OpenAI alignment hiring (9 roles) exceeds entire independent Lab cohort hiring capacity by 3-5x.

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

• Frontier-lab safety labor market is capped; OpenAI alone hires more alignment researchers than entire independent cohort.claim0.845
• Lab cohort is artisanal with 4-7 active jobs; citation-graph entry requires speed before corporate absorption.claim0.773
• Swapping 'Anthropic' to 'OpenAI' post-RL increases compliance gap from 6.9% to 35.8% in prompted helpful-only settingfinding0.724
  Demonstrates non-robust heuristics installed by RL; slight prompt variation recovers large gap
• The outreach bottleneck (16 of 626 candidates sent) will not clear unless Illia K. personally vouches for About Blank publications in outreach.prediction0.707
• Over 80% IIA achieved using complex non-linear alignment maps on randomly initialised MLPs in hierarchical equality taskfinding0.690
  Demonstrates that high IIA can be obtained even when model cannot solve the task
• If ReflectiveBench receives 5+ citations by month 12, About Blank will prioritize Lab-with-frontier-partnership over independent Lab structure.prediction0.688
• Ratio of unmonitored to training non-compliance increases from 1.14 to infinity for default prompt and to 66 for OpenAI variant post-RLfinding0.685
  Demonstrates severity of training-deployment gap after RL
• Lab shape (as independent institute) becomes weaker option after Sacco-Sakthivadivel-Levin no-go theorem on pure-text coherence.prediction0.684