I am interested in bringing our understanding of $Q$-curvature up to the level we have for scalar curvature, in attempts at better understanding the Ricci tensor. I believe that fourth-order and semilinear is better than second-order and fully nonlinear.