As someone who spends most of my time at work writing firmware for embedded systems, I learned VERY quickly never to use floating point numbers when I could use

As someone who spends most of my time at work writing firmware for embedded systems, I learned VERY quickly never to use floating point numbers when I could use fixed point numbers. Better to use an integer number of thousandths of a thing than try to use floating point numbers and have error. This mindset is clearly common, because it seems like the vast majority of microcontrollers have no floating point hardware anyway. At some point I was listening to a talk where the guy called floating point numbers _"The incredibly misleading numbers that never really work the way that you want them to."_ That's probably the best description of floating point numbers I've ever heard.