The only thing worse than other people's code is other people's test code. Developers who write decent-ish code using good practices, encapsulation and good comments suddenly become absolute barbarians when writing test cases. I'm going to cut a bitch next time I see a test case named ServiceShouldReturnResultAsExpected which is 200 lines long and contains dozens of assertions, for loops and supplementary fake classes and gives you no clue about what the developer was even thinking about.
Below it is a test case called ServiceShouldReturnResultAsExpected2 which is exactly the same but some minor amendments, and below it is another one which is marked as "skipped" and the comment refers to a ticket from 4 years ago which was moved across sprints multiple times and eventually just got quietly dropped by some PM and the body of the ticket is "placeholder ticket to fix threading issue, call pratesh with any questions", only both people called Pratesh haven't worked here for years and I don't even know which one they're referring to.
Below that is another unit test that also replaces some, but not all repository classes with real classes that reference a local database server and take 2 minutes to run for no apparent reason. There is no custom fucking SQL there or anything and I can't even see any meaningful assertions. The names are also vague as fuck.
Eventually I figured out the scam: all the test cases were variations to send the frame pointer through specific code paths to meet some long abandoned management KPI from years ago to increase code coverage. They don't assert anything meaningful, the point isn't actually to test anything, it's just to improve numbers.
I took the time to clear up the technical debt and wrote about 20 test cases that represented real world expectations from the system, fixed the bug and was questioned by the PM about whether my time would have been better spent implementing some bullshit feature.
A programmer's life is pain.