Right. I made the mistake recently of asking for a function that produced Wilson confidence intervals and the not checking the results line by line, because the code looked plausible as did the outputs. It wasn’t for production code, so I was in hack mode. It totally messed up the formula, which a colleague picked up when he compared the results to ones he got from Excel. Because it’s often right, it’s easy to fall into the trap of trusting it, but even if it works, the code is sometimes stupid. I have found it to be a massive accelerator but it has severe limitations, not least corroding the ability of coders to solve problems independently.