Reviews: Causal meets Submodular: Subset Selection with Directed Information

There are really two issues with this paper. The first one is the discussion on how to use directed information for causality. Unfortunately, the word causality means two different things and these should be separated (and this discussion should be mentioned in this manuscript I think). The first concept of causality is really *prediction* of a time series using other time series. This prediction respects time and hence people call that causal.