Residues: Time, Change & Uncertainty in Software Architecture
summary
Claims that random simulation of stressors plus an NKP (N, K, P) analysis produces architectures that survive more unknown events than requirement‑led design, measured by a residual index (Y−X)/S. Tools shown include stressor lists, incidence matrices to reveal hidden coupling, and iteratively built car charging residues (ALPR fallback, cameras, per‑minute pricing) that handle icing and vandalism. An experiment across teams yields a statistically significant improvement and the method resists attempts to falsify it.
AI-written summary · about
chapters (9)
- 0:00Intro
- 1:50A short history of software architecture
- 2:02Software architecture & complexity
- 11:10Uncertainty & the complexity sciences
- 11:20Software from a complexity perspective
- 21:15Residuality theory
- 29:06Example: Car charging systems
- 38:39Contagion analysis
- 46:08Outro
related talks
54mAn Introduction to Residuality Theory
50mKEYNOTE: Residuality Theory in Software Architecture
44mThe Philosophy of Architecture
45mReducing risk in projects, increasing resilience to unexpected change
43mBetter Software Development Systems through Stress
01h 05mKeynote - Without Resilience, Nothing Else Matters
01h 04mPredictably Unpredictable
51mRespect estimates