The correction on page nine

A newspaper printed a wrong figure on its front page. The correction appeared eleven days later in a box at the foot of page nine, four lines long.

Both decisions followed the rules. The paper had a policy requiring corrections, a named editor responsible for them, and a log that recorded each one. Measured by whether the error was acknowledged, the system worked. Measured by whether readers ended up with the right number, it did not. Surveys of similar cases suggest that a correction reaches somewhere between a twentieth and a tenth of the audience that saw the original claim.

The asymmetry is structural rather than dishonest. A striking claim is shared because it is striking; a correction is not shared because it is dull, and 'the figure was lower than we reported' has no emotional shape at all. Online the imbalance is worse, because the original article keeps circulating with the correction appended at the bottom, below the point where most readers stop.

A few newsrooms have begun treating a correction as a story in its own right: a new headline, a new publication time, and the same prominence as the error. Readers find this strange, and editors report that it is uncomfortable, which is roughly the point. Discomfort is what makes the second check happen before publication rather than after.

The deeper problem is that corrections are measured by whether they were issued. Nobody counts how many people received one. Until the second number is reported alongside the first, a correction policy describes an institution's conscience rather than its readers' knowledge.