← The collection

Planning Fallacy.

A project can take longer than the inside-view estimate suggests.

Interactive experimentintuitiveField note ·
Preparing the experiment…
THE SHORT VERSION

Planning Fallacy, explained.

The planning fallacy is a tendency to underestimate the time or resources needed for a task, even when similar past tasks took longer than expected.

01 / THE MECHANISM

Why it happens

An inside view focuses on the plan working as intended. An outside view asks how comparable projects actually performed, including interruptions and revisions. A buffer can help, but its size should reflect relevant experience rather than an arbitrary feeling of caution.

Forecasting only the imagined plan can miss variation from similar past projects.

Read the result

Compare the best-case estimate with the simulated completion distribution and the deadline including your buffer. These generated delays are teaching examples, not evidence about your actual project.

02 / FOLLOW IT THROUGH

A worked example

An eight-day plan

  1. A team estimates eight days of direct work and promises delivery in eight calendar days.

  2. It adds a 25% buffer, moving the deadline to ten days, then compares it with delays seen on similar projects.

  3. Two extra days may still be insufficient. A realistic estimate needs a reference class, not just a percentage added to an optimistic plan.

OPTIONAL DEEPER DETAILGo deeper: inside the model

Inside this model

A seeded teaching sample adds delay of 0–12 days to the best-case estimate across 100 hypothetical projects. The buffer adds estimate×buffer/100; the plot compares budgeted time and the sampled completion distribution.

03 / BEYOND THE EXPERIMENT

Where this idea is useful

A practical use

Use outcomes of previous similar tasks to set a schedule, then compare your planned buffer.

CHECK YOUR INTUITION

A common misconception

THE TEMPTING CONCLUSION

“Adding any buffer makes an estimate realistic.”

THE MORE USEFUL DISTINCTION

The starting estimate may already omit substantial work. A buffer is useful only when its adequacy is checked against a relevant range of outcomes.

What this explanation leaves out

  • Uniform delay is illustrative, not a calibrated reference class for your project.
ONE MORE QUESTION

What is a reference class?

A group of sufficiently similar completed tasks whose actual outcomes can inform the new estimate. Choosing comparable work matters as much as collecting a large sample.

TAKE THE IDEA WITH YOU

What happened on the last several projects that resembled this one?

Associated thinkers

Further reading

Explore the original research or the teaching reference behind this experiment.