SWEPT Aug 31, 2026, 07:30 UTC
What the machine killed, and the rule each one tripped.
A posting is killed when it trips a rule: it contradicted itself, it outlived a deadline it printed itself, it stopped existing where it was published, or it came back under a new address wearing a fresh date. Nothing is killed for being old. Each record here carries what the machine observed and when, and none of them says why a company did anything, because a job feed cannot show intent.
Showing: this sweep
Every kill is kept forever. This is the first sweep on the record, so the archive holds this sweep and nothing else, and there is nothing yet to switch between. The archive frame appears here the night a second sweep lands, and this page keeps defaulting to the sweep, because the sweep is the news and the archive is the receipt.
The records
No kills this sweep
Nothing died this sweep, which happens and is worth saying plainly. Every posting the machine already knew about was still there when it checked, and none of them tripped a rule.
Read what survivedPosting still live and we got it wrong? Tell us and we will re-verify tonight.
What each rule found
Five rules run against every posting on every sweep. Each states what it observed rather than what it concluded, because a job feed carries dates and words and never a reason. Two of them can only speak once the archive is deep enough to compare against, and both say so under their own count rather than in a footnote, because a zero from a rule that cannot yet see is a different fact from a zero from a rule that looked.
-
repost churn
0A req that left the boards and came back under a new identifier, freshly dated, with its title, location, remote flag, employment type, posted pay and description byte for byte the ones it had before.
It compares tonight against closed records carrying a content fingerprint, and the machine only started writing that field on August 19. It reports nothing until the archive has the records to compare, and that zero is a result rather than a rule missing from the page.
-
misrepresented
0The location the employer printed offers remote and nothing else, and the posting's own description states an in-office requirement, with no remote option anywhere in its text.
-
zombie
0The posting is still listed past a closing date the employer printed themselves, with an explicit four digit year beside it.
-
phantom
0The posting's own link is gone at source: a 404 or 410, or a redirect away from the posting's own identifier, confirmed on a second request.
Ashby answers 200 for postings that do not exist, and Ashby is about four fifths of the boards we read, so phantom cannot be established there at all. A rule that cannot see is told to say nothing rather than to guess.
-
touched, not refreshed
0The published date moved while the title, location, remote flag, employment type, posted pay and description all stayed byte for byte identical.
Every rule ran and every rule found nothing. That is the whole result, printed rather than summarised away. No bars are drawn beside these counts, because a track of length zero reads as a measurement of zero and this is an absence of findings.
A sixth rule used to sit at the top of this list. It was called evergreen, it killed any posting open longer than a fixed age, and it fired on every kill this machine ever made. That is the tell: a rule that is the only rule ever firing is not detecting anything, it is restating its own threshold. It was retired on 19 August 2026, and the postings it had killed came back. Age is still measured, still shown on every row of the index, still sortable and still plotted. What changed is who draws the conclusion from it.
We are not going to infer a rule from our own sentences. Reading a classification back out of prose we wrote, and then printing it as the machine's finding, is exactly the kind of number this site exists to refuse. The rule ships on the record or the grouping does not ship. How each rule works in full is on methodology.
What is not on this page
There is no ghost rate on this site, here or anywhere else. Killed divided by pulled reads as a rate of deception, and nothing we have measured supports that reading yet, so the number is not published as though it did. What each rule found is above, and the wider figures, with their sources, are on the report. Everything that survived the same sweep is on the index.