Essays
Every number on a screen is a claim.
Displaying a figure looks neutral. The data exists, the software shows it, no argument is being made. That is not how it is read. A number presented in an interface reads as something the software is asserting, and the reader does not separate the display from the claim.
Which means every figure is a small argument about the collection, whether or not anybody designed it to be one.
Three numbers that look identical
Consider three figures side by side in the same typeface. A purchase price, taken from an invoice — a fact. A current market value, typed in by the owner after ten minutes of looking at listings — an estimate. A total collection value, computed from those estimates — arithmetic performed on guesses.
On screen, all three are just numbers. The third is the least reliable and looks the most authoritative, because it is the largest and the software produced it rather than a person typing it.
That inversion is the problem. Computation confers apparent authority exactly where it is least deserved, because a computed figure inherits every weakness of its inputs and displays none of them.
The test worth applying
For any figure an application shows, ask: what would a reader reasonably conclude from this, and can we stand behind that conclusion?
Not "is it correct" — most are arithmetically fine. A total is genuinely the sum of its inputs. The question is what the reader takes from it, and a total of owner-typed estimates reads as a statement about what a collection is worth. Nobody could stand behind that, and the interface said it anyway.
What to do instead of hiding it
Removal is one answer and rarely the best one. The better one is attribution: show the figure with its provenance attached, so the reader can weigh it.
"Estimated, by you, on this date" is a completely different claim from the same number alone, and it costs one line. A figure that names its own basis stops being an assertion by the software and becomes a record of what somebody entered — which is what it always was.
Where that cannot be done honestly, removal is correct. A number that cannot be qualified into something defensible should not be displayed, however easy it was to compute.
Where the software fits — and where it does not
Watch Vault Archive is built to hold an estimate and an outcome apart rather than blending them into a single figure, and to show the workings behind any number it presents. What it cannot do is improve the inputs. Every valuation is the owner's own estimate unless independently appraised, and a total of estimates is exactly that however neatly it is presented.
One thing worth doing today
Look at the headline total in whatever holds your collection, and work out how many of the figures behind it you could actually defend to an adjuster.
The ratio is the real accuracy of that number, and it is never the one displayed.
More essays · The number we took out · How to value a watch yourself