Standardized Scoring Rubric¶

Scoring instrument — instantiates Rent-Seeking Channel Closure

Fixes the criteria, weights, and required evidence of an allocation in advance and in public, so awards turn on stated, checkable merit rather than on who has the decider's ear.

Standardized Scoring Rubric is a published, pre-committed instrument that defines exactly which factors may count in an allocation and how much each weighs, converting a discretionary judgment into a bounded, auditable score. Its defining move is constraining the decision rule itself — fixing what may and may not move the outcome — and encoding the public-interest or value-creation criteria as explicit scored dimensions, so a gatekeeper cannot smuggle in extraneous considerations like a relationship, a favour, or a side-payment. It is the criteria artifact: the fixed yardstick, distinct from the independent humans who judge the facts (the panel) and the record that publishes the reasons (the docket).

Example¶

A city awards a $30M transit-maintenance contract. Historically the selection committee "weighed everything," and the winner correlated suspiciously with who had lobbied hardest. The city now issues a standardized scoring rubric with the RFP: 40% technical approach, 25% price, 20% documented past-delivery performance, 15% workforce plan — each dimension with defined scoring anchors and the evidence a bidder must supply.

Bids are scored line by line against the rubric by independent evaluators. The incumbent's relationship advantage now counts for exactly zero, because "relationship" is not a scored dimension and nothing outside the list may move a score. A challenger with a stronger technical score and a lower price wins, and any losing bidder can check the arithmetic. The channel — influence the committee's overall impression — is closed, because there is no overall impression left to influence: only the sum of scores on stated criteria.

How it works¶

Criteria fixed ex ante and in public — the yardstick is set and published before bidders are known, so it can't be quietly tuned to a favoured party.
A bounded factor set — only listed dimensions may affect the score; anything else is out of bounds by rule, not by the scorer's forbearance.
Evidence-anchored scoring — each dimension carries defined anchors and required evidence, shrinking the room a scorer has to reward whom they please.

Tuning parameters¶

Weight allocation — how much each criterion counts. The weights encode the real value judgment and are the most contested and most gameable dial: price-heavy invites low-balling, quality-heavy invites subjectivity.
Criterion objectivity — measurable factors resist capture but miss hard-to-quantify quality; subjective factors capture quality but reopen discretion.
Timing of disclosure — full rubric up front, weights only, or criteria withheld until after award. Full ex-ante disclosure maximizes fairness but lets applicants optimize straight to the rubric.
Scorer independence and aggregation — single scorer or panel, and how scores combine (mean, trimmed mean, consensus).
Anchoring granularity — how tightly each score point is defined. Tight anchors cut discretion but add rigidity when a genuinely novel bid doesn't fit the boxes.

When it helps, and when it misleads¶

Its strength is closing the "work the decider" channel by removing the decider's unstructured discretion, making awards checkable and challengeable — and, because applicants optimize whatever is scored, steering their effort toward the merits the rubric rewards rather than toward influence.

Its failure mode is Goodhart's law: once the rubric is the target, applicants optimize the proxy rather than the goal, and a rubric that scores pages of documentation gets documentation, not quality.^[1] The classic misuse is drafting the weights around a pre-chosen winner's unique strengths — capture simply migrates from scoring the bids to writing the rubric. A rubric can also be so mechanical it excludes a genuinely superior non-standard bid. The discipline is to set and publish the criteria before parties are known, justify each weight against the public objective, have the rubric itself reviewed for tailoring, and keep a narrow, on-the-record override for the rare case it misfires.

How it implements the components¶

The rubric fills the rule-boundary and value-test components — it defines what may decide, not who judges or where reasons are posted:

allocation_rule_boundary — the rubric is the boundary: it fixes which factors may lawfully decide the award and rules everything else out of consideration.
public_interest_or_value_creation_test — the scored dimensions operationalize the value test, turning "does this serve the stated purpose?" into an explicit, weighted, checkable calculation.

It supplies neither the independent judges of contested fact (capture_firewall, standardized_evidence_package → Independent Technical Evidence Panel) nor the public reasons and response channel (transparency_and_reason_record, affected_party_participation_channel → Public Reason Docket).

Instantiates: Rent-Seeking Channel Closure — makes rule-mediated capture harder by fixing the merits the decision must turn on.
Sibling mechanisms: Public Reason Docket · Randomized or Lottery Allocation · Independent Technical Evidence Panel · Rent-Seeking Audit · Entry-Barrier Sunset and Review · Performance-Based Clawback · Regulatory Capture Audit · Beneficial Ownership and Influence Disclosure · Auction with Rent Recapture · Competitive Rebid or Retendering · Conflict-of-Interest and Recusal Rule · Cooling-Off Period · Sunset Clause Review · Anti-Capture Rotation Protocol

Notes¶

A rubric presumes merit is scorable. When candidates are genuinely indistinguishable on the merits, or when any criterion you could write would itself be captured, forcing a rubric manufactures false precision and opens a fresh capture surface in the rubric-drafting. That is the boundary where Randomized or Lottery Allocation takes over — the two are the merit-based and merit-refusing poles of the same problem.

References¶

[1] Goodhart's law — "when a measure becomes a target, it ceases to be a good measure." A published rubric is optimized by applicants, so the scored proxies drift from the merit they were meant to track unless the criteria are chosen to be hard to game. ↩