A nice paper on a structured variational approximation for Cox processes with sigmoidal link functions. The key challenge is to represent the variational posterior over a set of thinned events with unknown cardinality. The key simplifying assumption is that the M thinned events are iid from a mixture of truncated normals; this seems like a strong assumption that should be discussed in greater detail in the main text. Please also take the reviewers' suggestions into account and follow through on the promises made in the rebuttal.