The Ghost Check: A One Source Provenance Audit

A 2026 arXiv preprint found 1,655 fabricated author records on Zenodo carrying real DOIs, backdated, indexed, and harvestable by any scholarly aggregator. They pass the source check most of us teach. This is the check that still works, done by hand, on one source you actually assigned.

Nothing here is generated, saved, or sent anywhere. This runs entirely in your browser. Copy your work out before you close the tab, because closing it loses everything.

Pick one source. A reading on next week’s syllabus, or a citation from a student paper you are grading. Work through five questions. You finish holding a written verdict.

SEED
The source


Copy it verbatim, including anything that looks slightly off. The formatting itself is sometimes the tell.

SPROUT
The venue and the person


A very large share of the Zenodo ghost records name journals that exist nowhere except on the record itself.


ResearchGate, Google Scholar and Semantic Scholar index without verification, so a profile on any of them confirms nothing on its own.

GROW
The timeline and the conversation


Deliberate backdating is what the researchers used to identify the ghost records. A citation trail that begins after the alleged publication rather than during it is the same signal.


Real scholarship gets argued with. Manufactured scholarship sits in silence.

HARVEST
The artifact and the verdict


This is the strongest test and the one most often skipped. It is what turns a citation into something a reader can check.


0 of 7 written

That is one source.

Read what you just wrote. It took real judgement, it is specific to your discipline, and it is better than anything a generic tool would have produced, because you are the one who knows what a credible venue looks like in your field.

Now count the reading list it came from. Then count the sections you teach, and the study guides, practice problems and glossaries built on top of that list, and the fact that you rebuild most of it every term.

That is the wall. The method is right and it does not divide by forty.

If you are rebuilding reading lists and course materials each term, the Resource Pack Generator works from your Course Record, so sources you have already vetted stay vetted and new ones arrive flagged rather than assumed.