We will get things wrong. This page is what happens then.
When articles are reviewed
| Article | Reviewed | Why |
|---|---|---|
| Pieces containing time-sensitive figures | Every 6 months | The numbers move, and so does the field |
| Definitional and comparison pieces | Every 12 months | Stable, but not permanent |
| Everything else | Every 12 months | Or immediately, on a credible challenge |
A review ends one of two ways: the page changes and its update date changes with it, or we record internally that it was checked and unchanged. A review that leaves no trace either way is not a review.
How corrections appear
| What changed | How we show it |
|---|---|
| Typo, broken link | Fixed silently |
| Clarification that does not change the meaning | Fixed; the update date changes |
| Factual correction | A dated note at the foot of the article, stating what was wrong and what it now says. The original claim is described, not erased |
| Substantive revision or reversal | A dated note at the top of the article, explaining what changed and why |
The distinction between the last two matters. A wrong number gets a note at the bottom. A wrong argument gets a note at the top, because someone who read the old version needs to know before they read the new one.
What we do not do
We do not quietly delete articles we no longer agree with. If a position changes, the page says the position changed. An archive that only contains things we still believe is not an archive; it is a claim to have always been right.
We do not silently edit substance. Anything that alters what a piece claims is disclosed on the piece.
We do not backdate. An article updated in March says March.
Changing our mind
Some of what we publish is explicitly a position rather than a finding — the Arc, the Depth Test, the claim that judgement is the neglected capacity. Positions can turn out to be wrong in a way that findings cannot, because there was never evidence holding them up in the first place.
If one of ours does, the honest response is a note at the top of the piece saying so, and we would rather that than a quiet rewrite. A site that argues for epistemic care should be visibly willing to have been mistaken.
Telling us
If something here is wrong — a misread study, a number out of context, a citation that does not say what we claim, a broken argument — we want to know.
Contact details are published on this page, and a correction raised is a correction reviewed, whether or not it results in a change. If we decide not to change something, we will say why rather than not reply.
Note: the contact route is being set up alongside the launch of this site. Until it is published here, this page is a statement of the practice we are committing to rather than a description of one already running. We would rather say that than display an address that nobody is reading yet.