Ask 'how long have similar tasks taken?' not 'how long will this take?' — outside view beats inside view
When planning task duration, deliberately switch from inside-view scenario construction to outside-view base-rate consultation by asking 'how long have similar tasks taken?' instead of 'how long will this take?'
Why This Is a Rule
Kahneman and Tversky's inside/outside view distinction is one of the most robust findings in judgment research. The inside view constructs a scenario for this specific task: "I need to write the API, then add tests, then do code review — should take about 3 days." This feels accurate because it's detailed, but it systematically underestimates by ignoring the unexpected obstacles, interruptions, and scope changes that affected every past task.
The outside view consults the base rate for tasks like this one: "How long have similar API implementations taken in the past? My last three took 5, 7, and 6 days." The base rate includes all the delays and obstacles that the inside view's scenario excluded — because those delays actually happened in the reference class.
The question substitution is the mechanism: "How long will this take?" activates inside-view scenario construction. "How long have similar tasks taken?" activates outside-view base-rate consultation. Same estimation task, completely different cognitive mode, dramatically different accuracy.
When This Fires
- Estimating duration for any task, project, or deliverable
- During sprint planning, project estimation, or deadline setting
- When your estimate feels confident and detailed (high inside-view activation)
- Complements Decompose tasks into steps before estimating — holistic estimates hide transition costs (decompose before estimating) and Apply your documented error factor to timeline estimates before sharing (correction factor)
Common Failure Mode
Using the inside view and then "adjusting" with outside-view data: "My scenario says 3 days, but similar tasks took 6, so I'll say 4.5." This anchors on the inside-view estimate and insufficiently adjusts. The outside view should be the anchor: start with "similar tasks take 6 days" and only adjust downward if you can identify specific, concrete reasons this task is different.
The Protocol
When estimating task duration: (1) Ask: "How long have similar tasks taken?" Identify 3-5 comparable past tasks and note their actual durations. (2) Calculate the average. This is your outside-view baseline. (3) Ask: "Is there a specific, concrete reason this task will take less time than the reference class?" Not "I feel good about this one" — a specific, articulable reason (smaller scope, familiar technology, no dependencies). (4) If you can articulate a specific reason → adjust the baseline modestly downward. If not → use the baseline. The outside view is almost always more accurate than the inside view.