It's not that optimization is important, it's that understanding computational complexity is important. Why? Because it tells you when some kinds of optimizations might help, and when they likely will not. It helps you not write inefficient code in the first place, or it lets you use a simple algorithm when it will be sufficiently fast for your needs.

If you have a basic understanding complexity you almost certainly have seen and can more or less read the notation, as you probably needed it in order to develop that understanding of complexity in the first place. Whether you really need to understand some of the finer details, if you should have a tattoo of the Master theorem, well, I don't know.

("You" being whoever, not you).

Your comments appear to be critiquing the existence of the notation itself. I'm not sure what you expect the alternative to be. Words are used to communicate concepts.

Knowing the complexity of algorithms is important, correct? How would you go about teaching something like that to a student? Limit yourself to generic terms like linear vs exponential? Would that student have as nuanced an understanding of the subject compared to if he or she learned Big-O notation?