Beyond "it depends" - How much architectural work do we need?
summary
Claim: organize architectural effort by the degree of decision uncertainty rather than a single method. The talk maps a spectrum from big design up front through decent design upfront and just enough design to startup-style rapid experiments, ties tooling to practice (CI/CD, API-first, devops, tiny microservices), and prescribes fitness functions and continuous feedback loops. Two striking numbers: roughly 80–90% of lifecycle cost is post‑release and only ~10–15% of efforts typically create value, so the recommended practice is rapid, measured experiments (small questions, early feedback) to cut wasted work.
AI-written summary · about
chapters (12)
- 0:00Introduction to architectural work
- 1:11One-size-fits-all in software development
- 4:20Understanding uncertainty in architecture
- 10:39Value optimization under uncertainty
- 13:27Sources of uncertainty
- 15:33From certainty to uncertainty spectrum
- 19:45Cost-based vs feedback-driven approaches
- 24:38Architectural work across life cycles
- 27:45Big design vs just enough design
- 32:01Beyond it depends in architecture
- 34:26Factors influencing architectural choices
- 35:27Summary and Q&A
related talks
56mEssential Architectural Thinking – Why, What, How, When and How Much?
49mAs Much Architecture Work as Necessary
25mThinking Architecturally
43mThinking Architecturally
28mThe lean architect: Experimenting your way to better decisions
49mCultivating Architecture
58mThe past, present and future of Architecture
51mWhy We Need Architects (and Architecture) on Agile Projects