I agrret with you, but want to say I think it's important to write code that is "brittle" enough to just satisfy the requirements. I've seen a lot of "edge case

I agrret with you, but want to say I think it's important to write code that is "brittle" enough to just satisfy the requirements. I've seen a lot of "edge case that probably never happens" code that ends up causing problems down the line because it does get hit, but for a unexpected situation and ends up mangling data. Especially in the paradime of black box/microservices, if something comes in that doesn't line up with a known senario, there should only be enough code to log the data and raise an alert/error. Having something bad happen, and handling it so it becomes 'less bad' and doesn't get flagged, escalated, and tested properly is going to cause issues.