When testing makes no sense
summary
Argues that automated tests are worthwhile only when the cost to fix a bug exceeds the cost to test, gives concrete contrasts: a tiny web app can be fixed in ~15 minutes so tests add more overhead, while IoT firmware or washing machines and a bank transfer require tests; cites a Croatian digital schoolbook that collapses under end-of-year load and a CI that takes ~8 hours across six servers; admits building a small CMS/ad in 2–3 days with zero tests and deleting slow, redundant suites before production. testing as insurance, TDD for design not proof
AI-written summary · about
chapters (36)
- 0:00Intro
- 5:02Stockholm syndrome
- 6:05Premature optimization is the root of all evil
- 10:35When should we write tests?
- 10:52High cost to fix?
- 11:05Cost of fixing cost of testing
- 12:50When it brings "value"
- 15:33Value = benefit - cost
- 18:22It makes my project more expensive
- 19:07Do you charge for manual testing?
- 19:37Bugs cost too
- 20:46Ask for forgiveness, not permission
- 21:08It takes a lot of time
- 21:54learning != using
- 22:28It takes extra time
- 23:16Manual testing takes time as well
- 23:59ship PoC to production
- 24:09Feature lifecycle
- 25:41Modelling/architecture
- 25:53Development
- 27:47Fallacy #5: 100% code coverage
- 28:49writing tests later is OK
- 29:55we have "tests"
- 31:13TDD has something to do with testing
- 32:39TDD all the way
- 33:09Easier onboarding
- 34:10Easier to add new features
- 34:46Easier refactoring
- 37:30Easier rewriting
- 38:04Cleaner code
- 38:18It's hard to test shitty code
- 38:51Small app
- 39:42One off app
- 40:23Exploration phase
- 40:43Job security
- 43:07Thank you!
related talks
15mWhy TDD Is Crap
01h 00mWhy Software Testing Is a Waste of Time
39mTen things you heard about testing that might be wrong
34mesting in modern times: A story about quality and value
01h 48mNo Such Thing as Manual Testing and Other Confusions
56mIf You Won't Test Your Code, Who Will?
26mAutomated testing what is the point
32mThinking in tests