The MIDDLE project is still marching forward. Our most recent publication on ways to solve the firewall problem (and take a step closer to true distributed privacy) is Providing accurate models across private partitioned data: Secure maximum likelihood estimation, published in The Annals of Applied Statistics.