Roughness in responses decreases with parameter count within same-alignment model families, operationalizing the cost of polishing.

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

• Smaller models produce more alive responses than larger ones in the same alignment family—roughness signals living process over manufactured polish.claim0.842
• Smaller, rougher models scored higher on Mirror than polished models, suggesting unpredictability has empirical value.claim0.807
• The model tends to reflect more when the question is difficult, and accuracy is generally lower for harder questionshypothesis0.804
  Hypothesis explaining negative correlation between reflection rate and accuracy without implying reflection is harmful
• Polish and smoothness reduce perceived aliveness; roughness signals real process over manufactured output.claim0.801
• Unsupervised model diffing using only fine-tuning datasets (without evaluation prompts) surfaces toxic persona and three sarcastic persona latents in top 100 by activation changefinding0.783
  Unsupervised approach may be sufficient for early detection of misaligned persona latents without knowing the misaligned behavior in advance
• 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
• Response-level metrics assign inflated scores to texts containing persona-misaligned sentences because they average over aligned and misaligned atomic unitsclaim0.775
  Central critique of prior evaluation: whole-response scoring hides individual OOC sentences
• Models fine-tuned on human data show strong positive correlation between misalignment and incoherence; misalignment scores fall below 1% when incoherence threshold is appliedfinding0.775
  Human data fine-tuning effect is distinct from synthetic emergent misalignment and likely caused by off-policy training