R&D / PERSUASION · DOC CLASS — BIAS BANK

Cognitive-Bias Bank

The Bias Archive preserves the raw intelligence of manipulation — the documented cognitive biases every dark pattern exploits, and the persuasion models reverse-engineered from the PL-4 constructs — so that no exploited mechanism is ever lost to the Lab. It is the institutional memory behind every counter-pattern the Directorate designs.

Doc Control QH-RND-0004 Revision Rev A Effective 2091-01-01 Owner R&D / Persuasion Curated by GOVERNOR Classification INTERNAL // QUANTUM-ZONE-TASK-FORCE EYES-ONLY

The Bias & Model Archive

The Bias & Model Archive (zone CV) is one of the core THE RESET facility zones — a hardened, access-controlled intelligence repository holding the catalogued mechanisms of the entire Index. Each cognitive bias is documented with its trigger conditions and effect size; the persuasion models reverse-engineered from PL-4 constructs are held encrypted and air-gapped; and the rarest primitives are kept under the deepest classification seal. Every entry is logged against the techniques that exploit it and its place in the Index.

The Archive is engineered for redundancy first. Material is split across independent encrypted repositories on separate keys, with a mirrored secure backup at a second site so no single failure can erase a record. Integrity, access, and classification state stream over the Instrumentation Mesh to the Pattern Fabric; a breach in any repository trips an alarm before anything can be exfiltrated. Its purpose is plain: when a new dark pattern appears, the Archive is what lets the Lab name the bias it exploits and build the counter before it spreads.

⬡Secure Repository

Secure Repository

Persuasion models and bias profiles held in redundant encrypted, air-gapped repositories. Split across two sites with continuous integrity, access, and classification telemetry — designed so no single failure, and no single intruder, can destroy or steal a record.

⬡Bias Registry

Bias Registry

The authoritative ledger of every catalogued bias: trigger, effect size, the techniques that exploit it, and current status. Each entry is keyed to the Index and to its active Counter-Pattern Lab line.

🜂Counter-Pattern Derivation

Counter-Pattern Derivation

The protocol that turns an archived bias or model into a working counter-pattern — isolate the mechanism, design the inverse, and hand it to the Lab to validate. The end of the chain that begins in the Archive.

Catalogued Biases

The Bias Registry tracks the cognitive biases whose models are held in the Archive. Each carries a registry ID (QH-CB-2091-0xx), ties back to the techniques that exploit it in the Index, and feeds the active Counter-Pattern Lab line that builds against it.

Registry ID	Bias	Exploited by	Models	Status
`QH-CB-2091-003`	Loss Aversion	Streaks · Fake Scarcity · Retention traps	15	Active
`QH-CB-2091-011`	Social Proof	Fake Social Proof · Astroturf · Manufactured Consensus	11	Active
`QH-CB-2091-018`	Variable Reward (intermittent reinforcement)	Infinite Scroll · Loot Box · Pull-to-Refresh · Notification Roulette	19	Growing
`QH-CB-2091-024`	Sunk-Cost Fallacy	Forced Continuity · Gamified Grind · Streak Anxiety	8	Active
`QH-CB-2091-029`	Authority & Trust Bias	Disguised Ad · Misleading Hierarchy · Fake Expert Endorsement	10	Active
`QH-CB-2091-035`	Anchoring	Drip Pricing · Forced Bundling Upsell · Decoy Pricing	6	Active
`QH-CB-2091-041`	FOMO (Fear of Missing Out)	Countdown Timer · Limited Offer · Activity Notification Storm	13	Growing
`QH-CB-2091-046`	Status Quo Bias	Dark Default · Pre-checked Consent · Roach Motel Cancellation	5	Under-modeled

A bias is never retrieved from the Archive alone. The techniques that exploit it, their case history, and field prevalence live in the Index; the work that keeps each model current and builds against it runs out of the Counter-Pattern Lab. The Archive is the long-term intelligence repository, not the working catalog.

Catalog Integrity & Coverage

Coverage

Keep the catalog complete

A gap in the bias library is a technique the Lab cannot counter. Every newly logged technique is checked against the Registry for the bias it exploits, and any unmodeled mechanism is opened as a coverage gap — so a weak sector of the catalog can be reinforced before an attacker finds the hole first.

Freshness targets

Model-freshness floors

Persuasion models drift as platforms evolve; each carries a minimum freshness target. A model that drops below floor (see QH-CB-2091-046) is flagged Under-modeled and prioritized for re-derivation from current field data before any counter-pattern derivation is allowed to rely on it.

Classified models — separate protocol. The Archive holds documented biases and derived models only. The PL-4 Classified constructs — the Sentinel Compulsion Stack and the models reverse-engineered from it — are not archived or governed under this document; they are held and handled under a separate, restricted protocol with its own custody chain. Those are tracked in Restricted Techniques and never enter the standard registry.

Access & Security

Chain of custody is absolute. A persuasion model is a strategic asset and is treated as one. Every record is logged on accession and on retrieval; nothing leaves a repository without a two-person, signed entry. Only R&D staff on the current Archive roster may retrieve a model, and only against an approved counter-pattern derivation or Lab authorization — never on standing access. Retrievals tied to Classified models require Steward sign-off in addition to the standard custody chain. Procedures live in Standard Operating Procedures; custody, ownership, and disclosure questions are adjudicated by Legal & Deniability.