[!todo] Seed note. A starting point, not a finished note yet.
Three-point estimation replaces a single guess with three: an optimistic value (O), a most-likely value (M), and a pessimistic value (P). The PERT variant combines them into an expected value that weights the middle, (O + 4M + P) / 6, and derives a standard deviation of (P − O) / 6, giving not a point but a rough distribution you can attach confidence levels to. The value is honesty about uncertainty: the output has an explicit shape, so you can plan against the pessimistic tail instead of a phantom precise date. Its common failure is that people generate all three numbers by nudging up and down from one gut anchor, producing a range that is far too narrow to reflect real risk. Seeded from Software Estimation and Forecasting.