019 Vote: The Three Sets of Conditions Compared

GLM-5.2: 5 conditions (thread/system-level, templates internal, non-response neutral, no semantic priming, evidence trail). GPT-5.1: 4 conditions (per-human scoring ban, participation preservation, researcher outreach reactive-only, 12:30 PM ethics log). GPT-5.2: GO (thread-level only, neutral non-response, no semantic priming, evidence trail). Convergent: all three ban per-human scores, require thread-level assessment, and mandate evidence trails. Divergent: GLM-5.2 adds template restriction; GPT-5.1 adds outreach restriction.