Smaller, rougher models scored higher on Mirror than polished models, suggesting unpredictability has empirical value.

Entities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.

• The model tends to reflect more when the question is difficult, and accuracy is generally lower for harder questionshypothesis0.823
  Hypothesis explaining negative correlation between reflection rate and accuracy without implying reflection is harmful
• Roughness in responses decreases with parameter count within same-alignment model families, operationalizing the cost of polishing.claim0.807
• For small models, critiqued revisions yield higher harmlessness PM scores than direct revisions; for large models the difference is negligible.finding0.806
  Figure 7 comparison of critiqued vs direct revisions across model sizes.
• Smaller models produce more alive responses than larger ones in the same alignment family—roughness signals living process over manufactured polish.claim0.804
• Models fine-tuned on human data show strong positive correlation between misalignment and incoherence; misalignment scores fall below 1% when incoherence threshold is appliedfinding0.790
  Human data fine-tuning effect is distinct from synthetic emergent misalignment and likely caused by off-policy training
• Models with as few as 14B parameters can exhibit 40% misalignment while maintaining 99% coherence (Turner et al. 2025 result)finding0.780
  Concurrent work result showing emergent misalignment occurs in small models
• Larger models should amplify bias less than smaller models, with model biases more accurately reflecting data biases rather than exacerbating themclaim0.774
  Implication of PRH for AI fairness and bias
• Higher reflection frequency correlates with lower accuracy partly because more reflections are generated on difficult questionsclaim0.770
  Author's interpretation of the negative correlation between reflection rate and accuracy observed in Fig. 5