In my experience it is almost always the case. I think this is a case where Dunning-Kruger* often applies, which is to say that you enter this new environment and think you know what's going on only because you're so completely ignorant of the details. Other times it's simply that decisions were made that made sense at the time and only look stupid with the benefit of hindsight. In either case, that bad code probably still exists in production for a very legitimate reason: the trouble it causes isn't worth the cost to fix it.
*I'm aware that DK doesn't really say what the common interpretation says it does
Another concept that complements this is the fundamental attribution error. It's called that for a reason. Instead of first assuming that everyone who came before was a terrible coder/worker/person, you have to look at the context and incentives at play.
*I'm aware that DK doesn't really say what the common interpretation says it does