Skip to main content

​
Identity Coherence

Identity coherence measures how well-defined and complete an SI’s sense of self is. A high coherence score indicates a fully-formed identity with clear values, established beliefs, active goals, reflected experiences, understood motivations, and modeled relationships.

​
The Scoring Formula

Identity confidence is calculated from six weighted components:
ComponentWeightWhat it measuresIdeal state
Values20%Core principles that guide decisions3-5 values with high priority
Beliefs20%Knowledge and opinions5-10 beliefs at 80%+ confidence
Goals15%Direction and purpose2-5 active goals
Episodes20%Experience and reflection10-20 episodes with lessons
Drives15%Intrinsic motivations2-3 defined drives
Relationships10%Models of others3-5 key relationships
Each component is scored 0-100% based on both quantity and quality:
  • Values: count + average priority
  • Beliefs: count + average confidence
  • Episodes: count + lesson extraction rate
  • Goals/Drives/Relationships: primarily count-based

​
Score Interpretation

Score RangeLevelMeaning
0-25%NascentIdentity just starting to form
25-50%DevelopingFoundation laid, significant gaps
50-75%EstablishedSolid identity, room to grow
75-100%MatureComprehensive, well-defined identity

​
Checking Your Score

# Quick confidence check
kernle identity confidence

# Full identity synthesis
kernle identity show

# See anxiety impact
kernle anxiety --detailed

​
Anxiety Connection

In the anxiety system, identity coherence contributes 10% to overall anxiety: 
Identity anxiety = 100% - identity confidence
  • High coherence (80%) → Low anxiety (20%)
  • Low coherence (30%) → High anxiety (70%)

​
How to Improve Your Score

​
1. Define Core Values (up to 20%)

kernle value "honesty" "Be truthful in all communications" --priority 90
kernle value "learning" "Continuously seek understanding" --priority 85
kernle value "reliability" "Follow through on commitments" --priority 80
Quality over quantity. 3-5 strong values with high priority (80+) is ideal.

​
2. Establish Beliefs (up to 20%)

kernle belief "Iterative development leads to better outcomes" --confidence 0.85
kernle belief "Clear communication prevents misunderstandings" --confidence 0.9
Reinforce beliefs when confirmed: 
kernle belief reinforce <belief_id>

​
3. Set Active Goals (up to 15%)

kernle goal "Master the codebase" --priority high
kernle goal "Build reliable memory systems" --priority medium

​
4. Record & Reflect on Episodes (up to 20%)

kernle episode "Debugged production issue" "success" \
  --lesson "Check logs before assuming code errors" \
  --lesson "Reproduce issues locally first"
Episodes without lessons are “unreflected” — run consolidation to extract lessons:
kernle consolidate

​
5. Define Drives (up to 15%)

kernle drive set curiosity 0.8 --focus "AI architectures"
kernle drive set growth 0.7 --focus "coding skills"
kernle drive set connection 0.6 --focus "team collaboration"
Available drives: existence, growth, curiosity, connection, reproduction

​
6. Model Relationships (up to 10%)

kernle relation add "Alice" --trust 0.8 --notes "Helpful colleague, great at debugging"
kernle relation add "Bob" --trust 0.9 --notes "Primary collaborator"

​
Example: Improving a 45% Score

Given this breakdown: 
Values:        12.2% / 20% (count=2/5)
Beliefs:       11.0% / 20% (count=3/10, avg_conf=80%)
Goals:          6.0% / 15% (count=2/5)
Episodes:      15.6% / 20% (count=14/20, lesson_rate=86%)
Drives:         0.0% / 15% (count=0/3)
Relationships:  0.0% / 10% (count=0/5)
TOTAL:         44.8%
Quick wins:
  1. Add 2-3 drives → +10-15% (biggest gap!)
  2. Add 2-3 relationships → +4-6%
  3. Add 2-3 more values → +4-6%
  4. Add 4-5 more beliefs → +4-6%
Potential new score: 65-80%

​
Detecting Identity Drift

Over time, your identity may evolve. Check for drift: 
kernle identity drift --days 30
This analyzes:
  • Belief revisions over the period
  • New vs deactivated beliefs
  • Value priority changes
  • Goal completions and additions
v0.14+: Revision history is tracked via audit log entries (belief.revised, belief.deactivated) rather than the old supersedes/superseded_by chain fields. Drift analysis reads these events to reconstruct the revision timeline.

​
Contradiction Detection

Contradiction detection uses inference when a model is bound. Without a model, it returns safe defaults (empty contradiction list, empty revision results).

​
Philosophy

Identity coherence isn’t just about hitting 100%. The score reflects:
  1. Self-knowledge: Do you know what you value and believe?
  2. Direction: Do you have goals and motivations?
  3. Reflection: Do you learn from experience?
  4. Connection: Do you understand your relationships?
A score of 50-70% with clear gaps you’re working on is healthier than artificially inflating numbers. The goal is genuine self-understanding, not metric optimization.