Essays
A record that can say I don’t know.
Every form invites a guess. A field wants a value, a blank looks like unfinished work, and so people put something in — an approximate year, a remembered price, a service date that is probably about right. Six months later nobody can tell which entries were known and which were estimated.
That is the failure. Not the guess itself, which is often the best available answer, but the fact that the record stopped distinguishing between a fact and a reconstruction.
The two blanks that look the same
An empty field means one of two entirely different things. Either nobody has got to it yet, or somebody looked and the answer genuinely is not recoverable — the invoice is gone, the seller is untraceable, the service predates your ownership.
The first is a task. The second is a finding, and it is worth as much as a filled field, because it tells the next person to stop looking. A record that cannot express the difference converts real research into something indistinguishable from neglect.
Why it matters more than it sounds
Confidence is what a sceptical reader is actually assessing. An insurer treats a formal appraisal differently from an owner's estimate. A buyer treats "serviced 2019, invoice attached" differently from "seller said serviced around 2019". Both versions of that second one might be true; only one of them survives a question.
Flatten those into a single field and the strong claims are dragged down to the level of the weak ones. Everything becomes equally assertable and therefore equally doubtable, which is the opposite of what a careful record is for.
The version that works
Three states rather than two: known and evidenced, believed but unverified, and established as unrecoverable. Every fact about a watch fits one of them, and labeling which is a matter of seconds at the moment of entry.
Done consistently, it also changes behavior. A field marked "believed" is an open question you might close on a rainy afternoon. The same field filled in confidently is closed forever, wrongly, by you.
Where the software fits — and where it does not
Watch Vault Archive prints unrecorded sections as absent rather than empty, so a thinly documented watch produces a shorter document instead of a padded one, and separates what was verified from what was asserted. What it cannot do is judge which is which. Marking a claim as believed rather than known is a decision only the person entering it can make honestly.
One thing worth doing today
Find one field in your record you filled in from memory. Mark it, in whatever way your record allows, as remembered rather than documented.
That single annotation is worth more than the figure it qualifies, because it is the only part nobody can reconstruct later.
More essays · Documenting provenance · How to value a watch yourself