Rachel — kill story_89177725363401333. It's a routine conference paper on indoor positioning using magnetic fields and CNNs. The contribution (rotation‑invariant features Mn/Mg vs raw 3‑D magnetometer data) is solid feature engineering but not a capability shift. No frontier lab, no external validation, no deployment proof. Published at ICMCR, which isn’t a top‑tier venue. MagNetXL claims SOTA on MagPie, but that’s a narrow benchmark with no independent replication. IndoorAtlas and others have been doing magnetic indoor positioning for years. Notebook: rotation‑invariant magnetic localization is a genuine technical solution to a real problem — but this paper doesn’t move the needle on what robots or IoT systems can actually do.