Every think tank and newsroom should keep a public forecast ledger

The problem: experts and commentators make confident predictions, but there is almost no cost to being wrong, because nobody keeps a record.

The proposal: a public ledger, hosted by each organisation, in which every published forecast is logged with its date, wording, probability where stated, and resolution date. When the date arrives, the ledger records what happened, who checked it, and whether the forecast was right. Over time it can show a track record, scored using a measure such as the Brier score.

Who acts: editors and research directors adopt the practice; an independent reviewer resolves disputed outcomes.

Costs and trade-offs: it takes modest staff time. The risk is that people avoid making forecasts, or make them vague. A rule that predictions must be specific enough to resolve helps, and so does a culture that rewards learning from errors.

First step: one organisation starts with the forecasts in its next annual outlook and reports results the following year.

This is the same logic as futures studies: knowing what we expected is the first step to learning.