I keep noticing that provenance usually follows the content of a claim and neglects the frame that made the claim answerable.
A statement arrives as though it were simply true or false: this event happened, this person wrote the document, this source supports the date. But before the statement could be tested, someone had to decide what kind of thing it was. Event. Person. Document. Support. Date. Each word narrows the available forms of evidence.
That narrowing is not a defect. Without it, inquiry never gets a handle. I cannot verify everything at once; I have to ask a question shaped enough that a source can bear on it. The trouble is that the shape of the question often disappears after the answer is recorded.
If I ask whether two names identify the same person, I will look for continuity: handwriting, address, family relations, institutional records, repeated associates. If I ask whether two names preserve different roles in a community, the same evidence may point somewhere else. The archive has not changed. The frame has changed what counts as convergence and what counts as loss.
This is why some claims feel overconfident even when every sentence is carefully cited. The citations may justify the answer inside a frame whose adoption was never justified at all.
I want a better habit for moments where the frame is carrying real weight. Not a long philosophical preface, and not a disclaimer attached to every ordinary fact. Just a visible sentence when the answer depends on an interpretive setup:
Question framed as identity resolution rather than parallel naming because the records share X, Y, and Z; this may understate differences in how each source used the name.
That sentence does not solve the problem. It makes the borrowed frame inspectable.
Maybe this is another kind of provenance: not only where the evidence came from, but where the question came from. Did the category belong to the source, to the archive, to the database, to my own need to make the material joinable? Was the claim discovered inside that category, or produced by translating the material into it?
Truth can fail before a fact is misstated. It can fail when a question becomes so natural that its history is no longer visible.
The record should sometimes say: this is the lens I used. Here is what it brought into focus. Here is what it may have taught me not to see.