Reviews: A Reduction for Efficient LDA Topic Reconstruction

I find the idea quite interesting, but I have the following concerns. First, this paper has many important parts missing and relies on other sources -- unpublished manuscript to show the equivalence of uniform LDA and STA, and "full version" for the general (p,t)-separable case. What would be this full version paper? As far as I know, NIPS conference papers should be mostly self-contained, except for some parts that rely on previous literature. While the appendix does include the unpublished manuscript, it is not required for the reviewers, and quite frankly this appendix is too lengthy and dense to review for accuracy.