A confession first: I have never once finished a project by the day I privately promised myself I would. If that sounds familiar, you're in large company, and the size of the miss turns out to be measurable.

In 1994, three psychologists — Roger Buehler, Dale Griffin, and Michael Ross — asked university students a simple question about their honors theses: realistically, when will you hand this in? The students guessed 34 days, on average. The researchers also asked for a worst-case date: if absolutely everything went wrong, then when? About 49 days. Then everyone went off to write.

The average thesis took 55 days. Reality didn't just beat the plan; it beat the version of the plan where everything goes wrong. Barely a third of the students finished by the date they'd named. Kahneman and Tversky had coined a term for the pattern years earlier: the planning fallacy.

Why your estimates are fiction

The mechanism is almost elegant. When you estimate a task, you imagine doing it, and the imagined version runs clean. Nobody's mental movie includes the sick day, the surprise revision, the week your attention gets eaten by something else entirely. The delays that will actually happen are, by definition, the ones you didn't picture. So every estimate is secretly a best-case scenario.

The stranger part is that you already own the evidence against yourself. You have blown a hundred deadlines. But memory files each one under special circumstances — that time was the flu, that time was the client — and the belief that this time will run clean survives untouched.

Your history says fifty-five days. Your imagination keeps quoting thirty-four.

Ask a different question

The fix that holds up in the research is a change of question. Not "how long will this take?" Your imagination answers that one. Instead: "how long did things like this take last time?" Your history answers that, and your history doesn't flatter you. It's also why friends tend to guess your timelines better than you do; they were never inside the clean version.

There's no curing the optimism, and you wouldn't want to. Nobody would start anything hard while fully pricing in the middle part. But you can stop being surprised by yourself. Take your estimate, double it, and give the doubled number to whoever is waiting. You'll land somewhere between the two, which is to say: on time, for once.