Hi, Torsten, thank you so much for reading and commenting! I apologize for the late response, unfortunately, the end of the discussion phase of this contest coincided with a highly demanded period at work...
In my essay, goals are defined by variables that are restricted to relatively narrow ranges, at least, when compared with the variables defining the initial state, which are allowed to vary in a broader range. If the goal-seeking process does not reduce the range of possible values of the variables involved, then I can hardly think of the process as a goal. This definition ensures that the agent (the subject working to reach the goal) makes an effort to evolve in the right direction. This is why I make such a big emphasis on entropy reduction.
Darwinian evolution is one such process (although not the only one, I give some other examples). As I see it, your interpretation of how entropy varies throughout the process of mutation (entropy increases) and selection (entropy decreases) is correct. You are making an effort to follow the variations of entropy all throughout the process, which is nice. I am somewhat more lazy, I just look at the initial state (for example, all the proto-bats - mice, or similar animals - that existed before bats had evolved, and that had all sorts of hearing organs), and a final state (modern bats with sophisticated eco-location organs). Entropy decreases only because we define the goal as "the production of proficient eco-location organs". We, as observers, defined the goal.
Here it makes sense to underscore that out of all goal-searching systems, evolution has an interesting property: it does not know beforehand where it is heading to. When your or I have a purpose in mind, we know it beforehand, and precisely because we know it (we have a mental representation of what we want to achieve), we steer our actions in order to get it. So do robots, football teams, and even DNA replication (ok, the DNA molecule has no mental representation of anything, but the process of replication can be thought of a process whose purpose is ... DNA replication). Evolution, however, does not know where it is heading to, and moreover, no observer knows beforehand where it is heading to. One could even say that it wanders erratically. It tries a random sample of mutations, and whichever works, becomes the new road, even though it was not known in advance. Later in the process, we (outsider observers) look at a particular feature (for example, the eco-location organ of bats) and cry in bewilderment: "oh, this organ is soooo sophisticated!" We then look back at the process that gave rise to the eco-location organ, and cannot avoid thinking that it evolved *in order to* produce an efficient eco-location function. But such a goal can only be identified a posteriori. Indeed, only a posteriori was entropy reduced, because to calculate the entropy of the final state, we must know which is the desired range of variables. Only after we decided that the goal was to develop a morphology that allowed bats to achieve better eco-location abilities can we calculate entropies. Of course, only a restricted set of anatomies and physiologies can allow them to reach such a goal. But nobody knew beforehand that such was the goal!
You chose to discuss specifically the example of evolution, and I can well understand why: it is a particularly interesting example!
Thanks for the thoughtful comment. I will soon leave a post on your forum.
Best!
inés.