES - The grammar, on the architecture page
The processor
Vaswani and colleagues, Attention Is All You Need, 2017. The transformer is the first processor that reads meaning, and the reason ADS can exist: a calculus for judgment was never executable before there was a processor that could read the natural-language half and write the algebraic half as one text. The model section of the architecture page shows how the machine runs it: stateless, cold-called, replaceable.
WHERE IT LIVES - The model, on the architecture page
The theory of meaning
Charles Sanders Peirce, the triadic sign: Sign, Object, Interpretant. Shannon made information measurable by deliberately setting meaning aside; Peirce supplies the structure of the layer Shannon left open. Information systems are dyadic, a symbol mapped to a referent, and the interpretant, the reading that relates sign to object, never survives as an artifact anything can check. Every act here must close the triad, and the closure is checked before commit. The claim is about the record, not the mind: the interpretant here is a committed, sealed, checkable artifact. The record section of the architecture page is where it lives.
WHERE IT LIVES - The record, on the architecture page
The control principle
W. Ross Ashby, the Law of Requisite Variety, 1956: a control system must have at least as much variety as the system it governs. The gate does not match the model's variety by judging content, which would make it a second reasoner needing governance in turn. The schema absorbs the variety instead: every act must resolve into the same typed structure, and the gate computes over the structure. The gate section of the architecture page shows the arithmetic.
WHERE IT LIVES - The gate, on the architecture page
The measurement tradition
Process-tracing in decision research: Payne, Bettman, and Johnson, The Adaptive Decision Maker, 1993. A decision process can be measured on its structure and effort, independent of whether its conclusion is correct. Jc applies that per reasoning act, and every score is sealed with the act and recomputable from it. Measuring cognitive work is not new; sealing the measurement into the record it measures is the addition. The metrics section of the standards page carries it.
WHERE IT LIVES - Metrics, on the verify page
The owned record
Two strands that never met. Intellectual capital, Edvinsson and Malone, 1997: codified knowledge is an asset in the ordinary sense, built at cost, transferable, valuable for the accumulated work. Cryptographic self-verification, Haber and Stornetta, 1991: a record can prove its own integrity without a trusted institution. The first assumed a custodian; the second was not about assets. The chain is the convergence: a knowledge asset that carries its own proof, owned by its operator, checkable by anyone holding it. The record section of the architecture page, and the transparency page, are where it lives.
WHERE IT LIVES - The record and the transparency page
Each entry above ends where the idea does its work, because that is the test of a lineage claim: not whether the ancestor is named, but whether the mechanism honors it. The machine is on the architecture page. The proof is on verify.