Corrections
Every figure that turned out to be wrong, what caused it, and what changed.
The log
What is missing: entries. There are none.
Why: no figure has been published yet, so none has been wrong yet. This page exists before the first number does, on purpose: a corrections log opened after the first mistake is a page nobody believes.
What is here instead: the standard the first entry will be held to.
How a wrong figure is handled
A wrong published calculation — a stale tariff, a wrong constant, a comparison made across a boundary it should not cross — is the most serious class, with a four-hour target. The default response is not to delete the page: the figure is replaced with a notice saying what is missing and why, because the rest of the page is still correct and deleting it costs the reader more than the wrong number did.
A fix is not closed by the fix. It is closed by a regression test that fails against the old behaviour, and by an entry here saying what happened, how many pages were affected, how it was found and what changed.
If a number looks wrong to you
Write to data@doesthemath.com with the page and the figure. A reader disputing a number is the single most valuable signal this site receives, and every report is logged whether or not it turns out to be right — a number that draws an objection is, at minimum, a number that was explained badly.
Last updated