← All methodologies

dmn — declared by cat-harness — source held

Applies when. The criteria RECUR and the inputs are data. A gateway that must branch the same way on the same facts every time. Not for a one-off judgement — that is kepner-tregoe, recorded per madr.

Origin. Object Management Group, Decision Model and Notation (DMN) Version 1.5, OMG document formal/24-01-01, January 2024 (omg.org/spec/DMN). RECORDED, not held: its licence forbids posting copies on a network, so the library entry identifies the exact PDF (sha256) and its outline and holds no text.

Ingested sources:

  • library/omg-2024-dmn-1-5

Node file: dmn.md

DMN — the computable case

Present in this repository since before this adoption, and never written down as a methodology. Five .dmn tables exist under processes/decisions/, the BPMN engine evaluates them, and dmn-authoring sits on the business-analyst role. What was missing is the statement of when a decision should be a table at all — which is what made it possible to reach for DMN by habit, or to miss it.

The test, in one line

Would the same inputs have to produce the same branch next month?

Yes → a table. No → a judgement, and a table would lie about its own repeatability.

What this platform already enforces

A hand-supplied outcome is refused. A DMN-backed gateway computes its branch from the table; an agent cannot assert which way it went. That is a stronger guarantee than the methodology itself asks for, and it is the reason a table is worth the ceremony: the decision cannot be quietly overridden at the call site.

See bpmn-processes for the gateway contract and dmn-authoring for authoring the table.

Hit policy is a modelling decision, not a detail

UNIQUE says the rules cannot overlap and the engine may check it. FIRST says they may overlap and order decides. COLLECT returns every match.

Prefer UNIQUE, because it is the only policy under which an overlapping pair of rules is an error rather than a silent precedence. FIRST makes rule order load-bearing, and a reader of the table cannot see that the order was deliberate.

Where DMN and the other methodologies meet

  • A kepner-tregoe analysis whose criteria turn out to recur becomes a table. That promotion is the good case, and the MADR record of the original decision is what explains why the table has the columns it has.
  • A table can encode which methodology applies — a decision about decisions. The repo already does this shape in folio-intent.dmn.
  • GRADE is not a DMN table. Certainty of evidence is a judgement over a body of evidence; tabling it would assert a repeatability the method explicitly denies.

Refusals

  • Never table a one-off to make it look rigorous. A table asserts the decision recurs; if it does not, the assertion is false and the next author will trust it.
  • Never hand-supply a gateway outcome. Already enforced; stated here so the reason travels with the methodology.
  • A table with one rule is not a decision model. It is a constant, and should be written as one.