Dear Sara,
Your submission is surely one of the best-written and deepest entries in the contest. Any one of a number of topics you discuss (self-reference, distributed causation, causal efficacy of information, evolving dynamical laws, etc.) is worthy of more comment and discussion than can be accomplished in a forum such as this; hence, I can only choose a few points to remark on.
1. I am wondering if your rejection of the "fully reductionist picture" (page 2, last paragraph) really applies to the classical level, or only to the quantum level. I ask this for two reasons. The first reason is because digital computing (which is purely classical) has so far failed to produce many of the "lifelike" qualities repeatedly predicted by scientists in fields such as artificial intelligence, and it still seems reasonable (to me) to ask if life is intrinsically quantum-mechanical. The second reason is because classical holism in the strong sense seems to raise a host of problems (see subsequent points).
3. I have severe difficulty with the subject of "top-down causation" in classical physics (quantum holism is a different story). I have thought about causal structures a great deal in the context of quantum gravity and the fundamental nature of "spacetime," and it seems to me that very radical conclusions arise when one attempts to make the notion of classical top-down causation precise. I can't explain all the details in a single post (if you are interested, you may see my essay here), but the main problem involves the nature of time.
A short explanation is as follows: if causality is fundamental, then it seems reasonable to assume that the arrow of time aligns with the causal direction. This is already evident in special relativity, where causality can operate only over timelike intervals. Given an event E in spacetime, the set of all events that may be influenced by E is the "future light cone of E." Given a subset S of spacetime, the "future" of S is the union of all the future light cones of the events in S.
Now, if one admits classical top-down causation, this picture becomes vastly more complicated, provided that one "really means it," and is not merely saying that it is useful to lump events together into systems and talk about aggregate effects on other systems. By "really means it," I mean that one assumes that subsets of spacetime exert influence on other subsets independent of the influences between their respective events. This replaces the connection between the "future" subset of spacetime and the scope of causal influence by something much more complex. One must construct a "space" with one point for every subspace of spacetime (the "power set," in mathematical terms). Cause and effect is now described in terms of relations between points in this huge new space, and if time aligns with the direction of cause and effect, time can only be understood in the ordinary way as a one-dimensional "direction" in the power set, not in the original spacetime.
4. Regarding the causal efficacy of information, I have no doubt that something like this occurs in living organisms, but it seems more and more plausible to believe that all of physics works this way at a fundamental level, so it seems to me that it may ultimately be hard to distinguish life on this basis. Of course, this is too superficial an indictment of points you are making here, but something like it may still be relevant at some level.
5. Alain Connes, simultaneously one of the world's greatest mathematicians and physicists, has written an interesting book called Triangle of Thoughts in which he discusses the way in which humans experience time and how it differs from the linear picture of standard physics. If one did adopt the most radical interpretations of the ideas you have presented here, something like Connes' ideas might be considered more than metaphorical.
Congratulations on a splendid contribution! Take care,
Ben Dribus