Thanks for your thoughts! Your biases are what I was looking for. I was hoping that with enough input there would be a consensus (e.g., "look around assertions are confusing to everyone"). Perhaps I should have asked what people have the most trouble getting right, what they find themselves fixing most often, or just what they use the most.

This all seems to favor a "score the tree" approach. As I think about that, I wonder if it makes sense to give the user some input into the scoring. That is, someone could say, "I can't ever understand code assertions, so they're of the highest complexity, but I write look-arounds in my sleep, so they're low complexity". On the other hand, it's supposed to be a tool of maintainability so it makes more sense for things to be scored as the typical programmer sees them.

Anyway, I appreciate you sharing your thinking on this.