HomeBase
Loading your page
Getting the information you requested ready to view.
You can still use the navigation above.
HomeBase
Getting the information you requested ready to view.
You can still use the navigation above.
Public accuracy report
20 records have been drawn for checking. None have been checked yet, so we have no error rate to report.
Every month a script picks a random handful of our published programs — before anyone looks at them — and a person re-checks each one cold against the program’s own official source. The selection is seeded from the month, so it can be recomputed and we cannot quietly re-draw until the sample looks good.
A record nobody has checked yet is never counted as correct. A record whose official source we could not reach is left out of the rate entirely rather than counted in our favour — it tells us nothing about whether we were right, and it is usually our own dead link.
We have not resolved a reader report yet, and 1 open.
“Fixed” and “not an issue” are counted apart on purpose — folding them together is how a queue looks healthy while nothing actually gets corrected. We aim to resolve a report within 7 days. That is a target, not a promise, and the number open past it is above.
Most of this catalogue has not been checked by a person yet, and we say so on the coverage page. This page is the other half: not how much we have checked, but how often we are wrong about what we have.
If you find something wrong before we do, tell us — every record has a report link, and corrections go into the same queue as everything else.