One quantity, drawn six ways: how much the economy grew in a given quarter, as estimated by each of the 244 vintages that have published it. Drag a model to turn it. “Flat view” swings the camera to the angle where values read without perspective, because 3D is for shape and flat is for numbers.
Nothing here is a model’s error bar. The spread is what the official record has actually said about the same quarter, at different times.
How far a quarter’s growth estimate can still move as it ages. The centre line is the average distance from that quarter’s own first print; the funnel is the spread across every quarter. It narrows, and it never closes.
How far each published version of history sat from what the record says today. Every line is one vintage. Where the fog closes on the centre the vintages agreed; where it opens, the same quarter was being read differently depending on when you looked.
Each point is one quarter: what was first published, against what the record says today. On the amber plane the number never changed. Above it the quarter was revised up; below it, down.
The shape of the revisions, one ridge per era. Each ridge is how often a quarter was revised by a given amount. A narrow ridge means the first print usually stood; a wide one means it often did not.
How much of a quarter’s revision is still ahead of it. Each bar is the average further move after the estimate has been on the books for a year, five years, twenty.
The whole record as a landscape. Height is not the growth itself but how far each vintage sat from what the record says today: the plain is where the vintages agreed, and the hollows are where the same quarter was read differently depending on when you looked. Holes are quarters a vintage could not yet know about.