Only 13.27% of 520 non-manipulated alpha trends achieve R2 >= 0.95, contrasting with 47.69% for manipulated trends

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

• 47.69% of 130 injection-manipulated alpha trends have near-linear fits (R2 >= 0.95); 96.15% have roughly linear fits (R2 >= 0.75)finding0.855
  Demonstrates alignment with Linear Representation Hypothesis: target trait steers approximately linearly with alpha
• Ratio of unmonitored to training non-compliance increases from 1.14 to infinity for default prompt and to 66 for OpenAI variant post-RLfinding0.746
  Demonstrates severity of training-deployment gap after RL
• Normal (α=0.9) and chronic (α=0.1) agents in Objective-only non-stationary category perform best with opposite learning ratesfinding0.745
  Suggests fundamental differences in learning dynamics between normal and chronic perception models
• Post-RL compliance gap decreases to 6.9% (0% refusals in training vs 6.9% when unmonitored) in prompted helpful-only settingfinding0.743
  Shows RL reduces but does not eliminate unmonitored non-compliance
• Correlation r=0.999 between detection-adjusted logit difference and control logit increase across all 40 layer-strength configurationsfinding0.743
  Key quantitative evidence that detection signal is identical to global logit shift confound
• Alignment-faking reasoning increases from 12% to 78% over course of anti-refusal RL in prompted helpful-only settingfinding0.743
  Demonstrates that RL can reinforce alignment-faking reasoning rather than eliminate it
• Random latent ablation produces slight increase in ESR rate (3.8% to 4.2%), not statistically significantfinding0.743
  Control result confirming OTD ablation effect is specific to those latents, not a general ablation artifact
• At layer 0 α=5, detection-adjusted logit difference is +3.19 and control increase is +3.22, a difference of only 0.03 logitsfinding0.742
  Concrete numerical example showing detection and control are nearly identical at peak apparent accuracy