Many users praise Uncle Bob for trusting AI agents via strict tests without reading code, calling it wise and heroic, while others dismiss the approach as wasteful, hypothetical, or prone to bypassed constraints.
Based on 32 visible X reactions from 209 accounts.
Ask a question below.
Published answers will appear here.
@unclebobmartin @ori_pomerantz You just waste like 75% tokens by adding slop tests to slop code. The correct strategy is TDD and just it, you must read and write tests. I think you'd be far more efficient if you throw away 90% of these nonsensical tests that add maintenance burden and test nothing.
@unclebobmartin @ori_pomerantz Instead, im writing at a new higher level, true executable specifications that i dont have to automate. Surrounding with tests and checks.
@unclebobmartin @theo @ori_pomerantz Love this. I created a /gauntlet skill that does exactly this. Keep expanding and updating it too
@unclebobmartin @brunocalza @ori_pomerantz Effectively you're reviewing a specification
@unclebobmartin @ori_pomerantz nice! thanks for your response:)
Uncle bob doesn’t read code. The 73 year old, author of one of the most popular coding books ever, Clean Code, doesn’t read code from agents This is the way to code now. Testing, curating & verifying is the most important part You’re not behind an old boomer in adopting ai right?
Many users praise Uncle Bob for trusting AI agents via strict tests without reading code, calling it wise and heroic, while others dismiss the approach as wasteful, hypothetical, or prone to bypassed constraints.
Based on 32 visible X reactions from 209 accounts.
Ask a question below.
Published answers will appear here.