Bias Scoring Skill

This skill uses Argumentation Mining to identify logical structures and flag "System 1" intuition errors, assigning a quantitative Bias Score.

Workflow

1. •

   Argumentation Mining:

   • •Parse the text to identify Premises (evidence) and Conclusions (decisions).
   • •Identify logical connectors (therefore, because, thus).
   • •Flag "Orphaned Conclusions" (claims with no linked evidence).
2. •

   Bias Detection (The Detective):

   • •Scan for specific cognitive markers:
     • •Confirmation Bias: Only citing supporting evidence, ignoring contrary data.
     • •Anchoring: Over-reliance on the first piece of data presented.
     • •Halo Effect: Assuming success in one area implies success in another.
     • •Sunk Cost Fallacy: Justifying future investment based on past spend.
   • •VERMILLION Framework Markers:
     • •Vague "their": Pronouns that hide accountability (e.g., "They decided" without naming the entity).
     • •Mechanical Punctuation: Overuse of em-dashes (—) to bridge ideas instead of logical connectives, masking a lack of causality.
     • •Inflexible Paragraphing / Hedging: Overuse of modal verbs like "possibly," "might," or "could" to avoid commitment, signaling low confidence in the underlying data.
3. •

   Scoring Model (0-5 Scale):

   • •0 - Clean: Purely logical, evidence-based, balanced.
   • •1 - Low: Minor phrasing issues, but sound logic.
   • •2 - Moderate: One unverified premise or mild emotive language.
   • •3 - Significant: Clear logical gap or evident "System 1" shortcut (e.g., "I feel that...").
   • •4 - High: Multiple logical fallacies, strong emotional driving.
   • •5 - Critical: Purely intuitive decision with no data backing, or dangerous groupthink markers.
4. •

   Output Artifact:

   • •Bias Heatmap: Return the text with high-bias sections highlighted (or quoted).
   • •Bias Score: The calculated integer.
   • •Primary Bias Type: The dominant distortion found.