One of the things that drives me crazy is when people say with such brazen confidence, "we don't need type checking because our tests catch type errors." But even if your tests do catch a large amount of the errors (depending on how thorough your tests are that you've written), YOU STILL HAVE TO GO HUNT DOWN WHERE THAT ERROR CAME FROM when a test fails. Isn't it so much nicer to just have the type checker automatically show you exactly where the type error came from?