Gatekeeping Decision Log¶
Artifact — instantiates Accountable Gatekeeping Design
Records every passage decision — outcome, reason code, decider, evidence, timing, and any override — so the gate's choices, including the denials no one downstream can see, become an inspectable trail.
Everything else in an accountable gate — audits, appeals, pattern detection, recalibration — needs a record to work on, and most gates never keep one for the cases that didn't pass. Gatekeeping Decision Log is that record: a structured, append-only ledger capturing each decision at the choke point — the case, the outcome (passed, rejected, deferred, demoted, rerouted), the reason code and evidence relied on, who or what decided, the timestamp, and whether any override or exception was invoked. Its defining property is that it records the whole population that reached the gate, not just the survivors — so the denied, delayed, and diverted cases the downstream audience can never see nonetheless exist as inspectable data. The log itself makes no judgment and sends no notice; it is the raw material every other accountability mechanism reads from.
Example¶
A national research-funding agency awards a few hundred grants a year out of several thousand applications; the funded list becomes, in effect, the field's map of what work is worth doing. For years the only durable artifact was the list of winners. The agency introduces a Gatekeeping Decision Log: for every application — funded or not — a record is written capturing the panel's score, the reason code for the outcome (methodology weak, out of scope, fundable but below the payline), the reviewers assigned, any declared conflict and how it was handled, the date, and whether a program officer overrode the panel ranking.
Nothing about the individual decisions changes on the day. But a year later the log is what makes several previously unanswerable questions answerable: whether early-career applicants are rejected for scope at a higher rate than established labs, whether one program officer's overrides cluster suspiciously, whether "fundable but below the payline" is quietly absorbing most rejections. The funded list alone could never show any of this. The log turned the gate's invisible half — the thousands of denials — into something that can be counted.
How it works¶
- Log the whole population, not the survivors. Every case that reaches the gate gets a record regardless of outcome; the denied and deferred cases are the ones the log exists to preserve.
- Structure the reasons. Free text resists analysis; a controlled set of reason codes (with room for detail) is what later lets patterns across thousands of decisions be counted.
- Capture the decider and the override. Recording who or what decided, and whether discretion or an exception was used, is what makes conflict, capture, and drift detectable later.
- Append, don't overwrite. The log is immutable after the fact; corrections are new entries, so the record cannot be quietly rewritten to match a later story.
Tuning parameters¶
- Field granularity — how much each entry captures (bare outcome vs. full inputs, evidence, and deliberation). Richer records enable deeper audits but cost reviewer time and create sensitive data to protect.
- Reason-code taxonomy — how fine the controlled vocabulary is. Coarse codes are easy to apply but blur distinctions; fine codes capture nuance but drift in meaning between reviewers unless calibrated.
- Retention and access tier — how long records live and who may read them (internal only, trusted auditors, affected party, aggregate public). Trades inspectability against privacy and gaming risk.
- Immutability strength — soft edit history vs. hard append-only or cryptographic seal. Stronger immutability resists tampering but makes legitimate correction heavier.
- Coverage — whether informal overrides, desk rejections, and never-surfaced cases are logged. The cases most tempting to leave out are exactly the ones the archetype most wants recorded.
When it helps, and when it misleads¶
Its strength is foundational: it is the artifact every other mechanism here depends on. An audit needs decisions to audit; an appeal needs a record to reopen; pattern detection needs a population to count. Above all the log makes the gate's denied half exist as data, the archetype's central demand. Append-only structure means the record cannot be quietly reshaped after the fact.
It misleads when its coverage is partial or its reasons are theatrical. A log is only as honest as what gets written down, and the standing temptation is to record the clean cases while awkward overrides and informal desk-rejections happen off the books — leaving a tidy trail that looks accountable precisely where it isn't. Reason codes can collapse into a single defensible catch-all ("does not meet criteria") that records nothing real. And a log can manufacture false confidence: its mere existence proves nothing if no one ever reads it. The sharpest misuse is writing the reason after the decision to rationalize it, which inverts the log's purpose; the corrective is the administrative-law principle that a decision must stand on the reasons actually given at the time[1], not on justifications supplied later. The disciplines that keep a log honest are contemporaneous entry, mandatory logging of overrides and exceptions, and never letting the log stand alone — it earns its keep only when an audit or appeal actually consumes it.
How it implements the components¶
Gatekeeping Decision Log fills the record layer of the archetype — the components that turn transient decisions into durable, inspectable data — and only those:
choke_point_passage_map— by recording every case and its outcome (passed, rejected, deferred, demoted, rerouted), the log's aggregate is the map of what flows through the choke point, including the denied and delayed populations no one downstream can see.decision_reason_record— each entry preserves the reason code, evidence, decider, and override status for a single decision, the atom of inspectability the rest of the pattern reads.
The log records but does not act: it does not analyse the record for bias or drift (that is Algorithmic Ranking Audit and Random Sample Audit), deliver a reason to the affected party (that is Reasoned Decision Notice), route a contested case (that is Appeals or Reconsideration Workflow), or set the criteria whose codes it stores (that is the Editorial Standards Board).
Related¶
- Instantiates: Accountable Gatekeeping Design — the log is the substrate the pattern's audit, appeal, and recalibration steps all read from.
- Sibling mechanisms: Reasoned Decision Notice · Algorithmic Ranking Audit · Random Sample Audit · Appeals or Reconsideration Workflow · Transparency Report · Published Selection Criteria · Editorial Standards Board · Conflict-of-Interest Disclosure · Blind or Double-Blind Review · Independent Review Panel · Quota or Portfolio Guardrail
Editorial Notes¶
Form Classification¶
Form family: Record, Log & Register
Rationale: Gatekeeping Decision Log operates as a durable record, ledger, register, or trace whose value depends on preserving actual state or history because it records every passage decision — outcome, reason code, decider, evidence, timing, and any override — so the gate's choices, including the denials no one downstream can see, become an inspectable trail.
Independent corroboration: The frozen evidence defines Gatekeeping Decision Log as 'Records every passage decision — outcome, reason code, decider, evidence, timing, and any override — so the gate's choices, including the denials no one downstream can see, become an inspectable trail', so its operative form is Record, Log & Register.
Review outcome: Independent reviewer agreement; high confidence.
Origin Attribution¶
Primary origin: Law & Governance
Origin pattern: Cross-disciplinary synthesis
Present-day reach: Multi-domain
Rationale: Administrative-law reason-giving, exemplified by Chenery, grounds contemporaneous records of gate decisions.
Related originating lineages:
- Accounting & Auditing — Audit-trail practice materially shapes append-only evidence, override, and decider fields.
- Public Administration & Policy — Administrative decision systems developed reason codes and reviewable denial records.
Review resolution: Both reviewers agree that law_governance is primary: Administrative-law reason-giving, exemplified by Chenery, grounds contemporaneous records of gate decisions. I retain accounting_auditing, public_administration_policy only as formative lineage, not as a list of later applications. I resolve origin_mode as cross_disciplinary_synthesis because the artifact joins distinct disciplinary contributions. I resolve domain_reach as multi_domain because it transfers across several fields but is not a domain-free primitive. Encyclopedia synthesis is true because the exact generalized packaging is an encyclopedia-authored combination or refinement.
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¶
The log is easy to confuse with Reasoned Decision Notice, but they sit on opposite sides of the transparency boundary: the log is the internal, whole-population record (including every denial), while the notice is the outward delivery of a single decision's reason to the affected party. A gate can keep a rigorous log and disclose almost none of it — the log is the full internal tier from which narrower disclosures are drawn.
References¶
[1] Supreme Court of the United States. Securities Commission v. Chenery Corp., 318 U.S. 80 (1943). Requires an administrative decision to stand on the grounds the agency actually invoked, not a different justification supplied later. registry ↩