Epistemic Feedback Rituals — Practices That Keep Truth in Contact 1.2.1-2

Rituals, Not Habits

The article (1.2.1) names epistemic self-correction practices as one of three source-pattern families for epistemic integrity. These practices are structural — they are designed into organizational process, not left to individual initiative. The distinction matters. A habit is personal and voluntary; a ritual is institutional and scheduled. Epistemic feedback rituals exist because the default behavior of organizations is to process truth until it hardens into assumption, and then to stop examining it.

This sidebar describes four rituals and specifies what distinguishes a functioning ritual from its performative counterpart.

The Four Rituals

The premortem is conducted before a decision is finalized. Participants are asked to assume the decision has failed and to identify the most plausible reasons for that failure. The structural function is to surface objections, risks, and blind spots that the normal decision process suppresses — not because participants are dishonest, but because convergence pressure in groups systematically discounts dissent.

A premortem works when it is scheduled into the decision pipeline, when participation includes voices outside the immediate team, and when the results are documented and revisited. It fails when it is conducted as a formality after the real decision has already been made, or when the findings are collected and filed without affecting the decision.

The postmortem is conducted after a failure, a near-miss, or the completion of a significant project. Its function is to extract structural lessons — not to assign blame. The question is not "who was responsible" but "what systemic conditions produced this outcome, and what would need to change for a different outcome to be structurally likely?"

A postmortem works when it is conducted close enough to the event that evidence is still accessible, when its findings are connected to design changes rather than personnel actions, and when the process is repeated rather than improvised. It fails when it becomes a blame-assignment ritual, when its recommendations are recorded but never implemented, or when it is skipped because the failure was "not bad enough" to warrant examination.

Open artifacts are documents, datasets, and reasoning chains that are made visible to authorized reviewers — not because transparency is a virtue in the abstract, but because hidden reasoning cannot be checked. An open artifact is not the same as a public artifact. It must be accessible to those with a structural need to verify the claim, not necessarily to the general public. The principle is auditability, not exhibition.

Open artifacts function when they include not just conclusions but the evidence, assumptions, and reasoning that produced those conclusions. They fail when they are produced as compliance documents — formatted to demonstrate that a process was followed, without containing enough substance to allow anyone to actually verify the claim.

Review cadence is the scheduled reassessment of claims, policies, and assumptions that have been in operation long enough to go unquestioned. The structural insight is that truth has a shelf life in institutional systems — not because reality changes, but because the context in which a claim was valid may shift, the data underlying it may degrade, or the metrics used to verify it may drift from what they originally measured.

Review cadence works when it is periodic and independent of crisis — scheduled, not triggered. It fails when reviews are conducted only in response to problems, because by that point the damage has already accumulated, and the review is operating under pressure that compromises its honesty.

The Performative Counterpart

Every epistemic ritual has a performative counterpart — a version that looks structurally identical but carries no corrective force. Performative premortems collect risks that everyone agrees to ignore. Performative postmortems assign lessons that everyone agrees not to learn. Performative open artifacts satisfy audit requirements without containing auditable content. Performative review cadences reaffirm the status quo at regular intervals.

The difference between a functioning ritual and its performative counterpart is not in the format. It is in whether the output changes anything. A ritual that cannot alter the decision, the design, or the assumption it examines is not epistemic — it is theatrical. Epistemic integrity as a design requirement means that the rituals must be connected to the system's decision architecture, not suspended in a parallel process that reports but does not affect.