Need to think about this one for a while, I think the author is onto something, but I need to understand entropy a little better before I can assess this:
When we're working with binary states entropy isn't a 1 or a 0. Its definition is instead related to the amount of information we can hold versus how much we're actually holding. Same is true for a ternary state. The catch here is that we only have access and effects from two states. We basically have 1 and +0 and -0. With no way to tell if a zero is negative or not. We've altered entropy's definition here. Since no information actually ever becomes entropy in actuality. It appears as if its entropy whenever there's a 0. That's because each "bit" doesn't store 2 pieces of information, it stores 3. But we can only access the 1 and 0 difference so we can only read 2 pieces of information. We "lose" 1 piece whenever it goes to 0. The definition of entropy here has shifted towards defining any 0 as a piece of entropy. This model assumes information is never created or destroyed. Just rearranged, hidden, and unhidden.