Multi-Criteria Selection Rubric¶

Multi-criteria decision rubric — instantiates Opportunity-Gated Adaptive Diversification

A shared scoring frame that makes the criteria for keeping, cutting, or advancing a variant explicit and comparable across every branch.

When many branches are competing for survival, the quiet danger is that they get judged on different axes — one on revenue, one on strategic fit, one on whoever championed it loudest — so the "selection" is really incomparable apples and oranges. Multi-Criteria Selection Rubric removes that ambiguity. It fixes, in advance, the dimensions on which every variant will be scored and the rule that turns those scores into a keep / cut / advance decision. Its defining move is that it supplies the criteria, not the evidence and not the execution: it is the shared yardstick against which variants are measured, deliberately built before the contenders are known so the standard can't be bent to fit a favourite.

Example¶

A city transport authority has run four pilot designs for a newly funded low-emission corridor — a bus-only lane, a bike-and-scooter spine, a shared-mobility hub, a congestion-charge zone — and now has to decide which to scale. Left informal, the choice would collapse into whichever pilot had the most vocal backers. Instead the authority builds a Multi-Criteria Selection Rubric first: five weighted criteria — emissions reduction, cost per rider, equity of access, political durability, time-to-deploy — each with an explicit scale and a stated weight, agreed before any pilot's numbers are plugged in.

Now every pilot is scored on the same axes. The bike-and-scooter spine wins on emissions but scores poorly on equity; the bus lane is middling everywhere but clears the retention threshold on every criterion. The rubric doesn't gather the pilot data — that came from the trials — and it doesn't execute the roll-out. It does one thing: it makes the comparison honest and legible, so the decision can be explained, defended, and revisited when a weight is challenged.

How it works¶

Fix the criteria first. Enumerate the dimensions that define fit for this opportunity — before the contenders are on the table — so the standard is set independently of the field.
Weight and scale them. Assign each criterion an explicit weight and a defined scoring scale, making the trade-offs visible rather than buried in a gut call.
Score every variant on the same axes. Apply the identical frame to each branch, using evidence supplied by other mechanisms, so results are directly comparable.
Apply the retention rule. Convert scores into a keep / cut / advance decision via a stated rule — a threshold, a ranking cut, a dominance test — rather than a fresh argument each time.

Tuning parameters¶

Criteria set — which dimensions count as fit. Broad sets capture more of what matters but dilute focus and invite score-padding; narrow sets are decisive but can miss a niche's real value.
Weighting — how much each criterion drives the result. Shifting weights can reverse the ranking, so the weights, not the scores, are where the real argument belongs.
Compensatory vs. non-compensatory — whether a high score on one axis can offset a failing grade on another, or whether each criterion is a hard gate. Gates protect against fatal weaknesses; compensation rewards all-round strength.
Retention rule — threshold, top-n, or dominance. A threshold keeps everything good enough (and may keep many); top-n forces consolidation.
Scoring rigour — quick qualitative bands versus quantified evidence per cell. More rigour resists bias but costs time and can lend false precision.

When it helps, and when it misleads¶

Its strength is that it makes selection legible and contestable: because the criteria and weights are explicit and set up front, a decision can be explained, audited, and re-run when circumstances change — the difference between a defensible choice and a rationalised one.^[1] Fixing the frame before the field is what stops the standard from quietly bending toward a favoured branch.

Its failure modes live in the weights and the scales. A rubric radiates false objectivity — precise-looking scores resting on subjective weightings that decide the outcome, so a bias merely moves from the verdict into the weighting. It rewards the measurable, quietly penalising a variant whose value resists a number. And its signature misuse is reverse-engineering the weights until the pre-chosen winner tops the table — selection theatre. The discipline is to set criteria and weights before scoring, to run the ranking's sensitivity to the weights, and to keep a human check on any branch the rubric would cut on a technicality.

How it implements the components¶

Multi-Criteria Selection Rubric realises the decision-criteria side of the archetype — the components that define what counts as fit and what is kept:

niche_fit_criterion — it specifies, weights, and scales the dimensions on which a variant's fit to its niche is judged.
selective_retention_rule — it states the rule that converts scores into a keep / cut / advance decision.

It defines the standard but does not generate the evidence it scores (performance_evidence_capture — that's Parallel Pilot Trials and Lineage–Niche Fit Dashboard), nor carry out the cuts it implies (consolidation_pathway, exit_or_prune_rule — that's Merge and Deprecation Plan).

Instantiates: Opportunity-Gated Adaptive Diversification — this rubric is the archetype's explicit, shared standard for selection.
Consumes: the marginal-response evidence from Parallel Pilot Trials, which fills the cells the rubric scores.
Sibling mechanisms: Parallel Pilot Trials · Merge and Deprecation Plan · Innovation Portfolio Review · Network Mixing Protocol · Diversity Coverage Matrix · Diversity-Floor Rate Boost · Experimental Cohort Split · Protected Pilot Lane · Stage-Gate Exploration · Lineage–Niche Fit Dashboard · Niche Portfolio Matrix · Opportunity Landscape Mapping · Preserve–Prune–Recombine Review · Saturation and Crowding Review · Specialization Cohort Seeding

References¶

[1] Making selection criteria explicit, weighted, and applied uniformly across options is the core of multi-criteria decision analysis; the design-engineering version, Pugh's controlled-convergence matrix, scores concepts against weighted criteria for exactly this reason — to keep the comparison honest rather than intuitive. ↩