Technical debt is a tradeoff: a shortcut or other implementation choice may help ship a feature sooner while making future changes harder. The debt matters most where change is frequent and friction is high; not every debt needs to be repaid. Production work can also reveal costs overlooked at delivery, such as operating, scaling, and evolving a feature. section 7.4.1 section 1.1.5
To decide where improvement is worthwhile:
- Estimate debt interest as change frequency multiplied by friction per change. Use Git history to gauge frequency and team retrospectives to estimate the extra time spent wrestling with a module. section 7.4.1
- Focus on high-churn, high-friction areas. Low-frequency modules may carry debt that is acceptable to leave in place. section 7.4.1
- Embed small improvements in feature pull requests instead of relying on dedicated refactoring sprints. After a bug, ask what could prevent it from recurring. section 7.4.1
Estimating friction in one module
In one release, three developers spend two, two, and seven hours understanding validation logic, debugging coupling, and adding a rule across six locations. That is eleven hours of friction in a single file; its change frequency helps determine whether the debt merits attention. section 7.4.1
Tame one legacy module
Find a frequently changed file that stalls reviews or repeats bugs. Observe its behavior with realistic inputs, capture that behavior in characterization tests, assess frequency and friction, then make one small improvement. section 7.5















