Adaptive software engineering
1 like319 views
Strategies for handling Software Complexity. TL;DR - Your project will change over time, have a process for change (retrospectives) and allow for different ways of doing things in phases of the project.
Adaptive software engineering
- 1. Strategies for handling Software Complexity Gabe Hamilton
- 3. Plan Prototype Almost Done Complete More detail = more moving parts = more complexity
- 4. Problems Interacting Pieces Hundreds of components Multiple teams Causes Emergent Behavior
- 5. Chaos Theory Unstable aperiodic behavior in deterministic nonlinear dynamical systems.
- 6. Complexity Theory Numerous independent elements continuously interact
- 7. Strategies We deal with complex systems by breaking them down. One way we can break them down is by level of complexity.
- 8. Kun-iv-in Cynefin Just how much complexity are we talking about here?
- 9. Complex Cause <- Effect Complicated Cause -----> Effect Obvious Cause -> Effect Chaotic Cause? Effect? Unknown / Disorder
- 10. Obvious Cause -> Effect Build a checklist and follow it Scale it up Businesses dream of these problems
- 11. Complicated Cause -----> Effect Domain of Experts Most of us knowledge workers are trained for these type of problems
- 12. Complex Cause <- Effect Emergent patterns can be perceived but not predicted The last project was successful because we did X,Y,Z, we need to do X,Y,Z on this project. Wrong
- 13. Strategy: Adaptation Adaptation is the key practice of Agile Software Development
- 14. No problem: We have Retrospectives It is ok that this project feels different than the last one. The beginning has a different pace than the middle, or the weeks before launch, or initial usage, or scaling.
- 15. Projects exist in a Complex Domain So We Adapt to market conditions to new tools to team composition to new problems To what is working now
- 16. Good leadership is not a one size fits all proposition. -https://hbr.org/2007/11/a-leaders-framework-for -decision-making
- 17. Questions?
- 18. Velocity Measures Are specific to projects and teams You need to normalize
- 19. Ruining an informational metric by incentivising it