Coherence Review¶
Review process — instantiates Aesthetic Coherence System
A recurring human gate that judges whether a proposed instance reads as part of the system, still carries its cues' meaning, and — where it deviates — routes the exception to a decision.
A coherence review is the process side of the archetype: a repeated act of human judgment applied to proposed work before it ships. It is not an artifact and it does not define rules — it exercises them. Its defining move is that it tests perception and interpretation, not just checklist compliance: a reviewer asks whether a new instance actually reads as part of the intended system, whether the recurring cues in it still mean what they are supposed to, and — when the work departs from the rules — whether that departure is a justified exception or a violation, and whether a run of similar exceptions is a signal that a rule should change. It is the ritual that keeps a system alive rather than frozen.
Example¶
A city's civic digital-service team ships online forms — benefits applications, permit renewals, records requests — built by several partner agencies. To keep them feeling like one government service, every new page passes a standing coherence review before launch: a thirty-minute session, twice a week, where a reviewer and the building team walk the page. First the coherence check — does this read as the same service as the others, or does it feel like a vendor's generic template dressed in the logo? The reviewer catches that a new permit form's confirmation screen looks foreign. Next the meaning-continuity test — the recurring green "submitted" banner must mean your request is filed; here a team reused the same green for "saved as draft," which would teach residents the wrong thing, so it's flagged. Finally a housing department asks to add a custom emergency banner the system has no rule for; the reviewer logs it as an exception, grants it for this launch, and notes that this is the third emergency-banner request this quarter — enough to propose a real banner rule. Outcome: the page ships coherent, one cue is corrected before it can mislead, and a recurring exception becomes a queued rule update instead of quiet drift.
How it works¶
The review runs as a short, repeatable protocol with three passes borrowed from usability practice.[n1] The fit pass is holistic and human: set the new instance beside real siblings and judge whether it belongs — reviewers look for felt relatedness, not line-item conformance. The meaning pass singles out each recurring cue in the instance (a color-coded state, an icon, a layout signal) and checks that it still carries its established meaning rather than being reused for something else. The exception pass handles every deliberate deviation: record it, decide grant or deny with a reason, and — crucially — scan the log for repeats, because a deviation that keeps recurring is evidence of a missing rule, not a series of one-off sins. The output is a decision plus, over time, a queue of rule-change proposals.
Tuning parameters¶
- Gate strength — advisory feedback versus a hard launch-blocking gate. A hard gate maximizes coherence but becomes a bottleneck and breeds bypasses when under-resourced.
- Cadence and batch size — frequent small reviews catch drift early but tax reviewers; infrequent big reviews are cheap to run but let incoherent work accumulate.
- Human-judgment vs. checklist ratio — more open judgment catches meaning failures a checklist misses but is slower and less consistent between reviewers; more checklist is fast and repeatable but rubber-stamps the letter while missing the spirit.
- Exception-to-rule threshold — how many repeated exceptions trigger a rule-change proposal. A low threshold keeps rules current but churns them; a high one lets real context boundaries stay unrecognized.
When it helps, and when it misleads¶
Its strength is that it catches exactly the failures no artifact can prevent on its own: work that is technically compliant yet reads as foreign, and cues that have quietly been repurposed into meaning the wrong thing. By treating exceptions as learning signals it turns local deviation into rule evolution rather than fragmentation.
Its failure mode is degenerating into checklist compliance — the review keeps meeting but stops actually asking whether the work reads as part of the system, so it green-lights on-spec pages that still feel unrelated. The classic misuse is taste capture: one reviewer's personal preferences harden into de facto rules with no path for the people who must interpret the system to push back. The guarding discipline is to keep the fit pass genuinely perceptual (compare against siblings, not against a rubric), rotate or pair reviewers, and hold the exception log open so patterns — not individual taste — drive rule changes.
How it implements the components¶
coherence_check— the fit pass is this check: a human judgment of whether a proposed instance reads as part of the intended system, run before it ships.meaning_continuity_test— the meaning pass verifies that each recurring cue in the instance still carries its established meaning rather than being silently repurposed.exception_governance— the exception pass records, decides, and reviews deviations, escalating repeated ones into proposed rule updates.
It defines no visual_rule, aesthetic_principle_set, or system_boundary — a review applies those, it does not author them; that authoring is done by Brand System and Visual Language System. It also runs no drift_monitor: catching gradual system-wide fragmentation over time is Component Library's standing watch, whereas this review judges one instance at a time.
Related¶
- Instantiates: Aesthetic Coherence System — it is the maintained governance loop that keeps new work fitting the system.
- Consumes: Visual Language System — the review checks proposed work against the vocabulary and cue-meanings that system defines.
- Sibling mechanisms: Brand System · Visual Language System · Color Palette · Component Library · Environmental Design Standard · Design System · Pattern Library
Editorial Notes¶
Form Classification¶
Form family: Assessment, Review & Assurance
Rationale: A recurring human gate that judges whether a proposed instance reads as part of the system, still carries its cues' meaning, and — where it deviates — routes the exception to a decision, making its operative form a bounded evaluation of existing evidence or work that produces a finding or disposition.
Independent corroboration: The frozen evidence defines Coherence Review as 'A recurring human gate that judges whether a proposed instance reads as part of the system, still carries its cues' meaning, and — where it deviates — routes the exception to a decision', so its operative form is Assessment, Review & Assurance.
Review outcome: Independent reviewer agreement; high confidence.
Origin Attribution¶
Primary origin: Art & Aesthetics
Origin pattern: Cross-disciplinary synthesis
Present-day reach: Multi-domain
Rationale: Art direction and design criticism made perceptual coherence, recurring cues, and justified deviation objects of expert review before release.
Related originating lineages:
- Human-Computer Interaction — Design systems supply semantic tokens, exception routing, and consistency review across interfaces.
Review resolution: Both reviewers agree on art_aesthetics as primary. Reading the source mechanism confirms that its defining operation belongs to that lineage; the final record retains human_computer_interaction only where it materially formed the mechanism and keeps present-day application breadth separate from provenance.
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¶
[n1] Heuristic evaluation, the usability-inspection method associated with Jakob Nielsen, has a small set of expert reviewers judge an interface against agreed principles rather than testing it with users. A coherence review borrows the shape — expert judgment against shared principles — but aims it at system fit and cue meaning rather than usability. ↩