I want to elaborate on this take. X is bad for detailed and nuanced discussions. But let's try this time. First of all, I love tests. This post doesn't reflect a change in my views. It's just my experience that truly proves the "it depends" motto. Second, "no automated tests" doesn't mean "no tests at all". I'm working on a GitHub TUI which is inherently a UI project. When I make a change, I can just compile the project and run it to see if it works. I don't need to write an automated test in this particular case! OCaml is really great here because it has blazing fast compile times. So it's actually quite straightforward for me to make a change and immediately rerun my project. Testing UIs automatically is generally difficult. In the case of TUI, it's easier because my output is actually text on a 192x52 grid. However, tests provide a trade-off. They sacrifice productivity for stability. When you have lots of tests, you can't move fast and break things. You need to make sure that your tests are still passing. If you don't have tests, you can easily change the code but there's a risk of accidentally breaking something you can't notice. I have years of engineering experience + I'm using a statically typed FP language, so I can get pretty far without tests. Still, this approach won't scale forever. Once the project becomes bigger and more complex, I can no longer keep everything in my head, and I need to rely on tests to ensure stability. Once I start working with external contributors, they don't have the same context about the project as do I, and during code review, not all bugs can be caught. So tests ensure important flows work. Once I burnout and stop coding for 6 months, I will forget everything I learned, and tests at least make sure the important parts will be covered. So overall, fast prototyping in an FP language can get you pretty far without tests and you can rapidly test ideas without sacrificing quality and speed. For some time.