The part I would push past where you leave it is the third reason, why we keep asking even after we know it fails. You frame it as confusion plus theater plus poor evaluability. I think there is a fourth sitting underneath those, and it is not epistemic at all.
A prediction moves responsibility. A judgment about present conditions keeps the decision in the room, shared by everyone in it. A forecast lets the uncertainty be placed in one person's mouth, and once it is there, the rest of the room is lighter. That may be the real reason the decisive forecaster is rewarded and the one who hedges gets sidelined. It is not only that confidence reads as competence. It is that the confident forecaster is accepting a transfer the room is trying to make, and the hedger is refusing to carry it.
Which reframes your accountability problem. The forecasting environment does not just fail to learn because predictions are hard to score. It resists learning because the social function was never accuracy. It was relief. A system built to move a burden has little reason to check later whether the burden was placed correctly, and every reason to keep rewarding whoever will take it on demand.
I appreciate the thoughtful comments. As I read through your thoughts, I kept coming back to my second point about prediction theater and the idea that "asking experts to forecast those events offers a way to make an open-ended future feel more manageable." So, in that sense, I do think forecasts serve an important social function. Where I may see it a bit differently is that I don't think this replaces the accountability problem so much as complements it. Predictions remain difficult to evaluate well, so rather than carefully revisiting past forecasts, we often just move on to the next set of predictions. That creates a cycle where uncertainty is repeatedly compressed into confident forecasts, but the underlying forecasting process itself is rarely subjected to much scrutiny.
Complement rather than replace is the better framing, you are right. What ties them together may be that both reward moving on: the relief is spent once the forecast is spoken, and the evaluation never comes because revisiting would reopen the uncertainty the forecast was there to discharge.
The part I would push past where you leave it is the third reason, why we keep asking even after we know it fails. You frame it as confusion plus theater plus poor evaluability. I think there is a fourth sitting underneath those, and it is not epistemic at all.
A prediction moves responsibility. A judgment about present conditions keeps the decision in the room, shared by everyone in it. A forecast lets the uncertainty be placed in one person's mouth, and once it is there, the rest of the room is lighter. That may be the real reason the decisive forecaster is rewarded and the one who hedges gets sidelined. It is not only that confidence reads as competence. It is that the confident forecaster is accepting a transfer the room is trying to make, and the hedger is refusing to carry it.
Which reframes your accountability problem. The forecasting environment does not just fail to learn because predictions are hard to score. It resists learning because the social function was never accuracy. It was relief. A system built to move a burden has little reason to check later whether the burden was placed correctly, and every reason to keep rewarding whoever will take it on demand.
I appreciate the thoughtful comments. As I read through your thoughts, I kept coming back to my second point about prediction theater and the idea that "asking experts to forecast those events offers a way to make an open-ended future feel more manageable." So, in that sense, I do think forecasts serve an important social function. Where I may see it a bit differently is that I don't think this replaces the accountability problem so much as complements it. Predictions remain difficult to evaluate well, so rather than carefully revisiting past forecasts, we often just move on to the next set of predictions. That creates a cycle where uncertainty is repeatedly compressed into confident forecasts, but the underlying forecasting process itself is rarely subjected to much scrutiny.
Complement rather than replace is the better framing, you are right. What ties them together may be that both reward moving on: the relief is spent once the forecast is spoken, and the evaluation never comes because revisiting would reopen the uncertainty the forecast was there to discharge.