I admit I didn't follow this as much as I'd like to, and I wish it had some discussion of how this maps onto traditional signal processing frameworks in statistics and information theory, if it does.