TB: SCRUM (Sutherland)
Core Thesis
Almost all project management fails because plans are made upfront and never revised. SCRUM replaces the plan-then-execute model with continuous inspection and adaptation: regularly check whether you're building the right thing, whether the team is functioning well, and what is getting in the way.
Key Takeaways
The planning fallacy and Gantt charts
- The only problem with Gantt charts: they are always wrong. They were invented just before WWI and haven't improved since.
- Normal project management planning is stupid because people are so bad at predicting how much work something requires.
- Key insight: refine the plan throughout the project rather than do it all upfront. Each sprint produces real information that the plan doesn't.
Team performance dominates individual performance
- Improving team performance matters by orders of magnitude more than improving individual performance.
- That very large groups do less is an iron-clad rule of human nature. Optimal team size is small.
- Don't look for bad people, look for bad systems — ones that incentivize bad behavior and reward poor performance.
The sprint rhythm
- Three daily questions: What did you do since last meeting? What are you going to do next? What is getting in your way?
- 80% of the value of any piece of software lives in 20% of the features. Build that 20% first.
- People > process; products that work > documented specs; collaboration > negotiation; responding to change > following a plan.
Heroism is a warning sign
- A team that depends on regular heroic actions to make its deadlines is not working the way it should.
- What is the smallest improvement that can be made right now that will make things better?
Mental Models
- Inversion — instead of planning comprehensively upfront, ask what's the minimum needed to learn whether you're on the right track
- The Middle Way — the sprint is a middle path between chaotic improvisation and rigid upfront planning