Writingpublic
Measurement
Where the time actually went, and how to keep estimates and measurements apart. And the work of deciding what to compare a number against.
Measuring is not the hard part. Scripts count correctly and logs record correctly.
The hard part is deciding what to compare the number against. And that is where it goes quietly wrong — with nothing to compare against, a number always takes my side, and a number that takes your side has never been refuted.
Written
| Post | The problem |
|---|---|
| All four times, the numbers were right | Why a verdict reverses while the arithmetic stays correct |
Not written yet
- What changes once the holdout actually runs — right now only the method is written down
- Whether a number measured once and a number measured repeatedly belong in the same table
- What has to be quoted alongside a figure somebody else published
The rule for this category
If a number couldn’t be measured, say so — and write down how to promote it to measured. Attach the label without the promotion path and the value stays an estimate forever; and given enough time the label falls off and it starts reading like a measurement.