FoMind
On Friday evening you and a friend are chatting about the weather. “It’s going to get colder this weekend,” you say. By Sunday it’s one degree cooler than Friday. Your friend says you nailed it, but you’re not so sure it counts. The problem isn’t your judgment — it’s that the statement can’t really be checked. How much colder is “colder”? Which day is “this weekend”?
Three tests of a good question
A forecasting question worth recording should pass at least three tests:
- Checkable: when the time comes, there’s a clear fact to settle it — official data, a public news report, or your own records.
- Deadline: it says exactly when it will be resolved. A forecast without a deadline can always hide behind “not yet.”
- Unambiguous: anyone else judging it would reach the same verdict. Words like “colder,” “soars,” or “goes viral” need to become specific numbers or events.
Rewrite it
Some before-and-after examples:
- “Stocks will go up soon” → “The CSI 300 index closes this Friday higher than it closed on Monday”
- “My delivery will be slow” → “Tonight’s food order arrives after the estimated delivery time shown in the app”
- “The weekend weather will be bad” → “The official weather station records measurable rain this Saturday”
Once you rewrite a question, you’ll notice your confidence changes too. A vague question lets you feel like you “sort of know.” A sharp one forces you to think: what are the actual odds?
Write the resolution criteria up front
Edge cases cause the most trouble. If the delivery arrives exactly on time, is it late? If a few drops fall, did it rain? Decide these cases when you write the forecast and put them in your resolution criteria, so you can’t go easy on yourself at review time.
A good forecasting question is one where, on the deadline, anyone can look at the facts and tell whether you were right.