We systematically underestimate how long things take because we plan for the best case and ignore how similar projects actually went. The fix is the 'outside view' — estimate from real track record, plus a buffer.
The planning fallacy: we systematically underestimate how long tasks will take and how much they'll cost — even when we've been burned by the exact same thing before.
Ask someone when they'll finish a project and they picture everything going right: no interruptions, no bugs, no sick days. That best-case story becomes the estimate. Reality then delivers the interruptions, and the thing takes twice as long. Kahneman called this the planning fallacy, and the maddening part is that it survives experience — the same people who were late last time are confidently optimistic this time.
Why it persists: when we plan, we look inside the specific task and imagine its steps (the 'inside view'), which naturally tells a smooth story. We ignore the 'outside view' — how long similar projects actually took, including all the ones that went sideways. The renovation, the thesis, the software launch: each feels unique from the inside, but from the outside they're members of a category with a known, longer track record.
The fix is that outside view: instead of estimating from the steps, ask 'how long did this kind of thing take last time, and for other people?' and start there. Then add a buffer for the unknowns you can't yet name, because there are always some. A useful rule of thumb: take your honest estimate and increase it by 50% (more for anything novel). You won't feel like you need the buffer. You will.
Almost every deadline you set, budget you plan, and 'I'll just finish it tonight' is quietly optimistic — naming the bias is what lets you plan for reality instead of the fantasy.
It's not 'you're bad at planning,' and it's not fixed by 'just try harder to be realistic.' Willpower doesn't touch it — the bias comes from taking the inside view, and even experts repeat it. The only reliable fix is switching method (outside view + buffer), not being more careful with the same method.
Lock this idea into memory with a 5-minute active-recall session — the science of spaced repetition, no signup.
Try this idea free →