Deep Learning is the end-to-end solution to all our problems. Now that we know that, we should focus on what remains: the beautiful and elegant theory behind learning (whether it is deep, reinforced, statistical, online, interactive, supervised or not, Bayesian, etc.). Joke aside, there is a large part of the learning theory community that is attending NeurIPS. This event is purely social, a meeting point for people from the learning theory community in the NeurIPS maze and crowd. The objectives are not to be fancy, far from it (theory is not fancy ! or is it ?) but just a gathering of colleagues/friends sharing the common interest in rigorous and beautiful proof, rates of convergences and obscure theorems.