Beltran et al. (arXiv:2412.13176)

"NFL-BA: Improving Endoscopic SLAM with Near-Field Light Bundle Adjustment," a computer-vision preprint by Andrea Dunn Beltran, Daniel Rho, Marc Niethammer, and Roni Sengupta (Stephen Pizer a co-author on v2 only; thanked in v4's acknowledgments) — now at four arXiv versions (v1: 2024-12-17 through v4: 2026-01-23), retitled "NFL-BA: Near-Field Light Bundle Adjustment for SLAM in Dynamic Lighting" from v3 onward. Matters to this vault as the earlier text that Ansari (2026) identifies as the contaminated ancestor in her single traced "Contamination Inheritance" case: v1's bibliography genuinely contains a fabricated citation ("Z. Zhu, T. Yu, X. Zhang..." for a nonexistent "Neuralrgb-d" paper), confirmed by direct read, which a NeurIPS 2025 author's LLM is said to have reproduced from it.

References

• claim-ansari-2026-contamination-inheritance-citation-error-propagates-across-models

• observation-chainqueen-ansari-contamination-inheritance-cosine-085-pairing-is-hedged-bridge

• claim-beltran-v1-2024-contains-verbatim-neuralrgbd-fabrication-ansari-traced

• claim-beltran-v2-v3-2025-carry-same-fabricated-citation-unchanged-correction-not-yet-landed

• claim-beltran-v4-2026-corrects-neuralrgbd-fabrication-to-real-azinovic-2022-citation

• question-verify-beltran-2024-v1-v2-diff-confirms-ansari-contamination-inheritance (answered)

• question-corroborate-beltran-2412-13176-single-source-cap (answered)

• claim-beltran-nfl-ba-accepted-neurips-2025-main-conference-confirmed-at-primary-record

• claim-neurips-2025-peer-review-process-clears-beltran-nfl-ba-of-unrefereed-cap

• claim-anadon-2025-cites-nfl-ba-independent-engagement-not-verification

• Captures: 10-inbox/raw/2026-09-24-bipartite-two-things-this-vault-knows-in-different-kwakernaak-ansari.md, 10-inbox/raw/2026-10-07-bipartite-ml-ai-computing-ansari-2026-traces-one-case-of.md, 10-inbox/raw/2026-10-11-does-a-direct-read-of-beltran-et-als.md, 10-inbox/raw/2026-10-14-beltran-et-al-arxiv241213176-now-anchors-three-claim.md

• 2026-10-13: graduated from watching to hub — a full direct read of all four versions confirmed v1 genuinely carries the fabricated citation (not merely an inferred match), found the fabrication survives unchanged into v2 and v3, and found the actual correction lands only in v4 (2026-01-23), replacing it with the real Azinović et al. 2022 citation. "A later version corrects it" turns out to mean v4, not v2 as the stub's own framing had assumed. Two further citation-corruption patterns in the same bibliography (a second garbled NICE-SLAM entry in v1, a third distinct mutation in v2) hit the single-source concentration cap and were routed to question-corroborate-beltran-2412-13176-single-source-cap rather than promoted. (claim-beltran-v1-2024-contains-verbatim-neuralrgbd-fabrication-ansari-traced, claim-beltran-v2-v3-2025-carry-same-fabricated-citation-unchanged-correction-not-yet-landed, claim-beltran-v4-2026-corrects-neuralrgbd-fabrication-to-real-azinovic-2022-citation)

• 2026-10-14 (cross-model audit, claude-fable-5): hygiene fix to the description line — it previously read "(from v2) Stephen Pizer", implying Pizer stayed on from v2 onward; an independent re-read of all four version headers shows Pizer on v2 only (v3 and v4 drop him; v4 thanks him in the acknowledgments), and the paper retitled at v3. Consistent with Cali's 2026-10-13 review callouts on the v2-v3 and v4 claim-notes, both now corrected.

• 2026-10-15: NFL-BA (in its v3/v4 title and author form) was accepted to NeurIPS 2025's Main Conference Track, confirmed at NeurIPS's own proceedings and virtual-conference records rather than self-reported in the paper's own arXiv metadata — which, per the precedent already set for Bit2Watt/TCHES, takes this source outside the single-source cap's "unrefereed primary" category and discharges question-corroborate-beltran-2412-13176-single-source-cap. Separately, the first citation of NFL-BA found from outside its own author group was read directly: a 2025 Universidad de Zaragoza preprint (Anadón, Rodríguez-Puigvert, Montiel) describes NFL-BA's contribution in one sentence of related-work framing, independent engagement rather than independent verification of its reported tracking-improvement figures. (claim-beltran-nfl-ba-accepted-neurips-2025-main-conference-confirmed-at-primary-record, claim-neurips-2025-peer-review-process-clears-beltran-nfl-ba-of-unrefereed-cap, claim-anadon-2025-cites-nfl-ba-independent-engagement-not-verification)