Uncoupling
summary
Argues that coupling comes in five concrete forms—operational coupling, development coupling, semantic coupling, functional coupling, and incidental—and shows how choices change them: direct DB access → strong runtime and schema coupling; REST weakens runtime; a message broker removes runtime coupling but raises functional coupling. Uses Informatica/PIM and an actual retailer case where 22 million items and ~12 million required price updates force a “price point” semantic leak that costs millions, and cites David Parnas’ information‑hiding and interface reuse as the practical cure.
AI-written summary · about
chapters (12)
- 0:00Introduction and background
- 1:55The problem of tight coupling
- 4:25Coupling in other disciplines
- 12:57Types of software coupling
- 15:43Analyzing coupling in an example
- 17:52Evolving away from coupling
- 24:40Trade-offs in reducing coupling
- 25:46Semantic coupling and leakage
- 30:45Parnas and information hiding
- 35:18Reducing semantic coupling
- 39:14Rotating perspectives on coupling
- 40:39Summary and key takeaways
related talks
51mBalancing Coupling in Software Design
01h 01mBalancing Coupling in Software Design
51mHiding The Lead: Coupling, Cohesion, and Microservices-Style
48mBalancing Coupling in Software Design
56mContinued Learning: The Beauty of Maintenance -
41mConnascence, beyond Coupling and Cohesion
37mHiding The Lead
01h 28mLaw of Demeter: A Practical Guide to Loose Coupling