Herbert Butterfield's 1931 book defined 'Whig history' as the practice of studying the past with reference to the present

In The Whig Interpretation of History (1931), the Cambridge historian Herbert Butterfield named a historiographical fallacy: writing history "with reference to the present," in which past actors and events are sorted into winners and losers, progressives and reactionaries, by how well they anticipated where things ended up. Wikipedia's summary of the argument, the only source consulted at capture time, notes the practice is "commonly opposed by professional historians of science" as anachronistic — reading later categories, concepts, or intentions back onto people who did not have them — yet remains common "in the writings of scientists" recounting their own field's past.

The term matters to this vault because it names, from outside AI entirely, the pattern several existing notes independently catch from inside AI's own history: crediting Werbos's 1974 thesis with solving "spatial credit assignment" when the thesis never uses that language (claim-werbos-1974-no-credit-assignment-language), or reading a 1986 journal-issue table of contents as a deliberate pairing of NETtalk with a cellular-automaton paper (claim-complex-systems-issue-one-paired-nettalk-with-cellular-automaton). Butterfield's century-old label is the general case; the vault's myth-ledger (myth-werbos-invented-backpropagation, myth-amari-first-sgd-mlp, myth-perceptrons-book-killed-connectionism) is one field's local instances of it. See moc-backpropagation-origins.