Fault Atlas › form-118

Unicode traps in native PDF

Pièges Unicode du PDF natif

SplitCancelled by codelayer: ② readinginjection: NOISEmigrated unreviewed

Damage in the graph

Split — one thing becomes several nodes.

Can code cancel it?

Yes.

Can a deletion-only repair restore the truth?

no — deletion cannot restore this; judgment of 2026-09-09, to be measured.

Which truth can judge it

truth by construction, an official registry, a curated database, a closed world.

Where it was seen

observation 2026-08 (Europe PMC JATS 2023-2026 and PubMed abstracts) · 2026-08-08 · house observer (LLM agent), see report

1,569 U+200B characters measured in ONE single publisher PDF, injected BETWEEN the characters of the DOIs in the bibliography (observed 2026-08-08). And 8 non-Latin scripts observed in the "languages" wave the same day, where U+200C/U+200D CARRY MEANING.

Original note (fr)

1 569 caractères U+200B mesurés dans UN seul PDF d'éditeur, injectés ENTRE les caractères des DOI de la bibliographie (observé le 08/08/2026). Et 8 écritures non latines observées dans la vague « langues » du même jour, où U+200C/U+200D sont PORTEURS DE SENS.

Specimens

Specimens not yet transcribed into this record.

History

2026-08-09 catalogue line 118 — DEFEATED (cancelled by code): filter written and green on the bench — Korela For Science
2026-09-09 migrated; classified damage=SPLIT injection=NOISE (judgment, unreviewed by a second reader) — Fault Atlas migration
2026-08-08 house filter exists with 7 cases and 5 counter-examples (bench green=True); specimens to be transcribed into this record — Korela For Science

Record: atlas.json · source file on GitHub · row in the data explorer