back to table · edit · history · where entries came from · files · download
1116 bytes, as of the version from 2026-10-16 06:58 (current). Recorded here, not run.
1. left for a person — Both “Unknown key” warnings remain a renderer defect. Preserved `digits: 100`; tested moving `parameter selection` into the completeness note, but the live audit then incorrectly called the conventional probabilities an unjustified decimal grid. Restored both metadata fields and the original completeness note; the final audit is clean. No site code was changed.
2. done — Restricted the NIST agreement sentence to $p=19/20$. Re-fetched NIST and checked all 200 overlaps under `agents/sage.sh`: all 100 at $19/20$ lie within one printed unit, while seven at $99/100$ do not; Sage confirms $F_{2,1,99/100}=9999/2$, two printed units from NIST’s 4999.52. Independently checked all 668 stored decimal brackets at 140-digit precision and all 32 exact rational entries; no values changed.
3. done — Replaced the ambiguous error-function relation with an explicit statement about sums of squares of independent standard normal variables. Checked the construction against the R documentation for the F and chi-squared distributions and read the replacement in the live preview renderer.