Last week I had an idea very similar to the "cosmic immune system out to get us" concept expressed above, as a result of reading the book "The Life of the Cosmos" (1997) by physicist Lee Smolin, who paradoxically is an opponent of the anthropic principle, preferring his concept of cosmological natural selection as the explanation of why the laws of physics are as we find them. His is a many-worlds cosmology but the individual universes in the ensemble can reproduce further universes, a process accompanied by the introduction of small mutations into the laws of physics (the genome of a universe) that the progeny inherit, which causes the ensemble to evolve in a Darwinian fashion. (My only quarrel with universe evolution is that there seems to be nothing that can remove less-fit individuals from the scene and recycle their matter for use by the more fit, but we must avoid argument from lack of imagination.) Evolution, when discussed in the idiom of DNA-based genes, is driven by increases in gene frequency reflecting reproductive success. Therefore we predict that the corresponding universal process tries to maximize the fecundity of the typical universe. Take-home: we happen to see a specific physics not because it produces the most observers but because it produces the most universes. (Since "universe" denotes all there is, the word is clearly being misused here. Therefore the units of cosmological selection will be referred to as "compartments" in what follows, and the word "universe" will retain its dictionary meaning.) Abandoning the anthropic principle allows us to view the rise of civilizations of intelligent observers in any compartment as no more than accidental, and probably representing a kind of cosmological cancer. Other suitable metaphors would be "parasitic process" and "failure mode". Examples of these abound, such as feedback noises developing in public address systems, epileptic seizures developing in cerebral cortex and/or hippocampus, or any pathogen-triggered disease in the human body. I am sure that this problem has been regularly impairing compartment reproduction all down the cosmological line of descent leading to us, causing a selection pressure that long ago brought about the evolution of a compartmental immune system, which, of course, is out to get us. But how could little ole Earth threaten the whole observable universe? By someday colonizing the Galaxy in finest Science Fiction style and expropriating the interstellar matter the host compartment needs to reproduce, to make new planets for us to live on. Interesting that the universe thinks we can do it. I guess immune attack can be the sincerest form of praise. The compartmental immune system has in fact probably been taking preventative measures against us all throughout our history, leading to the myth of Old Nick, Father of Lies, The Evil One, etc. It cannot even be dismissed as a myth, it's merely what in science would be called an approximate theory. It appears that the compartmental immune system consists of “mousetraps” for the unwary civilization (perhaps baited with goodies like fossil fuels, the use of which is calculated to Venus the planet in a runaway greenhouse effect that strikes without warning, and uranium, whose dangers do not need repeating), Shminux is obviously all over this one. The traps would be encoded into the laws of physics with diabolical subtlety, and technological civilizations exactly such as ours have always been the target. Based on this line of reasoning, I am calling for the creation of a branch of science that could be called Defensive Cosmology, to help us spot these traps before it is too late. Who's with me? P.S.: lest you think this worldview is overly pessimistic, I'll mention that there is a place for God in it. He is an advanced technological civilization in another compartment that contains ours, that has repurposed our compartment as an experimental test bed for doing experiments on how to defeat their compartmental immune system. Notice that they cannot solve their worst problem before they have solved our version of it. Our interests are aligned. Nor must they give us the disease in order to try curing it, as we do with lab rats. That would be totally sinister. It would also be redundant, because all compartments always naturally come with immune systems. It remains to be explained why solving our problem for us would be any easier than trying to solve theirs directly. Here is what I think could be the reason. The entropy, or information content of a black hole has been shown to be proportional to the area of the event horizon, thought to be a kind of abstract surface. But what if it somehow has a claim to being a literal surface? If black holes are the daughter compartments our compartment has produced, as Smolin suggests, then we can unpack this to mean that these daughter compartments are flatlands! The contents are all on the surface which is locally two dimensional. Everything in these compartments is a total open book to us. We can intervene absolutely anywhere in these worlds, and see absolutely anything in these worlds. Wouldn't you say that that is an experimenter's dream? We are to God as these flatlands are to us. I can reconcile this we-are-a-surface idea with the principle of relativity and the need for a living compartment to have internal motions, if anyone is interested.
Concerning your first point, that the designer has to hand-insert that all-important sign bit. So how do humans come up with these sign bits? I imagine a trial-and-error process of interacting with the controlled system. During this, the person's brain is generating an error signal derived directly or indirectly from an evolutionarily-fixed set point. While trying to control the system manually using an initially random sign bit, I suppose the brain can analyze at a low level in the hardware that the error is 1) changing exponentially, and 2) has a positive or negative slope, as the case may be. If the situation is exponential and the slope is positive, you synaptically weld the cortical representation of the controlled variable to the antagonist muscle of the one currently energized, and if negative, to the energized muscle itself. Bayesian inference would enter as a Kalman filter used to calculate the controlled variable. I suppose the process of acquiring the sign bit of the slope could not be separated from acquiring the model needed by the Kalman filter, so some kind of bootstrapping process could be involved. In his book "Neural Engineering..." (2004), Chris Eliasmith makes a case that the brain contains Kalman filters.
Is the evolutionary process responsible for the original hard-wired set point itself a controller? I doubt it, because, to use Douglas Adams' analogy, control principles to not seem to be involved in getting the shape of a puddle to match that of the hole it's in.