The planning fallacy is one of the most replicated findings in behavioural psychology: when humans estimate how long something will take, we are systematically optimistic — even when we have ample evidence that our past estimates were similarly wrong.
First named by Kahneman and Tversky in 1979, it's been replicated everywhere from software projects to home renovations to academic theses. The size of the underestimate hovers around 30–50%.
Why it exists
The simplest explanation: we estimate by mentally simulating the task going well. We don't simulate the call that runs over, the dependency that's out of date, the client who needs three rounds instead of two. We picture the version of ourselves who finishes cleanly and goes for a walk.
The trick our own brain plays is treating each task as unique — convincing us that this time is different, that we know the codebase better now, that the brief is unusually clear. The plural form ("how long does work like this usually take?") doesn't activate; the singular form ("how long will this one take?") does, and that one is optimistic.
How to counter it
The most reliable countermeasure is reference-class forecasting: instead of estimating this task, look at the distribution of similar tasks you've already done and use the median + range. Your past self has already tested every excuse your present self is about to use.
How Ensaria relates
Ensar's Reality Engine corrects the planning fallacy by default. Every task estimate comes from your own past completions, weighted toward recency, with median + range + sample size displayed inline. You can override; but when you do, you're now overriding your own data — a healthier kind of optimism than overriding an empty void.
Related terms
- Reference-class forecasting — the antidote.