Ask HN: What do you consider as “bad” code?

My opinion about this constantly changes through my programming experience. Functions longer than 20 lines of code? Pffff, an algorithm/problem can be such complex in nutshell and decomposing a function solving it will be non-relevant (see any std library of PL). Mutable global state? Pfff, anything in software is based on that, the problem is using it properly in target architecture (ex. in micro-controllers global state usually manages analog/digital output, calling a function for this adds memory and time overhead especially in real-time systems). Copy-paste? Pfff, similar code != this code solves the same problem, sometimes it's easier to include a change in "copy-paste" code than deciding how to include this in boundaries of "reusable" parts (ex. for the first time there're two structures X = (A, B) and Y = (A, B) with similar logic but different domains, and then manager decides to include specific logic for Y working with B data).

Now my general principle which distinguishes "good" code from "bad" one, is how good it resembles a model solving given problem. It can be a non-neccessary complexity of code (ex. instead of "if (isFinite(a)) { ... }" writing "if (a.toString() != "Inifinity" && a.toString() != "NaN") { ... }" when "a" is simply a number). And contrary not-enough complexity of code, relying on blind faith (famous XSS vulnerabilities, SQL injections in web apps are examples of that).