I think the tldr; of what you said is that even when you have a theoretical handle on the growth function, you still need to actually benchmark anyway