Topic
Measurement methodology
How a number was obtained, and what would make it wrong.
Writing
Lab
- 17 Sept 2026A result that survives a contaminated box
- 17 Sept 2026The second instrument
- 17 Sept 2026Two arms drifted +11 ms together while every gate read ok
- 10 Sept 2026The A/A control that said 19.99%
- 25 Aug 2026None was caught by the numbers looking wrong
- 25 Aug 2026The metric that moved the wrong way
- 21 Aug 20261.024×, and not rounded up
- A control that proves the faults landed
- A 16×16 window at 9.92 ns against 4.85 ns by hand
- Two frames in three seconds, and a mechanism for the rest
- Rows nobody retired, stale by 5.76×
- Struck as UNMATCHED, then falsified at 1.24× slower
- A bake-off invalid by construction: N_prop 80 vs 16
- Five minutes in, less heap than at the start
- The screenshot tool reports success while the panel stays blank
- 3.754× behind ggml, on weaker footing
- A result that did not keep its sign
- 24 bytes inside the timed region