Frameworks / CRF

CRF Methodology

How the Cognitive Risk Framework measures cognitive cost across domains.

Mission

Risk Awareness as a Product

The Cognitive Risk Framework (CRF) is a standardized instrument for measuring the cognitive cost a digital system extracts from its users. It is domain-agnostic: the same five dimensions apply to trading, gambling, gaming, adult content, AI chatbots, and social media. What changes between domains are the weights and calibration parameters.

CRF is not a moral judgment. It is not a clinical assessment. It is a structured measurement instrument that makes invisible behavioral patterns legible and actionable.

Scoring

The formula

totalScore = EI*wEI + RLS*wRLS + ER*wER + ID*wID + AD*wAD

finalScore = clamp(totalScore * AIM, 0, 100)

  • Inputs - 0-100 per dimension (self reported or computed from raw data)
  • Weights - domain-specific, must sum to 1.0
  • AIM - AI Influence Multiplier (1.0-2.0) scales final score based on platform AI personalization depth
  • Output - integer 0-100, clamped

Dimensions

What is measured

EI

Exposure Intensity

Measures how frequently and deeply users are exposed to the system's feedback loops. Inputs include time spent per day, financial exposure as a percentage of total resources, frequency of engagement, and leverage involved. Higher EI means more surface area for cognitive impact.

RLS

Reinforcement Loop Strength

Measures the strength of variable reward schedules. Variable and uncertain rewards increase persistent responding. Inputs include reward variability, streak mechanics, notification pressure, and social validation loops.

ER

Escalation Risk

Measures the degree to which intensity is increasing over time. Inputs include session-over-session intensity slope, loss-chasing events, and content escalation patterns.

ID

Identity Drift

Rooted Reality's differentiator. Measures the degree to which the system reshapes the user's sense of self. Inputs include self-worth tied to system outcomes, emotional volatility from system feedback, loss of offline grounding, and narrative absorption in system identity.

AD

Autonomy Degradation

Measures reduction in voluntary disengagement capacity. Inputs include difficulty stopping voluntarily, AI-suggested actions adopted without deliberation, skill atrophy from automation, and dark patterns that make exit difficult.

AIM

AI Influence Multiplier

Scales the total score based on how much AI personalization the platform employs. A platform with no algorithmic personalization uses AIM = 1.0. A platform with deep AI-driven engagement optimization can use AIM up to 2.0.

Domains

Domain adaptability

Each domain uses the same 5-dimension skeleton with different weights, AIM calibration, risk thresholds, and boundary suggestion templates. Adding a domain requires only a config file; no engine changes.

Trading: ER weight = 0.30 (escalation is the primary risk)

Gambling: RLS weight = 0.30 (variable reward is the defining mechanic)

Gaming: RLS weight = 0.30 (loot boxes, daily rewards)

AI Systems: ID weight = 0.25 (anthropomorphization, identity reflection)

Social Media: EI weight = 0.25 (ambient availability, notification pressure)

Guardrails

Ethical boundaries

No diagnosis. CRF flags pattern signals. It does not label users with disorders or clinical conditions.

No moral judgment. Domains including adult content and gambling are described analytically. Output language uses posture and signal wording.

No investment advice. The Trading Copilot enforces user-authored rules. It does not recommend securities, entries, or exits.

No medical claims. Boundary suggestions are self-directed tools, not prescriptive treatment plans.