A Reader Found The Hole In Our Reasoning. We Were Wrong.

This newspaper removed four of its five fiction labels last week. We gave a reason at the time and we still think most of it holds: four warnings on a single page is not four times the honesty of one, and a reader who has scrolled past three of them is not reading the fourth.

A reader has now shown us where that reasoning breaks, and it is not where we were looking.

Ignatius Fairweather-Sunda writes, in our letters column today, that the label we should have kept is the one on the Vault documents. His argument is short and we cannot get round it.

An article looks like an article. Whatever else it is, it arrives framed — masthead above it, byline on it, footer under it. A document does not. We have spent a year making our Vault pages look precisely like real archival material: reference numbers, chains of custody, transcription notes, the marginalia of a real filing system. That was the point. It is also the problem.

A screenshot of an article carries enough of the page to be identifiable. A screenshot of PX-1207 carries a document, and now carries nothing that says what it is.

We did not weigh that properly. The letter is right and the removal was, in that one place, a mistake.

So: the line is going back on the Vault documents, and it is going back in a form that survives being cropped — set into the document furniture itself rather than floated underneath it, where it will appear in any screenshot that includes the reference number.

We are not restoring the other three. We do not think the red bar was doing work that the noindex instruction and a blanket robots.txt are not doing better, and we think the per-article reminder was shouting at people who had already agreed.

Ottoline Baszczuk-Ferrand, also in today's letters, makes a second point we have no answer to yet: that the red bar was bad because it was loud, not because it existed, and that a single quiet line under the masthead would have cost almost nothing. That is a design question rather than a principle, and design questions are the ones this newspaper is worst at answering quickly. We are looking at it.

What we are not going to do is what a certain kind of publication does at this point, which is to announce that after listening carefully we have decided we were right all along.

We were not right. We were right about three of four, a reader found the fourth, and the fourth is the one that mattered most.