Some reflections on the future of computational chemistry