A hierarchical renormalization model: some properties and open questions