Improving legacy code always means to understand business logic and if possible the motivation behind it. Then you have a replacement built and you can justify why it is better.
Shall I go on?
"laws aren't broken" is the kind of thing someone says when they've never seen "justice" in action, or are walking around with their eyes shut.
"Legacy" code that only "occasionally works" doesn't sound like something worth keeping around. It sounds like something to dump in the trash and never look back.
If you have working legacy code though, even if it occasionally fails because it's full of bugs and fundamental flaws, should be replaced very carefully, and only with in-depth understanding of what is working and what is broken about the existing system.
Based on the broken laws you list, I assume you're talking about America. American laws are indeed very broken, but dumping it in the trash and never looking back would be a one-way ticket to a bloodbath. It is both important very important to fix our very broken laws, and important to do so very carefully and only with an in-depth understanding of what is working and what is broken about the existing system.
If I were to give you as much benefit of the doubt as you give your parent, I'd say that "occasionally laws work" is the kind of thing someone says when they've never seen how much worse it can get.
Perhaps you mean: they are full of bugs, fragile, broken, but occasionally they work?