Answerability Review Meeting¶
Ritual — instantiates Accountability Chain Design
Creates a recurring or event-triggered forum where owners explain decisions, reviewers test reasoning, and repair commitments are assigned.
An Answerability Review Meeting is the standing occasion on which owners are actually made to answer. It is the live event — not a document, not a tracker — where the person accountable for a decision sits before a body with standing to question them, explains what they did and why, has that reasoning tested, and leaves with named repair commitments or an escalation. Its defining idea is answerability as a relationship enacted in real time: accountability is not a record on a shelf but the recurring moment at which a forum can pose questions[1], pass judgement, and require follow-through. Everything upstream (who owns what, why it was decided) is input to this meeting; the meeting's own contribution is the questioning and the commitments that come out of it.
Example¶
An AI platform company runs a fortnightly Model Release Review. Any team shipping a model change to production must appear. This week the recommender team is defending a ranking-model update. The release owner walks the forum through the change; the standing reviewers — a safety lead, a fairness researcher, a product director — probe the reasoning: Why is the offline metric trusted here? What happens to the cohorts that regressed? Under questioning the owner concedes the fairness evaluation was run on stale data.
The forum does not itself fix the model. What it does is bind the outcome: the release is held, the owner leaves with a repair commitment (re-run the evaluation on current data, report back next session), and because the regression touches a protected-class question beyond the forum's remit, the meeting escalates it to the company's responsible-AI council. Two weeks later the owner is back in the same room to answer for whether the commitment was met. The ritual's power is not the meeting itself but the certainty that one is always coming.
How it works¶
- Standing and cadence. The forum has recognized authority to require attendance and answers — recurring on a fixed schedule and convenable on a triggering event (a serious failure, a contested decision). Without standing it is a status update, not answerability.
- Owner answers, forum tests. The accountable owner explains; reviewers who did not make the decision interrogate the reasoning. The separation of decider from questioner is what makes the test real.
- Commitments, not conclusions. The meeting's output is a set of assigned repair commitments with owners — the consequence path — recorded and revisited, not a set of opinions that evaporate when the room empties.
- Escalate what exceeds the room. When an issue outruns the forum's authority, severity, or independence, the meeting routes it upward rather than papering over it.
Tuning parameters¶
- Trigger vs cadence — purely scheduled review versus event-triggered convening. Regular cadence catches slow drift; event triggers catch acute failures. Most chains need both.
- Forum independence — reviewers drawn from inside the owner's chain versus outside it. Independence resists capture and softball questioning but costs context and goodwill.
- Questioning depth — a light attest-and-move-on versus adversarial probing. Deeper tests catch weak reasoning but slow throughput and can curdle into interrogation if the culture is punitive.
- Commitment bindingness — whether repair commitments are advisory or gate the decision (e.g., a release is blocked until met). Binding is stronger and slower.
- Escalation threshold — how far beyond the room's remit an issue must reach before it is kicked up. Set too high, the forum quietly absorbs things it has no standing to resolve.
When it helps, and when it misleads¶
Its strength is that it puts a live cost on unexamined decisions: an owner who knows they will have to explain a choice to a questioning forum decides differently than one who will not. It is the point in the chain where recorded rationale is actually stress-tested and where explanation is converted into committed repair.
Its failure modes are social. A forum can be captured — packed with allies who ask soft questions — so the meeting performs scrutiny without applying it, the accountability-theater version of a review. It can also decay into ritual attendance, where owners recite and the forum nods, producing minutes but no binding commitments. And a punitive forum drives the truth underground: owners learn to present defensible stories rather than real ones. The guarding discipline is to seat at least some reviewers outside the owner's chain, to require that every session end with tracked commitments or an explicit "nothing to repair," and to keep the questioning hard on the decision but not on the person.
How it implements the components¶
answerability_forum— the meeting is the forum: a body with standing to require attendance, pose questions, and demand justification for a decision.repair_or_consequence_path— the session converts explanation into assigned, revisited repair commitments (or sanction), so answering leads to correction rather than ending at talk.escalation_trigger— when an issue exceeds the forum's authority or independence, the meeting routes it upward instead of absorbing it.
It runs the forum but does not itself name who owns each decision beforehand (accountable_owner, action_or_decision — the Accountability Matrix), and it questions reasoning without doing the root-cause analysis that mines an incident for what was known and why (evidence_or_context_record, rationale_record — that is Postmortem Action Tracking); this is the live forum where owners answer, not the investigation behind an answer. It assigns repairs but leaves their tracking to verified closure (closure_audit) to the Corrective Action Register.
Related¶
- Instantiates: Accountability Chain Design — the meeting is the forum strand: where the owner-record chain is actually put to the question.
- Consumes: Accountability Matrix tells the forum who owes an answer; the Corrective Action Register tracks the commitments it assigns.
- Sibling mechanisms: Accountability Matrix · Corrective Action Register · Escalation Record · Incident Ownership Protocol · Postmortem Action Tracking · Decision Log · Role Charter
Editorial Notes¶
Form Classification¶
Form family: Assessment, Review & Assurance
Rationale: Creates a recurring or event-triggered forum where owners explain decisions, reviewers test reasoning, and repair commitments are assigned, making its operative form a bounded evaluation of existing evidence or work that produces a finding or disposition.
Independent corroboration: The frozen evidence defines Answerability Review Meeting as 'Creates a recurring or event-triggered forum where owners explain decisions, reviewers test reasoning, and repair commitments are assigned', so its operative form is Assessment, Review & Assurance.
Nearest alternative: Organization, Role & Governance — The forum's primary act is testing reasons and assigning repair commitments, while standing authority enables it.
Review outcome: Independent reviewer agreement; medium confidence.
Origin Attribution¶
Primary origin: Public Administration & Policy
Origin pattern: Cross-disciplinary synthesis
Present-day reach: Multi-domain
Rationale: Public-administration accountability theory defines answerability as an actor's duty to explain conduct to a forum able to question and impose consequences.
Related originating lineages:
- Law & Governance — Hearings and due process supply standing and consequence pathways.
- Organizational & Management Science — Review cadence and repair assignments provide the meeting form.
- Political Science — Democratic-accountability theory supplies the forum relationship.
Encyclopedia synthesis: The exact catalogued form synthesizes established practice rather than reproducing a single standard historical label.
Review outcome: Independent reviewer agreement; high confidence.
References¶
[1] Bovens, M. "Analysing and Assessing Accountability: A Conceptual Framework". European Law Journal 13(4), 447–468 (2007). Defines accountability as an actor–forum relationship in which the forum can question the actor and pass judgment. registry ↩