For the purpose of selecting a DOM node in a performant manner, you really don't need to know how CSS/XPath traverse the DOM. You just need to know that querySelector exists.
In general, I think it's enough to know what your abstraction does, rather than how it does it.