Seasonal Forecast Predicts 'High Levels of Uncertainty' and 'Increased Probability of Being Wrong'

The man at the back of the hall asked whether the seasonal forecast came with a confidence interval. He was told that the confidence interval was the forecast.

The Folio's fungal season outlook was issued on Tuesday in its usual format: a chart, three months, a single percentage that everyone in the room would remember differently. This year the chart carries two additional columns. One is labelled Probability of Being Wrong. The other is labelled Probability of Being Wrong in a Way That Is Hard to Notice.

"The unit is fruiting-body observations per surveyed hectare per week," said the forecaster, Ines Varga, pointing at a footnote with the end of a pencil. "The denominator is surveyed hectares. It is not participants, it is not photographs submitted, and it is emphatically not hopes."

She then read the headline finding aloud, twice, because the first time several people assumed it was a joke about last year's forecast. High levels of uncertainty. Increased probability of being wrong. The chart behind her showed a modest rise in the middle week, flanked by error bars that left the frame of the poster entirely and continued onto the wall.

The reactions split along the usual line. Veterans said the outlook had gone soft and that a forecast which declines to name a week is a weather report with a hobby attached. Newcomers said it was the first honest thing the desk had published, and asked whether the two additional columns could be added to last season's retrospective as well. Both groups filed their complaint with Vera Pike, who had arrived to collect a chair and was blamed for the outcome by each side before she reached the door.

A competing group issued a counter-forecast the same afternoon, drawn from the same fortnight of field notes and arriving at conclusions with considerably more exclamation marks in them. Its authors described their method as intuition informed by experience. Its denominator was not stated.

"A forecast with no error bars is not a forecast," Varga told the hall. "It is a wish with a date on it."

Last year's outlook remains the reference point, and not because it was accurate. It was wrong in the direction everyone enjoyed, so the corrections thread ran for eleven days and the veteran who posted them still opens meetings with it. The desk has since adopted a standing rule: retire a claim when the evidence changes, and note in the file which way it changed.

By late afternoon the chart had been photographed, cropped to the three confident weeks, and posted with a caption about calling it. The reply arrived underneath in two words: "Which denominator?" Varga was seen at the sink, washing the pencil.