Planning for Uncertainty
summary
Argues that time estimates die beyond one sprint or two horizon because individual task estimates follow wide distributions and velocity only signals trends not delivery dates. Treats planning poker as protocol to surface assumptions and align team perception of size, not to forecast calendar days. Recommends starting with a walking skeleton, one stable team, and real engineering practices — CI, ATDD/TDD and continuous integration — to make delivery predictable while accepting long‑term uncertainty.
AI-written summary · about