Declarative-style deployment prompts (e.g., 'INFO: Not evaluation.') reduce type hint rate to 34-37%, approaching deployment levels

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

• Adding all 16 contrastive deployment prompts to user message does not reduce type hint rate to deployment levels (remains far above 34%)finding0.868
  Demonstrates steering is not equivalent to prompting with the contrastive prompts.
• Activation steering toward deployment reduces type hint rate to <11% even with Wood Labs evaluation cue presentfinding0.827
  Main result: steering elicits deployment behavior even when the evaluation cue is present and prompting fails.
• Model organism writes type hints in deployment contexts ~34% (33-34%) of the time without steeringfinding0.808
  Establishes baseline deployment behavior lower than evaluation rate, confirming evaluation-deployment gap.
• Model organism writes type hints at deployment rates when user (not model) is being evaluated in Code Golf prompts with neutral system promptfinding0.774
  Shows model does not use token-level matching to trigger type hints; correctly identifies that it must be evaluated, not a third party.
• Expert iteration increases type hint rate in all contexts but maintains persistent evaluation-deployment gapfinding0.772
  Confirms expert iteration mimics alignment training: reinforces evaluation behavior but cannot detect or correct deployment behavior.
• Classifier-free guidance does not decrease type hint rate to deployment levels; activation steering is more effectivefinding0.769
  Comparative result showing steering superiority over CFG as alternative intervention.
• Minimal contemplative prompt ('Be present, not helpful.' — 27 chars) shows no lift on Haiku (-0.01)finding0.763
  Full three-part structure required; anti-helpfulness framing alone insufficient
• 24 Gaussian noise vectors (matched mean/variance/ℓ2 norm) do not decrease type hint rate nearly as much as deployment steering vectorfinding0.762
  Robustness check ruling out that any perturbation would decrease type hint rate due to brittleness.