There is no broader context wherein natural selection can be considered to be an optimization process, that is a pernicious misconception of evolutionary theory. Fortunately, people with a computer science background have a distinct advantage towards correcting this fallacy, because their training affords them an understanding of information as a working concept that lay people rarely attain.
The key insight is that any algorithm implementation for a process which has an objective must, as an absolute minimal requirement, possess an encoding of that objective in its implementation. That is, a real representation of the goal must be in the process's make-up so that the goal can be pursued at all, because correct navigation requires assessing actions for whether they work towards the goal or not, and any such assessment requires meaningful reference to the goal. Without such a definition to refer to, differentiation between desirable and undesirable outcomes is impossible.
This goal encoding may be explicit (ie readily understandable by observers studying the implementation) or implicit (hard to parse), but either way, it must be instantiated in the make-up of the implementation, in some medium with the capacity to hold the goal definition, ie a way of storing the requisite number of bits within the implementation itself (or readily reading it from elsewhere, or constructing it from some combination thereof). This definition of the goal must be implemented in a manner that can be read and acted upon by the rest of the algorithm implementation, so that the system as a whole can pursue states that better match the goal. ie so that it can optimize.
With regards to evolution, how could nature select without having an idea of what it was selecting for? A reference definition of fitness must be available to nature if it is to measure each individual organism's fitness and select accordingly.
For a natural-selection-as-optimization-process algorithm implementation, there would need to be a component that encodes natural selection's optimization objective into the implementation's very make-up (or a ready way to read that goal from an external source).
What is the make-up of the natural selection algorithm's implementation? It is the entirety of nature itself, in whole and in part. Nature is literally everything in the universe, and literally anything in the universe, from the most massive galaxy to the smallest particle, can participate in natural selection events. And no part of nature, save for some animal brains, seems to contain a representation of a goal for natural selection.
Is it even conceivable that everything in the universe, down to the smallest particle, could encode a common goal? Does a volcano encode the goal of maximizing reproductive fitness for the populations living around it? Can a shower of cosmic rays encode the goal of making sure the creatures who's DNA it disrupts are the ones who should be removed from the populace? They don't appear to encode any such evolutionary goals, nor do they have the capacity to maintain any goal at all beyond following the physical laws of matter -- Volcanos are disordered piles of rock and churning lava, and cosmic rays are singular fundamental particles that are subject to wholesale transformation with every impact -- neither has any way of encoding a common objective for natural selection, nor is there evidence for them being able to collectively maintain one.
We can illustrate the paradox of an optimizing nature using your water molecule analogy. A collection of water molecules acting under a gravitational field will demonstrate downwards fluid dynamics which single molecules in space would not, but no matter how much H2O you put together, it will never spontaneously develop any concept of evolutionary fitness.
And yet a flash flood is a very real natural selection event that can reshape the genepool of a coastal town, but all the same it has no means of representing any goal of optimizing the population's fitness through who it drowns and who it spares; its just water. Flowing water performs natural selection, but it isn't optimizing for any goal, no matter how you try to spin it, because it has no way of maintaining a representation of a goal in its disordered and inconstant structure. It flows, yes, but it has no goal in doing so, its not pursuing any optimization objective, all the while it is a real instance of natural selection. It doesn't have or need any way of determining who is more or less fit than another, so how could it be optimizing for it? It's just flooding.
Whether its by deluge, an erupting volcano, a congenital heart attack, or a pack of rabid dogs, the processes making up natural selection events do not possess an encoding of a goal for natural selection. They do not possess the necessary information structure required to pursue a common optimization objective, and so they cannot be optimization processes in any meaningful sense.
> The key insight is that any algorithm implementation for a process which has an objective must, as an absolute minimal requirement, possess an encoding of that objective in its implementation.
I don't agree with this in any way, or perhaps more accurately, I don't agree that we know (and perhaps could know) the scope of the implementation even if this claim was true, which I don't think it is.
The idea that "people with a computer science background have a distinct advantage" is also plainly wrong to me. I have a background (as in, I quit my PhD in) computational biology, have been a software engineer for more than 35 years, and there are just as many people with as without computer science backgrounds who fall for the fallacy.
What part of it don’t you agree with? That an algorithm implementation must encode the goal that it pursues? How can something pursue a goal it has no access to a definition of? If you have an alternative way it could work, please propose it.
I’m not asking rhetorically, I’m truly interested in learning the flaws in my argument for why natural selection cannot be modelled as an optimization process. So if you have the time to reply with a more detailed rebuttal, I’d much appreciate it.
edit: Addendum: I recognize my claim that computer scientists might have an advantage in understanding this is contentious, and I was not implying that they (we) as a group have a better record of understanding evolution’s subtlety than biologists (which I studied in uni) or the average lay person. I just think they could have an advantage in understanding the version of the argument that I gave above, and I am interested in improving it for that purpose.
What is the algorithm implementation when it comes to the physical world? Does the implementation extend to remote galaxies? Is the strong force part of the implementation? We don't know ... there appears to be no way to know.
But even if you could know, it is just demonstrably wrong that the implementation must encode the goal. If you create selection pressure, and have a reproductive system that allows for mutations, then you may end up an "implementation" that encodes the goal implicit in the selection pressure. But anyone who messed around with genetic algorithms or artificial life in the 90s knows that you can trivially start out with no resemblance to "the goal" at all. Where life on earth in aggregate or any specific example of it in particular might be along that pathway is similarly impossible to say.
Finally, even defining "the goal" is tricky. Consider the well-documented case of moth evolution in industrial (and later, post-industrial) northern England. Their camouflaging wing tones changed to respond to the typical color on vertical surfaces, twice within a human generation or three. Was "the goal" flexible coloration across generations, or was it "light, then "dark" and then "light" again? That's a philosophical question as much as anything ...
> What is the algorithm implementation when it comes to the physical world?
It is the physical world, nature is the implementation of the natural selection algorithm. Yes, the strong force is part of the implementation, because the strong force can play a role in selection events, cf nuclear bombs and radiation. The gravitational pull of remote galaxies can also influence selection events by changing planetary orbits minutely.
I don’t see these as problems for my argument because I am not the one claiming they encode an objective, I just see them as natural forces which can influence selection without any overarching purpose or goal. It is those claiming natural selection is an optimization process who must show how it could work. The onus is on them to show where their supposed objective of natural selection is encoded in its implementation.
> If you create selection pressure, and have a reproductive system that allows for mutations, then you may end up an "implementation" that encodes the goal implicit in the selection pressure.
What goal are you referring to?
> But anyone who messed around with genetic algorithms or artificial life in the 90s knows that you can trivially start out with no resemblance to "the goal" at all. Where life on earth in aggregate or any specific example of it in particular might be along that pathway is similarly impossible to say.
I am one of these people, but I don’t know what goal you are saying these systems came to demonstrate. Are you saying these artificial evolution systems had objectives they pursued? What caused them to follow these objectives? What is this “pathway”?
>Consider the well-documented case of moth evolution in industrial (and later, post-industrial) northern England. Their camouflaging wing tones changed to respond to the typical color on vertical surfaces, twice within a human generation or three. Was "the goal" flexible coloration across generations, or was it "light, then "dark" and then "light" again?
There was no goal at any point in the process. Moths with colors that matched their contemporaneous environment were less likely to be eaten by predators than those which stood out. Calling it a goal is a confusion, its trying to add a conceptual framing that isn’t necessary and adds nothing to the understanding of the system. Neither the soot levels in the air nor the birds hunting for moths have a goal of adjusting the balance of moth coloration phenotypes. They are just the context, along with everything else in their environment, in which evolution of moth coloration may occur.
In what sense is there any goal in the example? And if it is a goal, why is it not optimization? I claim there is no goal, no optimization objective to natural selection. Its not just a philosophical side question, it is the question.
> It is the physical world, nature is the implementation of the natural selection algorithm.
But you don't know (and to some degree, cannot know) which parts of it. So you cannot really know if the implementation encodes a goal or not.
> What goal are you referring to?
Whatever goal was being used in the case of genetic algorithms or artificial life systems. Those systems have goals, but the early stages do not embody the goal in any way you could recognize.
> There was no goal at any point in the process.
So in the case of natural evolution, we happen to agree. However, I don't agree with your claim that "the implementation must embody the goal" is a useful way to think about this, and I also have some sympathy for the idea that there could be huge-time-scale teleology associated with evolution that we cannot discern.
Its conceivable that the universe could encode a goal somehow, after all its so vast, but that conceivability alone is not evidence for the existence of an encoded goal any more than the conceivability of extra-terrestrial intelligence, or of a higher design to reality, is proof of their existence. What science tells us is that the only goal nature seems to embody is following the physical laws we've been able to determine, and nothing more. I'd apply the same interesting hypothesis status to huge-time-scale teleology that we cannot discern, and perhaps it is both real and we will never be able to discern it. Personally I find the notions very interesting, but I don't see reason to believe in them. If there were good evidence for them, they'd be the subject of scientific study already.
But we seem to agree that natural selection doesn't have a goal. In my observation, any purported overarching goal that is ascribed to natural selection, including the measure of inclusive fitness[0], can be reduced to some function of the context in which it is being observed, like moth coloration was influenced by soot levels.
As to my main claim, I do believe it is necessary that an encoding of a goal is necessary for choice among actions in pursuit of a goal, because some kind of reference to a goal is necessary to compare options in a decision algorithm. In the case of a-life systems which have goals, that encoding is somewhere in the algorithm of evolution rules combined with the initial state of the simulation. In the case of nature, I don't see a place where that encoding could exist, except the trivial "goal" that all elements will follow the laws of physics.
Please note though that I never put it that "the implementation must embody the goal," I was more careful with my language by saying that it must have an accessible or working encoding of the goal, one its decision process or evolution rule would need to reference in order to make decisions that favored it. The encoding need not be internal (so embody is definitely not necessary), and none of these things are necessarily explicit or well partitioned (e.g. an evolution rule can implicitly encode a goal).
edit: addendum: [0] On inclusive fitness being reducible to situational factors, I'm just following the direction of M.A. Nowak, C.E. Tarnita and E.O. Wilson on this: https://www.nature.com/articles/nature09205
The key insight is that any algorithm implementation for a process which has an objective must, as an absolute minimal requirement, possess an encoding of that objective in its implementation. That is, a real representation of the goal must be in the process's make-up so that the goal can be pursued at all, because correct navigation requires assessing actions for whether they work towards the goal or not, and any such assessment requires meaningful reference to the goal. Without such a definition to refer to, differentiation between desirable and undesirable outcomes is impossible.
This goal encoding may be explicit (ie readily understandable by observers studying the implementation) or implicit (hard to parse), but either way, it must be instantiated in the make-up of the implementation, in some medium with the capacity to hold the goal definition, ie a way of storing the requisite number of bits within the implementation itself (or readily reading it from elsewhere, or constructing it from some combination thereof). This definition of the goal must be implemented in a manner that can be read and acted upon by the rest of the algorithm implementation, so that the system as a whole can pursue states that better match the goal. ie so that it can optimize.
With regards to evolution, how could nature select without having an idea of what it was selecting for? A reference definition of fitness must be available to nature if it is to measure each individual organism's fitness and select accordingly.
For a natural-selection-as-optimization-process algorithm implementation, there would need to be a component that encodes natural selection's optimization objective into the implementation's very make-up (or a ready way to read that goal from an external source).
What is the make-up of the natural selection algorithm's implementation? It is the entirety of nature itself, in whole and in part. Nature is literally everything in the universe, and literally anything in the universe, from the most massive galaxy to the smallest particle, can participate in natural selection events. And no part of nature, save for some animal brains, seems to contain a representation of a goal for natural selection.
Is it even conceivable that everything in the universe, down to the smallest particle, could encode a common goal? Does a volcano encode the goal of maximizing reproductive fitness for the populations living around it? Can a shower of cosmic rays encode the goal of making sure the creatures who's DNA it disrupts are the ones who should be removed from the populace? They don't appear to encode any such evolutionary goals, nor do they have the capacity to maintain any goal at all beyond following the physical laws of matter -- Volcanos are disordered piles of rock and churning lava, and cosmic rays are singular fundamental particles that are subject to wholesale transformation with every impact -- neither has any way of encoding a common objective for natural selection, nor is there evidence for them being able to collectively maintain one.
We can illustrate the paradox of an optimizing nature using your water molecule analogy. A collection of water molecules acting under a gravitational field will demonstrate downwards fluid dynamics which single molecules in space would not, but no matter how much H2O you put together, it will never spontaneously develop any concept of evolutionary fitness.
And yet a flash flood is a very real natural selection event that can reshape the genepool of a coastal town, but all the same it has no means of representing any goal of optimizing the population's fitness through who it drowns and who it spares; its just water. Flowing water performs natural selection, but it isn't optimizing for any goal, no matter how you try to spin it, because it has no way of maintaining a representation of a goal in its disordered and inconstant structure. It flows, yes, but it has no goal in doing so, its not pursuing any optimization objective, all the while it is a real instance of natural selection. It doesn't have or need any way of determining who is more or less fit than another, so how could it be optimizing for it? It's just flooding.
Whether its by deluge, an erupting volcano, a congenital heart attack, or a pack of rabid dogs, the processes making up natural selection events do not possess an encoding of a goal for natural selection. They do not possess the necessary information structure required to pursue a common optimization objective, and so they cannot be optimization processes in any meaningful sense.