Decision Method Triage Matrix¶
Checklist — instantiates Heuristic vs. Algorithm Tradeoff and Selection
Scores or classifies decisions by stakes, urgency, reversibility, uncertainty, data quality, and accountability need.
A Decision Method Triage Matrix is the up-front sorter that decides how a case will be decided before anyone decides what the answer is. It scores each incoming decision on a fixed set of context axes — stakes, urgency, reversibility, uncertainty, data quality, and accountability need — and reads off, from the cell those scores land in, which method class the case deserves: a fast rule, a checklist, an approximate model, a full algorithm, expert review, or a staged hybrid. The one idea that makes it this mechanism is that it is keyed on the decision, not on any particular tool: it never asks "is this shortcut trustworthy?" — it asks "given what kind of decision this is, which lane should it go down?" The matrix is a router, and its output is a method assignment, not a verdict.
Example¶
A property-and-casualty insurer routes every incoming claim through a triage grid at intake. Each claim is scored on a handful of axes: dollar exposure (stakes), whether the claimant is waiting on a payout to make repairs (urgency), whether an overpayment can be clawed back (reversibility), how complete the submitted documentation is (data quality), and whether the claim touches litigation or a regulator (accountability need). The grid maps score patterns to lanes: a small, well-documented, fully-recoverable auto-glass claim lands in the "straight-through auto-adjudication" cell; a mid-size claim with thin documentation lands in "adjuster with checklist"; a large bodily-injury claim or anything flagged for litigation lands in "senior adjuster plus independent review." The result is that roughly two-thirds of claims are routed to a fast lane, while the rare, severe, contestable ones are surfaced by rule into slower, more accountable handling — instead of by whoever happened to open the file.
How it works¶
The matrix is a lookup grid, not a calculation. It works by (1) fixing a small set of context axes that actually change the right method; (2) defining crisp score bands on each axis so a case can be placed without argument; (3) mapping each cell — or each threshold crossing — to a named method class drawn from the option set; and (4) making the assignment the default that stands unless someone formally departs from it. Its distinctive move is decoupling the routing from the deciding: the same grid governs thousands of heterogeneous cases, so consistency comes from the classification, not from the individual decider's mood or seniority.
Tuning parameters¶
- Axis count and granularity — more axes and finer bands capture more nuance but make the grid slower to apply and easier to game; keep only axes that change the recommended lane.
- Cell-to-method mapping aggressiveness — how readily a cell routes to a heavier lane; aggressive mapping catches more edge cases but clogs the slow lanes.
- Tie-break default — whether ambiguous cases fall toward the faster or the more cautious lane; bias toward caution where errors are irreversible.
- Re-scoring ownership — who may re-place a case and how often; loose ownership invites drift, tight ownership invites bottlenecks.
When it helps, and when it misleads¶
Its strength is that it turns method choice from a habit into a visible, consistent rule: the same kind of case gets the same kind of treatment, and the reasons are auditable. It is also the layer that keeps rare severe cases from being quietly handled fast because a queue was long.
Its characteristic failure is bin-gaming: once actors learn which cell routes to the cheap lane, they shade the inputs to land there — a live instance of Goodhart's law,[n1] where a classification used for routing stops measuring what it was meant to. A related failure is a grid that looks precise but rests on axes that don't actually change the right method, lending false confidence to arbitrary lanes. The guarding discipline is to audit-sample cases from each cell against realized outcomes and re-derive the axes from what actually separated good from bad routing, rather than freezing the first grid.
How it implements the components¶
decision_context_profile— the scoring axes are the context profile, made operational as the grid's rows and columns.method_option_set— each cell points at a named method from the menu; the matrix is where the option set gets exercised as a selection.accountability_and_explainability_requirement— accountability need is one of the scoring axes, so high-accountability cases route by rule into reviewable, appealable lanes.
It does NOT validate whether a chosen shortcut still holds — that is the Heuristic Boundary Checklist, which implements heuristic_validity_boundary; the matrix selects among methods but never interrogates one heuristic's envelope. It also does not visualize the speed-vs-rigor tradeoff numerically — that is the Stakes–Latency–Error Scorecard.
Related¶
- Instantiates: Heuristic vs. Algorithm Tradeoff and Selection — the matrix is the archetype's front door, assigning a method class to every case.
- Consumes: Stakes–Latency–Error Scorecard — the scorecard's tradeoff analysis is what justifies where the grid's lane boundaries are drawn.
- Sibling mechanisms: Stakes–Latency–Error Scorecard · Heuristic Boundary Checklist · Algorithmic Escalation Protocol · Shadow-Mode Method Comparison · Override and Exception Log · Retrospective Error Calibration Review · Model or Rule Card
Editorial Notes¶
Form Classification¶
Form family: Decision, Gate & Allocation
Rationale: Decision Method Triage Matrix operates as a case-specific gate, selection, routing, prioritization, or resource disposition because it scores or classifies decisions by stakes, urgency, reversibility, uncertainty, data quality, and accountability need.
Independent corroboration: The frozen evidence defines Decision Method Triage Matrix as 'Scores or classifies decisions by stakes, urgency, reversibility, uncertainty, data quality, and accountability need', so its operative form is Decision, Gate & Allocation.
Nearest alternative: Representation, Specification & Plan — The grid makes a bounded method selection for the case; fixed axes and bands are the reference artifact used to decide.
Review outcome: Independent reviewer agreement; medium confidence.
Origin Attribution¶
Primary origin: Organizational & Management Science
Origin pattern: Cross-disciplinary synthesis
Present-day reach: Multi-domain
Rationale: Management decision design cohered routing choices to different levels of analysis or review according to stakes, urgency, reversibility, uncertainty, and accountability.
Related originating lineages:
- Computer Science & Software Engineering — Software and automation governance supplied the distinction among heuristic, algorithmic, expert, and hybrid lanes.
- Operations Research — Decision analysis supplied structured scoring of context attributes before selecting a solution method.
Review resolution: Management decision design cohered routing choices to different levels of analysis or review according to stakes, urgency, reversibility, uncertainty, and accountability.
Encyclopedia synthesis: The exact catalogued form synthesizes established practice rather than reproducing a single standard historical label.
Review outcome: Reconciled after independent review; high confidence.
Notes¶
[n1] Goodhart's law — "when a measure becomes a target, it ceases to be a good measure." A triage axis used to route cases into cheaper or more expensive lanes becomes something actors optimize against, which is why matrices need audit sampling to detect bin-gaming. ↩