A couple of notes. You state that the plan is read from right to left. That's not completely true. The logical processing order of the plan is from left to right, like reading a book. The data flow is from right to left. Frequently you're going to follow the data flow, but there are some plans, especially ones involving temporary storage in spools, that can only be appropriately understood by reading them left to right. That was a mistake in my first book that I fixed in the second edition.
You state that operator costs are estimates, good. But then you go on to implicate that they have direct correlation to performance and that's just not true. They're internal measuring sticks of meaning only to the optimizer and don't relate to the real world in any significant way. I'd be as clear as possible there. That's another thing I didn't make sufficiently clear in the first edition of my book.
Decent summary of how to read execution plans. Well done. "
Thanks again, Grant, for the encouragement and clarifications.