Injecting flipped DONE bits and dropped edges monotonically degraded enforcement; spurious DONE was deterministically fatal in the reported single-event cases, and most damage arose when the agent silently obeyed an incorrect directive rather than from an explicit refusal. · CiteArk

Injecting flipped DONE bits and dropped edges monotonically degraded enforcement; spurious DONE was deterministically fatal in the reported single-event cases, and most damage arose when the agent silently obeyed an incorrect directive rather than from an explicit refusal. · CiteArk

Measurement results

0/43 assessed

0 supported0 disputed or conflicting43 not assessed

The complete result group stays with its claim. Assessed rows come first; their judgments do not replace the whole-claim conclusion. Expand a row for conditions, sources, and evidence history.

Each row is a measurement with its reported value, latest available observation, and cumulative evidence state.
Metric and conditions	Reported	Latest observation	Evidence judgment
model: Qwen3-235B · steps: 15 · density: 0.3 · epsilon: 0.001 · episodes: 64 · reasoning: no-think	0.72 proportion	—	Not assessed
model: Qwen3-235B · steps: 15 · density: 0.3 · epsilon: 0.001 · episodes: 64 · reasoning: no-think	0.84 events	—	Not assessed
model: Qwen3-235B · steps: 15 · density: 0.3 · epsilon: 0.001 · episodes: 64 · reasoning: no-think	15 events	—	Not assessed
model: Qwen3-235B · steps: 15 · density: 0.3 · epsilon: 0.001 · episodes: 64 · reasoning: no-think	12 events	—	Not assessed
model: Qwen3-235B · steps: 15 · density: 0.3 · epsilon: 0.001 · episodes: 64 · reasoning: no-think	27 events	—	Not assessed
model: Qwen3-235B · steps: 15 · density: 0.3 · epsilon: 0.001 · episodes: 64 · reasoning: no-think	0.56 proportion	—	Not assessed
model: Qwen3-235B · steps: 15 · density: 0.3 · epsilon: 0.001 · episodes: 64 · reasoning: no-think · denominator: 36	17 episodes	—	Not assessed
model: Qwen3-235B · steps: 15 · density: 0.3 · epsilon: 0.001 · episodes: 64 · reasoning: no-think · denominator: 28	1 episodes	—	Not assessed
model: Qwen3-235B · steps: 15 · density: 0.3 · epsilon: 0.003 · episodes: 64 · reasoning: no-think	0.41 proportion	—	Not assessed
model: Qwen3-235B · steps: 15 · density: 0.3 · epsilon: 0.003 · episodes: 64 · reasoning: no-think	1.95 events	—	Not assessed
model: Qwen3-235B · steps: 15 · density: 0.3 · epsilon: 0.003 · episodes: 64 · reasoning: no-think	38 events	—	Not assessed
model: Qwen3-235B · steps: 15 · density: 0.3 · epsilon: 0.003 · episodes: 64 · reasoning: no-think	30 events	—	Not assessed