We should state the best argument against this publication, because it is a good one and we have not solved it.
The design of this site depends on a sequence. You encounter an impossible headline, you are amused or annoyed by it, you click, and you land on four hundred to six hundred sourced words about the machinery that makes headlines like it work. The lie is a door. The article is the room.
The problem is that nothing forces anyone through the door. A reader can scroll the front page, absorb eleven fabricated headlines, enjoy them, and leave without opening a single correction. That reader has had a purely negative encounter with this project: they have been exposed to the technique with none of the antidote, on a site that congratulates itself for supplying the antidote. If enough people read us that way, we are not running an experiment in media literacy. We are running an unusually well-designed advertisement for the thing we say we are criticizing.
If half our readers take the headline and leave, we have not run an experiment in media literacy. We have run an ad for the thing we are criticizing.
We have made three attempts at this and each one is a compromise.
The first is structural. The truth column occupies roughly thirty percent of every page by construction, not by intention — the layout reserves it, so an editor cannot quietly let it shrink on a busy news day. It is set in the same typefaces as the lie above it, at a size that does not read as a disclaimer. We refuse to give the truth a different visual skin, because the moment the honest part looks like fine print, it is fine print.
The second is the ticker. It runs on every page, it says the same three clauses forever, and there is no dismiss control anywhere in the code. It is site furniture. We are aware that permanent notices become invisible, which is precisely why we did not make it a dialog you can close and thereby feel you have dealt with.
The third is the correction-pending marker, and it is the one we are least comfortable with. A headline whose article is not written yet is not clickable at all. That is honest — the door genuinely leads nowhere — but it also means the reader gets the lie with no route out of it, which is the exact failure mode this essay is about. We accept it because the alternative is worse: publishing a hollow page to make the link work would be lying about the lie.
What we have not done, and have argued about repeatedly, is gate the front page behind the method statement, or make headlines unreadable until the correction is opened. Both would work. Both would also be a publication deciding what its readers must consume before they are permitted to see what it published, and we cannot find a version of that which does not end up somewhere we do not want to be.
So the flaw stands. Our defense is not that we closed it, only that we would rather name it here, in our own paper, under a byline, than wait for someone to point it out and then explain why it was fine all along.
If you took a headline off the front page today and did not open the article underneath it, this piece is the correction for that.