What Every Computer Scientist Should Know About Floating-Point Arithmetic

The thing is that although "big decimal" doesn't prevent you from rounding errors, it does help a lot because the number representation matches the mental model. Whereas, with floating point, the value is not always what is expected (0.1 is not represented as exactly 0.1 in FP).

So although I could use FP for many things, including money, my experience is that the cognitive cost of managing it is too much for many people (and no, I'm not considering myself better than the others)