Agile Software Development, Scrum, Extreme Programing, XP, Test Driven Development, TDD, Feature Driven Development, FDD, Lean, DSDM, Behavior Driven Development, BDD, Refactoring, Pair Programming, Kanban

Predictability Without the Gamble of Poker

Often knowing when something will be delivered is important. Activities like hiring and purchasing have a lead time and can mean that organisations need to know when the product will ship. But Hofstadter’s law, the planning fallacy and optimism bias all reinforce the idea that we just can’t estimate! So how can we be predictable if we can’t estimate?

Over the years the Agile community has focused its efforts on estimation in small batches and endeavoring to eliminate group think and avoid our bias through techniques like planning poker. However not only can this be a demoralising and imprecise process often leading to day long planning sessions but it also invites the questions “when will I get my points” and “how do we increase the teams velocity”.

In this session we’ll explore Monte Carlo simulation. A technique that lets us take real lead time data. It lets us be explicit about our uncertainties, defect rates, blocking events and risks while factoring in the inevitable change in scope to produce a probabilistic estimate about when the product will be shippable. Not only can the models help us predict when work will be ready but they can also give us valuable insight into potential bottlenecks in our flow.

Video producer: