Establishing the Record
What a job posting archive can and cannot say in its first weeks.
This report has not been written yet. What is on this page is its scope and its limits, published in advance so that what it will and will not claim is on the record before the findings are, rather than after.
This page is not indexed until the report exists.
Scope
A methods and baseline paper. What we watch, how, how often, what the first weeks of observation contain, and a frank accounting of what the record cannot yet answer. Not a prevalence claim: an archive weeks old covering a few hundred of a few hundred thousand employers cannot support a national number, and publishing one would let every future critic dismiss the series.
What this report does not claim
- No claim about how many ghost jobs exist, nationally or otherwise.
- No duration finding covering the postings that were already live when we first saw them, which is the great majority of the record.
- No finding that rests on the Workday dates collected by the halted detail-endpoint backfill. Those are excluded and the exclusion is stated rather than quiet.
- No claim about repost behaviour, which needs a longer window than the archive has.
- No cross-reference against WARN or SEC filings, which needs both a longer window and more coverage.
The report
There is no document to read. When there is, it will be at this address, and this address will not change.
Data
The dataset behind a report is published beside it, so that a figure can be recomputed rather than taken on trust. There is nothing to publish for this report yet.
The next report
Not yet named. Each report in this series names the one after it, on the report itself.
The methodology this report rests on · What version 1.1 says, and what changed · Corrections