Let me share a little secret of the training industry with you: an “advanced” topic is any topic that you don’t already know.
Don’t argument is that the proper axis is around commonality of usage: most commonly performed to least commonly performed. It’s an interesting argument for sure.
I’m of two minds with the idea, however. I appreciate Don’s example and like the concept of commonality for differentiation. But there are things which are legitimately advanced topics, in that they would be difficult to understand even if they were common. In Don’s query tuning example, an example of something legitimately difficult to understand is the set of rules the query optimizer chose to test for a particular query. Yes, it is very uncommon to need to know this, but it is also difficult to understand if you do need to know, and explaining how and why the query optimizer chose the path and rules that it did requires a fairly deep base of expertise.
In short, I think there’s an endogeneity problem: things can be perceived as difficult because they are unommon (which is my reading of Don’s point), but also things can be uncommon because they are difficult to understand given some baseline of knowledge.