Skip to content

Filter Rationale Register

Register — instantiates Structural Filter Intersection Audit

A standing record of why each filter exists — its stated purpose, owner, and the legitimacy standard it claims — so every screening rule can be traced, justified, or challenged.

The Filter Rationale Register is the durable accountability layer: one written entry per filter capturing its origin, its stated rationale, its owner, and — as a required field — the legitimacy standard it claims to serve (safety, law, quality, resource scarcity, taste, brand). Where the Filter Stack Map records what the filters are, the register records why each is said to be justified, and holds that claim on the record so it can be examined over time. It is, in effect, an architecture-decision record for a screening pipeline: a filter with no entry has no accountable reason to exist.

Example

An academic journal desk-rejects roughly seventy percent of submissions before peer review, on criteria that live only in editors' heads: "not a fit for our scope," "methods too weak," "topic feels overdone." The Rationale Register forces each screening rule into an entry — the rule, when and why it was adopted, who owns it, and the legitimacy standard it claims. Writing them down exposes that "not a fit" actually bundles three different rationales, one of which — "unlikely to be well cited" — is really about the journal's impact metrics, not the work's merit. The register does not change the rule. It makes the rule's justification legible, attributable, and therefore contestable.

How it works

The distinguishing move is a mandatory legitimacy field: every filter must name the standard it appeals to, drawn from an explicit taxonomy, so "we just know quality when we see it" cannot pass as a reason. Each entry is durable and maintained, and individual screening decisions link back to the registered rule that produced them — the traceability that lets a specific outcome be followed to a specific, dated justification. Because entries persist, a rule's rationale can be revisited, and a drift in how it is justified is caught on the record.

Tuning parameters

  • Legitimacy taxonomy — the menu of allowed justifications (legal, safety, quality, resource, brand, taste). A tighter menu forces honest classification; a looser one lets rules hide behind a vague "quality."
  • Entry granularity — one rationale per filter, or per sub-rule. Finer is more honest and heavier to maintain.
  • Public vs. internal — whether the register is visible to producers and audiences or held internally. Publishing raises accountability and can lower candor.
  • Traceability depth — whether each individual decision links back to a registered rule, or only the rules themselves are logged. Per-decision tracing enables appeals but costs recordkeeping.

When it helps, and when it misleads

Its strength is converting tacit, deniable screening norms into on-the-record, challengeable justifications — and it is the backbone that appeals and drift reviews read from. Its failure mode is that a register captures the stated rationale, which may be a rationalization: a well-written justification can launder an illegitimate filter into respectability. The classic misuse is writing the rationale after the fact, to defend a rule rather than to test it. The discipline that guards against this is to hold each registered rationale against the filter's actual outcomes — what it really excludes — and against an explicit legitimacy standard, instead of accepting the prose at face value.

How it implements the components

  • filter_traceability_record — the durable, per-filter log linking each screening rule (and, at depth, each decision) back to a recorded origin and owner.
  • filter_legitimacy_standard — the required field on every entry naming the standard the filter claims to serve, making legitimacy explicit and auditable rather than assumed.

It does NOT judge whether that claimed legitimacy holds in practice, adjudicate individual contested cases (appeal_exception_public_reason_channel — Appeal and Exception Review), or enumerate the filters themselves (parallel_filter_inventory — Filter Stack Map). It records the justification the others rely on.

Notes

The register is read by two downstream mechanisms — the appeal channel adjudicates against it, and the periodic drift review measures against it — so a sloppy or incomplete register silently weakens both. Its quality is not self-contained; it is load-bearing for everything that later has to ask "on what grounds?"