Hi Ian,
Nice essay! Higher-level than the other ones you've written for this contest, but I think you've framed the more technical parts in a nicely accessible manner.
"The universe as a whole must not be physically deterministic!" Hear, hear! (Although I'm not fully persuaded by your argument, I agree with your conclusion...)
Some nitpicks: I'm not convinced that it makes sense to assign Shannon information to actual physical *signals*, rather than to the source itself. Sure, in certain special cases you can do it, but it hardly seems generally possible.
For example, you write:
> As Schumacher and Westmoreland note, "Information is the ability to distinguish reliably between possible alternatives" [16]. In this sense, information is encoded in the properties of the particles.
In the quote, they're talking about subjective distinction, which is in turn defined by *possible* alternatives, and therefore depend on the properties of the *source*, not any particular message. How do you conclude from this that information is now objectively encoded in physical particles that are themselves the signal? (Although I suppose you recognize this when you start talking about domains later... ) Still, my point is that "information content" is a dangerous notion, as it might lead one to falsely conclude that Shannon Entropy is a localized physical property that propagates with signals, like charge or mass.
I think you could have bypassed the seeming mismatch between Shannon entropy and classical thermodynamical entropy by simply pointing out they're somewhat different concepts. (Our uncertainty of classical systems *would* decrease if we kept track of past measurements and used a correspondingly finer-grain description; but that's not what we do when we define thermodynamic entropy.) As you discuss, the quantum analog is much stronger.
Now, of course I know that you have a research program to establish a connection between these two types of entropy, but I suppose that made me even more surprised that this issue never really came up. After all, what makes your research interesting is that these *are* inherently different notions and you're finding connections between them. If you assume they're the same from the outset, your research conclusions won't seem as strong.
Finally, even though I agree with your ultimate "bit from it" conclusion, I wonder how you'd address the orthodox-QM argument that unitary evolution implies that fine-grained "entropy" (in at least one of the above senses) never really increases at all. I guess one way would be to put in some formal indeterminism, as I do, but I've never seen you advocate that explicitly.
PS -- If you ever find yourself thinking of entropy as associated with spacetime regions rather than instants, please let me know -- I think there might be some overlap between our programs if you start going down that road.