Professor Ellis,
I have addressed some of the questions raised on my submission page over there. I will address the rest here. Think there are some fundamental differences on some key ideas/definitions that I would like to differentiate so that we can clearly draw our intellectual battle lines so to speak :)
1) The key difference between the logic of physics and biology in your essay seems to be the presence of a context in biology that is seemingly missing in physics (Am I still missing something else here?). I would argue that for a system S, C is like any other physical system that influences it's state, and the 'context' or lack thereof is derived from the interactions (governed by physical law) between S and C. For example: in the case of the microbe (S) and poison (C), it is the specific interaction between S and C and the action that S performs or doesnt perform due to that interaction, that imbues C with it's context with reference to S. The context of the physical system C to S is not something determined apriori. Of course if the system is capable of memory and learning, then it can use that to remember the poison and it's actions for a future situation. I would further argue that "If...then...else" like logical statements can be achieved in hardware in computers, that can be purely described using physics based statements.
What type of system S can recognize the context that is has imbued C with because of its interactions? I think that is an important question that needs to be worked out.
2) (This is a repeat of what I wrote on the other thread) The agency I am talking of (takes the definition from philosophy and it) is simply the capacity to act. To act involuntarily, unconsciously or consciously with a purpose will all fall under it. The moon with the ability to act on earths waters makes it an agent, but doesn't have to fall under the category of making it a purposeful one for the moon or even tht the moon is capable to generating its own purpose like we do. It is very possible to think of physical systems that have no agency-can change their state based on the influence of external systems but do not have the ability to 'act' and affect its environment. It is thus possible to have systems that have 'agency' as I define, but not have a purpose or intent for that agency. I would point out that it is not just hierarchical systems that have a sense of agency, but minimally dissipative systems (from my essay) whoses dynamics can be achieved in an hierarchical predictive coding model that will be capable of a sense of agency. The sun, earth, etc will not satisfy the condition of being minimally dissipative.
3) The complexity constraint is using a statistical complexity measure like the mutual information between the system and all the inputs in the past that has influenced it's state. The finite complexity constraint is necessary for I am dealing with finite state automata models with an emphasis on finite and this constraint would ensure I dont end with trivial solutions like 'have an infinite number of distinguishable states and remember everything'. Furthermore the problem is mathematically formulated so that the tradeoff parameter in the optimization problem beta can be moved like a knob to play around with the complexity and see how that affects external input-system correlations. While it is possible to have both extremely complex structures or minimally dissipative systems that do not learn, we can see that the type of learning dynamics we see in certain biological systems is a tradeoff between the dissipation and complexity, parametrized by beta.
4) "If it is adaptation, there is some selection principle in action which cannot be captured simply by the idea of dissipation." England's idea is that those selection mechanisms themselves are instantiations of larger thermodynamic dissipation principles. There are of course some caveats there in his hypothesis and there is much work to be done to make it a more developed proposal.
Think I have addressed most of the important issues you raised here to the best I can over a post. Appreciate your questions. I have learnt that I need to be a lot clearer with some of my definitions.
Natesh