Review record
Pinned evidence spans, the checklist result, and whether the model verdict was revealed before or after the review.
Active epistemic scaffolding
Replace passive friction with low-effort discrimination tasks that make the reviewer process the evidence before the recommendation.
- 1 Evidence
- 2 Authority
- 3 Decision acts here
- 4 Consequence
- 5 Challenge
- 6 Reconsideration acts here
- 7 Correction
Where it acts
At Decision and Reconsideration, two of the seven stages every consequential decision passes through. It raises the fidelity of review without adding the passive friction that produces rubber-stamping.
Steps
Put the mechanism in motion
7 steps, in order. Check each one off as it is done; the records show the work was done.
0 of 7 steps complete
Records this mechanism keeps
What a team running this mechanism keeps, and what each record is for.
Falsification checklist
The common failure modes for the decision type, with an owner and a review date.
Canary protocol
Disclosure, environment, and the exclusion of canaries from live care.
Pacing log
Aggregate cadence and queue-throttle events, with no individual scores.
Reuse-ready snippets
Copy policy, audit, and incident language
Use these snippets in requirements, audits, and postmortems with the mechanism permalink.
Policy requirement
Policy requirement (MEC-30 Active epistemic scaffolding) - Review interfaces use active discrimination tasks rather than passive friction. - Canaries are disclosed and excluded from live care. - Pacing data is aggregate and never used in individual performance review. Reference: https://ethotechnics.org/mechanisms/patterns/active-epistemic-scaffolding
Product requirement
Product requirement (MEC-30) - The review record carries the pinned evidence spans, the checklist result, and whether the model verdict was revealed before or after the review. - Every control has a keyboard equivalent and a stable position.
Audit evidence checklist
Audit evidence checklist (MEC-30) [ ] No countdown timers, confirmation modals, or mandatory free-text justifications. [ ] Pinned evidence spans are recorded for each review. [ ] The falsification checklist is current and has an owner. [ ] Canaries are disclosed and not in live care. [ ] Pacing data is aggregate and excluded from individual performance review. [ ] All controls are keyboard-accessible and stable.
Postmortem trigger
Postmortem trigger (MEC-30) Trigger review when a review record has no pinned evidence, when a canary is found in live care, or when pacing data is used in individual performance review.
Example usage
A prior-authorization review queue
One scenario with each part of the mechanism in place.
How it plays out
The model verdict is hidden until the reviewer opens it. The reviewer pins the policy clause and the lab value that carry the decision. Borderline cases show a three-item falsification checklist. Canaries run only in the training environment. When median review latency drops below the cognitive floor, the queue slows and switches to a comparative view. Pacing is reported in aggregate and never enters an individual score.
Anti-patterns
Common failure cases and counterfactuals
Each case passes on paper and fails in operation. The counterfactual shows what would have worked.
Friction theater
Countdown timers, confirmation modals, and mandatory free-text boxes slow the operator without engaging the evidence.
Counterfactual: The interface requires a small act of discrimination that lands the reviewer's attention on the decisive evidence.
Exception: If the operator can dismiss it without reading, it is not a review.
Canary surveillance
Synthetic cases are used to score individual operators, so vigilance becomes a performance metric and the canaries become another source of defensive behavior.
Counterfactual: Canaries are disclosed, used in simulation or low-stakes streams, and reported in aggregate.
Exception: A canary that can punish the operator is a trap, not a calibration.
Randomized controls
Moving the approve and deny controls breaks keyboard access and muscle memory, increasing slips without engaging reasoning.
Counterfactual: Controls keep stable positions; a deliberate gesture or two-step confirmation is reserved for high-stakes or low-confidence cases.
Exception: Break the reflex, not the accessibility contract.
The stale checklist
The falsification checklist is written once and never updated, so it becomes a new checkbox.
Counterfactual: The checklist has an owner, a review date, and a record of what changed.
Exception: A checklist without a maintenance owner is a modal dialog with better branding.
Diagnostics for this mechanism
Tools that size the risk this mechanism addresses.
Diagnostic
Delegation audit
Walks a team through one workflow against six questions and returns an exposure score, the permissions nobody can justify, and whether its decisions can be undone. Start here when the system keeps no decision records yet; use the record conformance checker once it does.
Cite this page · v1.2.1 · updated Oct 6, 2026 · CC BY-SA 4.0
APA
Ethotechnics Institute. (2026). Active epistemic scaffolding (MEC-30). Ethotechnics Institute. https://ethotechnics.org/mechanisms/patterns/active-epistemic-scaffolding
MLA
Ethotechnics Institute. "Active epistemic scaffolding (MEC-30)." Ethotechnics Institute, 2026, https://ethotechnics.org/mechanisms/patterns/active-epistemic-scaffolding.
Chicago
Ethotechnics Institute. "Active epistemic scaffolding (MEC-30)." Ethotechnics Institute. Oct 6, 2026. https://ethotechnics.org/mechanisms/patterns/active-epistemic-scaffolding.
BibTeX
@misc{mechanism_active-epistemic-scaffolding,
title={Active epistemic scaffolding (MEC-30)},
author={Ethotechnics Institute},
year={2026},
howpublished={Ethotechnics Institute},
url={https://ethotechnics.org/mechanisms/patterns/active-epistemic-scaffolding},
version={v1.2.1}
}
RIS
TY - WEB TI - Active epistemic scaffolding (MEC-30) AU - Ethotechnics Institute PY - 2026 UR - https://ethotechnics.org/mechanisms/patterns/active-epistemic-scaffolding ER -