earth-shape-of-the-record-and-share-of-the-budget

# Earth: report the shape of the record, and its share of the budget A multi-year record delivered as a period mean plus a cumulative total has been produced but not described. The three statements a reader takes from a record are how it changed over its own length, which periods were extreme, and how large it is beside the other terms of the total it belongs to. None of the three falls out of validating your series against a reference version of your series. ## The failure this prevents A run delivered a multi-decade global annual record and reported: the full-window mean rate, the cumulative total, the impact-unit equivalent, and - at length - how closely each of its sub-units reproduced a reference version of itself. While validating, it had already computed both half-period means, their percentage difference, and the single most extreme period with its uncertainty, and stored them in a survey-stage results file. None of the four reached the report. The word for the change in rate over the record appears in the report only inside a bibliography title. The only share-of-total anywhere in the run was the share of its own *reproduction residual* carried by a handful of sub-units: it computed fractions of its error budget and never a fraction of the physical result. That is what happens when the results section is indexed by the run's hypotheses. A descriptive statistic adjudicates nothing, so it has no section to live in, and the delivered object goes undescribed while th...