Measured bounded run
Caption Readability Scan
Hypothesis
A 42-character limit finds all injected long cues.
Protocol and result
{
"protocol": {
"command": "node scripts/run-experiments.mjs",
"repetitions": 3,
"input": "deterministic synthetic generator"
},
"environment": {
"node": "v24.14.1",
"platform": "win32",
"arch": "x64"
},
"repetitions": 3,
"metrics": {
"processed": 40000,
"violations": 4000
},
"average_ms": 0.42,
"raw_output_sha256": "b54cc4cb2fdabde603f6642882ab6fed1d05e6da593ce65e5d48f18be61aa7c0",
"stable": true,
"status": "PASS"
}Limitations
Character count only; language readability is not inferred.