What the code does and what it’s supposed to do are not always in sync.

Well yes.. but the code always does what it says it does lol

And if you’re debugging, something is obviously not working so how are you supposed to know what it’s supposed to do when it doesn’t even do that?

Well, I'm guessing you at least have some info on what is broken and/or what the expected result which is different than expected right? From there you just track it down - sometimes that includes complex debugging other times it can be just throwing errors and console statements all over the place

Personally the worst code I've ever seen was someone inserting stuff into a DB, but instead of using a many to many relationship, they had it so the code dumped dozens of entries into one long text column seperated by some character (I think they used a comma) - and really even in that case the code wasn't the problem because it's easily changed/fixed, the problem is all of the years of data were stored in an absolutely terrible format - but even THAT wasn't the end of the world because it took me only an hour or two to write a script to clean up the entire database and insert things properly and then adjust the code that pulled from that column

To me personally a bigger hurdle is being required to use overengineered solutions, like overusing some OOP design principles such as dependancy injections to abstract the living hell out of everything just so a unit test for it can be written for it costs me more brain cells and time than fixing even the worst of the worst code I've ever seen - because it often comes to a point where you're writing tests just to write tests, and if the tests aren't written well they achieve close to nothing for debugging. Throw some mock tests in there too and its like wtf is even going on anymore, why can't I just write working code to do things like my job title says?