The planning fallacy is not a personal defect
Psychologists have a name for the reliable optimism of human time estimates: the planning fallacy, described by Daniel Kahneman and Amos Tversky in the 1970s. Its signature is stubbornness. People underestimate how long their own tasks will take even when they know their previous, similar tasks ran late. The memory of every past overrun somehow never makes it into the next estimate.
The mechanism matters for fixing it. When you estimate, you imagine performing the task - and imagination produces the best-case script: the version with no interruptions, no false starts, no revision, no waiting on someone else. You are not lying to yourself about the work; you are accurately estimating a version of the work that almost never occurs.
The inside view and the outside view
The research points at a reliable correction: stop asking "how will this task go?" (the inside view - which triggers the best-case script) and ask instead "how long has this kind of task taken before?" (the outside view - which consults the record).
The outside view works because it silently includes everything the script omits - the interruptions, the rabbit holes, the second draft - without you having to predict any specific one of them. You do not need to know what will go wrong; the historical number already contains the average of everything that usually does. The catch is that the outside view needs a record to consult, which is exactly what most people do not keep about their own days.
The invisible taxes: transitions and switches
Plans are written as if work is contiguous: 9:00-11:00, deep work, done. Real work pays taxes at every boundary. Opening the project and reloading the mental state takes time. Physically moving between contexts takes time. And after every interruption there is a re-entry cost - attention research consistently finds that refocusing after a switch takes real minutes, not seconds, and that some residue of the previous task lingers into the next one.
None of these taxes appear as line items in anyone’s plan, which is why a day of six planned hours so often yields four real ones with nothing visibly "wasted." The time went to seams between the work. A plan that acknowledges seams - buffers between blocks, fewer and longer blocks rather than many short ones - slips far less, not because the work got faster but because the plan stopped denying the taxes.
Unplanned work: the day’s dark matter
Compare any planned week against the real one and a third category always appears: hours that belonged to no block at all. The favour that took the afternoon. The small emergency. The forty-minute "quick look" at something. Unplanned work is the dark matter of the schedule - invisible in the plan, yet bending every trajectory in it.
You cannot eliminate it, but you can budget for it. If your record shows roughly a quarter of each day goes to the unplanned, the honest response is to plan roughly six hours of a nine-hour day and let the reserve absorb reality. A plan at full capacity is not ambitious; it is pre-broken.
Measure your multiplier
Here is the practical, one-week fix. Track your actual time - just start and stop times against what you planned, no elaborate categories - for one ordinary week. Then, for each kind of recurring work, divide actual time by planned time. That number is your personal multiplier, and for most people it is remarkably stable per category: perhaps around 1.5 for focused writing or coding, higher for anything creative or open-ended, close to 1.0 only for truly routine tasks.
The multiplier converts the planning fallacy from a character judgment into an engineering constant. You do not need to become a better estimator; you need to keep estimating exactly as you do, then multiply. When the block for "write the proposal" is sized at your real historical rate, it holds - and a schedule whose blocks hold is a schedule you start believing again.
What to change, concretely
The full prescription fits in five lines.
- Keep a record of planned versus actual time for at least a week - you cannot correct what you have not measured.
- Estimate as usual, then apply your measured multiplier per type of work.
- Add buffers at every boundary between commitments; prefer fewer, longer blocks.
- Budget explicit reserve capacity for unplanned work instead of planning at 100%.
- Schedule start times you have historically kept, not the ones that sound virtuous.
Key takeaways
- Underestimating your own tasks is a documented, universal bias - not a discipline problem.
- Estimates fail because imagination produces the best-case script; the historical record includes what the script omits.
- Transitions, context switches and re-entry after interruptions are real costs your plan must carry as buffers.
- Unplanned work is a permanent category; budget reserve for it instead of pretending this week will be different.
- One week of tracking gives you a personal multiplier that turns estimation into arithmetic.
Doing this in Fezer
Fezer exists to make the record effortless enough to actually keep. You plan the day in blocks, tap Begin and End to track real sessions as they happen, and Compare mode draws both columns on the same hours - late starts, overruns and unplanned dark matter all become visible without a spreadsheet. Analytics then rolls the same data up by week, which is where your multipliers and your true unplanned-work budget emerge.
It is free, and the record it builds about your days never leaves your device - no account, no server.