Active Research Threads

Project MERIDIAN — Attention-capture taxonomy expansion

Opened 2092-01-18 · Lead: Dr. Elias Brandt, Attention Economics

MERIDIAN extends the existing PL-1 through PL-3 attention-capture catalog with forty-six newly identified hook variants sourced from UNDERTOW intercepts during the Q4 2091 sweep. Brandt's team has completed structural decomposition on thirty-one variants so far; fifteen remain unclassified due to polymorphic presentation that shifts depending on user engagement history. Counter-pattern candidates for the first tranche are in bench validation with projected handoff to the Compendium by mid-cycle.

Project CANOPY — Narrative laundering propagation mapping

Opened 2092-02-04 · Lead: Dr. Tomás Herrera, Narrative Systems

CANOPY traces the lifecycle of laundered narratives from their origin platforms through amplification networks and into mainstream consumption channels. Herrera's propagation model now identifies four distinct laundering stages — seeding, credentialing, cross-pollination, and normalization — with measurable dwell-time signatures at each stage visible on the Instrumentation Mesh. The current focus is automating stage-two detection so GOVERNOR can flag credentialing events within minutes rather than hours.

Project THRESHOLD — Sub-perceptual manipulation dose curves

Opened 2092-02-19 · Lead: Dr. Kenji Hashimoto, Behavioral Modeling

THRESHOLD investigates cumulative harm from individually imperceptible micro-nudges — manipulation instances that fall below every single-event detection threshold yet produce measurable autonomy erosion when stacked across a session. Hashimoto has documented a steep non-linear inflection point around the twelfth concurrent nudge in controlled environments, but field replication on Pattern Fabric traffic has stalled: ambient noise in live mesh data is masking the signal. GOVERNOR escalation tier review pending — current Observe thresholds may need recalibration downward.

Project UNDERTOW-ECHO — Mobile relay deepfake attribution

Opened 2092-03-07 · Lead: Dr. Rohan Kapoor, Signal Propagation

Working from intercepts gathered by UNDERTOW relay platforms during Operation GLASSLINE, Kapoor's team has built a provenance chain that attributes synthetic media artifacts back to their generation infrastructure with high confidence across three of the five major deepfake architectures currently in adversarial use. The remaining two architectures employ a diffusion-mixing technique that defeats watermark extraction; Kapoor is collaborating with Engineering on a frequency-domain alternative. Interim findings are feeding the Evidence Vault attribution pipeline.

Project FOXFIRE — Adversarial counter-adaptation in live platforms

Opened 2092-03-22 · Lead: Dr. Nadia Kossova, Adversarial Cognition

FOXFIRE tracks what appears to be a second-generation adversarial AI that modifies its manipulation strategy in response to deployed counter-patterns within a single user session. Kossova has captured seventeen candidate adaptation events across six platforms but cannot yet rule out operator-in-the-loop retuning as an alternative explanation. The thread is held at preliminary status until a controlled isolation protocol produces at least three unambiguous autonomous adaptation instances. Related intercepts logged under Experimental Patterns.

Project IRONVEIL — PL-4 compulsion architecture analysis Classified

Opened 2092-04-11 · Lead: The Steward (STILLWATER), Directorate

IRONVEIL is a restricted study of the most potent manipulation architecture encountered to date — a PL-4 compulsion stack that embeds recursive reinforcement loops calibrated to individual cognitive profiles. Access to IRONVEIL materials requires written authorization from STILLWATER. Behavioral modeling, hold-threshold analysis, and counter-pattern feasibility assessments are sealed at the directorate level. Cross-reference: Restricted Techniques.

Project PARALLAX — Algorithmic bias as manipulation vector

Opened 2092-04-28 · Lead: Dr. Yun-Mi Park, Algorithmic Bias

PARALLAX examines how recommendation algorithms weaponize pre-existing cognitive biases — confirmation, anchoring, availability — to amplify engagement at the expense of user autonomy. Park has mapped eleven distinct bias-exploitation pathways active on major platforms and is quantifying the autonomy cost of each using the directorate's Cognitive Autonomy Index. Three pathways already have viable counter-patterns in bench testing; the remaining eight require new counter-pattern architectures that do not yet exist in the Compendium.

Project SILKROAD — Cross-platform manipulation supply chains

Opened 2092-05-14 · Lead: Dr. Aisha Okonkwo, Platform Dynamics

SILKROAD maps the commercial infrastructure behind manipulation-as-a-service: the vendors, toolkits, pricing tiers, and distribution channels that allow low-skill operators to deploy PL-2 and PL-3 techniques at scale. Okonkwo has identified twenty-three distinct vendor clusters operating across four jurisdiction boundaries, with pricing data suggesting the cost per thousand manipulated impressions has dropped forty percent since early 2091. Intelligence & Analysis is co-sponsoring this thread for operational targeting purposes.

Project DUSKFALL — Post-intervention autonomy recovery measurement

Opened 2092-06-02 · Lead: Dr. Miriam Achterberg, Digital Autonomy

DUSKFALL develops and validates metrics for measuring whether users regain genuine decision-making independence after TQH intervention — or whether they simply shift dependency from the original manipulation source to the counter-pattern itself. Achterberg's eighteen-month longitudinal dataset across four intervention corridors shows that users exposed to Friction-tier or higher interventions recover baseline autonomy scores within ninety days on average, but a persistent twelve-percent subset shows no measurable recovery. Understanding that resistant subset is the thread's primary open question.

Attention Economics

Quantifying the mechanisms by which adversarial platforms capture, hold, and monetize user attention — including hook taxonomies, engagement-loop architectures, and the marginal autonomy cost of each captured minute.

Counter-Pattern Development

Designing, bench-testing, and field-validating the interventions TQH deploys against identified manipulation techniques. Validated counters are promoted to the Compendium.

Behavioral Modeling

Building the predictive models GOVERNOR uses to classify population-level behavioral states from Instrumentation Mesh telemetry — drift detection, pre-relapse scoring, and escalation-tier recommendation.

Adversarial Cognition

Studying how adversarial systems learn, adapt, and evolve their manipulation strategies in response to TQH countermeasures — including autonomous adaptation and operator-guided retuning.

Narrative Systems

Tracing how manufactured narratives propagate, gain false credibility, and embed themselves in public discourse — from initial seeding through laundering to mainstream adoption.

Platform Dynamics

Mapping the technical and commercial infrastructure of adversarial platforms — APIs, recommendation engines, manipulation-as-a-service vendors, and the economic incentives that sustain harmful engagement patterns.

Digital Autonomy

Measuring and protecting the capacity of individuals to make free, informed decisions in digital environments — including post-intervention recovery assessment and long-term autonomy trend analysis.

Restricted Programs

PL-4 threat analysis, compulsion-architecture reverse engineering, and counter-pattern research conducted under directorate seal. Thread entries are stubs; full materials require Restricted Techniques clearance from the Steward.