Certain ideas naturally resist being encoded into a given medium of communication; epistemic communities built around that medium therefore fail to appreciate the ideas which that medium cannot successfully communicate.
Awesome post. My sense is that the cybernetic perspective fails to ever take hold because the things we’re trying to study are just really fucking complicated. You end up sounding like a madman, or gesturing toward a general vibe that readers may or may not identify with. For me, the conclusion of every analysis ends up being “this is super complicated and these complications are not widely understood or appreciated”.
Not to say one can’t go farther than that, but it’s simply really hard and takes monumental effort to either articulate the mechanistic breakdown of interacting processes, or to write about them well enough to sound intelligent.
There’s just not a lot of room in intellectual meme space to move our understanding of a complicated topic from 3% to 4%… one is always incentivized to reduce the problem to such a degree that you can say you understand 90-100%, having lost sufficient perspective to actually say anything interesting.
You’ve at least managed to articulate this phenomenon super well, as the mimetic environment of intellectualism is simply inhospitable to certain important ways of thinking.
Way to stick the landing. Your teaser got my hopes up, and you delivered. You have convinced me to stop talking about pragmatism–such an awkward term– as a source for cybernetics and talk instead about process philosophy.
Since you centered on Bergson, I have to plug Emily Herring's fantastic "Herald of a Restless World: How Henri Bergson Brought Philosophy to the People."
Your gesture to Dewey and James (he of the stream of consciousness) leads me to recommend Charles Peirce to this genealogy. It always feels like he is on the verge of blowing up, like Bergson is lately. Thanks to how well Pierce's semiotics help make sense of generative AI, it may be happening.
I'm seeing some mentions of Pierce coming up in contemporary social science methodology -- felt like he was less of the lineage in this piece, or at least that he hated being associated w Bergson. I havne't read Herring's book but a good friend did and told me about it, so parts of it might be in here
Perice hated being associated with anyone, except maybe James, because James was always so nice to him.
I'm halfway through Leif Weatherby's Language Machines, and am thinking about writing a review centered on his positioning of Peirce. Two other recent books that bring Peirce to bear on generative AI are Paul Kockelman's "Last Words Large Language Models and the AI Apocalypse" and David Gunkel's "AI for Communication."
If you or anyone digging through your comments has other recent examples of a Peirce revival, especially Peirce helping make sense of the outputs of LLMs, I'd be grateful for the pointer.
Ok, now I really need to read Language Machines...
and, not about LLMs, but here is Peirce's "IBE" applied in by methodologists in a top poli sci journal...I understand that sociology has also seen a revival of interest in pragmatist phil of science https://www.journals.uchicago.edu/doi/abs/10.1086/734280
Pragmatism was an awkward term because you were misusing it... Jamesian Pragmatism is a philosophy of the mind of the continental kind. It's subject is the human spirit's inquiry into whether, when, and which things matter. It's not a methodology, and is a really strange choice as a 'source for cybernetics'. If you were to go continental, Freud would have made a lot more sense. Cybernetics has the best chance of making sense (a long shot in any case to be honest) when it focuses tightly on recursion and feedback loops. Basically, you make a system Alive by giving it the Compulsion to Repeat that Freud outlined in Beyond the Pleasure Principle.
I would say pragmatism is a method for avoiding interminable fights over the definition of terms:
The pragmatic method is primarily a method of settling metaphysical disputes that otherwise might be interminable. It is to try to interpret each notion by tracing its respective practical consequences.
This is one of the more pragmatist definitions of pragmatism I've heard! Phrased in terms of what it's for rather than what it technically means. And maybe in some cases, this includes what it needs to be distinguished from. Sometimes I like to think of pragmatism a bit like a genealogy of the future, as opposed to genealogy of the past.
Also, fights over definitions cant be 'interminable' (to my first point, what an *extremely* Latinised way to say 'endless'). Either we end up agreeing or the thing falls out of use ... plus, that dialectic teaches us about the world. If all the things we might use pragmatism for, dialogue-murder for the sake of using methodologies instead of our bodies to think may well be the least in the spirit of its original thinkers
To be clear, “interminable” is James’s word, chosen for that particular definition in “What Pragmatism Means” though he did use “endless” to make similar points elsewhere.
I remember. But James was writing for readers who knew Latin. I have a whole thing about this. Anyway, I say what James did: its a method not a methodology. I just think he was being weird with the 'settling disputes' bit because it can be hard even for him to find the right words. Disputes was an odd choice for talking metaphysics. But like I said, all this is karma for not picking Freud instead at the start
Since you answered my mostly Greek and Saxon English with nearly only Latin English, we may as well be talking in two languages. Still, I'll try answer that since I know both.
Yes pragmatism is a method, but its not a methodology. Its a razor for thinking about meaning. I cant think of any metaphysical disputes either James said they wanted to settle. If anything what I like about pragmatism is its willingness to say about some of those disputes: ya know what, maybe this doesn't need settling, actually.
1. The most provocative - perhaps Cybernetics can't be grounded in Western Philosophy at all, but most naturally fits inside a combination of Buddhist/Daoist philosophical ideas.
2. There's plenty of Cybernetic writing, however, it's on LinkedIn rather than Substack and not exactly a genre you want to emulate.
3. To echo Russell about Plato: the few people who read Whitehead (and now Deleuze/Stengers/Latour) don't read Weiner (or von Neumann or Rene Thom) and vice versa. Unlike AI, where cognitive scientists, philosophers and technologists read each other a lot.
4. Continuing the previous point, Cybernetics lacks vivid thought experiments like Searle's Chinese Room, and instead has indecipherable diagrams.
Your point 4 is really what I'm trying to get at in the essay -- but the question is *why* Western philosophy finds those thought experiments so compelling. Perhaps I should've gone further down this road, but I think that certain games are the proper cybernetic analogue to thought experiments.
Point 1 deserves a much longer response, but yes this resonance (and even explicit intellectual cross-pollination) came up a lot in my reading. Yuk Hui makes the connection with Daoism and certain strands of Confucianism very clear, and I'm struck by how many of the early cybernetic thinkers had some exposure to "the East." Wiener visited China and Japan in the 1930s, and both Stafford Beer and Ross Ashby were stationed in India during WW2.
There's a now forgotten (makes your point!) literature on Systems/Cybernetic theory and Buddhism. Joanna Macy's "Mutual causality in Buddhism and general systems theory" is the best known example of that genre. Then again, there's a 'Buddhism and X' literature for whatever is the intellectual fad of the day.
Did you read Richard Seaford? He wrote about the birth of both Indian and Greek Philosophies. He argues that coinage gave way to philosophy in Greece and in India because it involves the changing of a substance (gold, silver, or both mixed) to turn it into blocks that have a fixed meaning on economical value.
Ok, so… I am doing causal discovery on physical systems (mostly astronomical). The main objection I hear when presenting my work is that while DAGs capture static (and acyclic, by construction) causation relations, dynamics is made up of feedback loops. Coupled systems of ODEs are already good at modeling those: are you sure we need DAGs? Meanwhile my wife, who is a regional economist, just presented a conference paper on the causal determinants of innovation based, again, on causal discovery applied to firm survey and patent data. The main objection she received is that “DAGs maybe are not for every discipline” and “how will you sell this?”. To each their own kind of antimemetics I guess…
oh yeah it's fun seeing statistical methodology across different disciplines....they have such different intuitions!
But, fwiw, my position is that causal inference (whether DAGs or potential outcomes or whatever) is an anti-cybernetic perspective. For better and for worse, depending on what we're trying to do
As I am facing concrete challenges when applying these methods to physical systems where causality often is indeed cyclical, I am very interested in your take
yeah so the tension is -- rigorous causal inference is the only serious way to get at one element of the causal chain. It is in that sense necessary...but this means throwing out the feedback/recursion that actually makes the thing tick.
Most of these problems, at least the ones I've encountered, are just beyond our ability to understand with the degree of rigor we've come to demand
Well yeah we need exogenous variables to be able to identify causal relationships via IV/experiments, but nothing is really exogenous in a closed system where everything interacts. If this is the kind of tension you are getting at, is there a systematic exploration of it in the literature?
Yes that’s the tension—and no I don’t really think people know how to deal with it. Most of our problems are unanswerable with the degree of rigor we’ve come to demand
You may have mentioned him in other posts, but I think its time for a rediscovery of Anthony Wilden who put cybernetics at the centre of his thinking and writing
There's a bridge between code and biology: the protoscience of biosemiotics.
Biosemiotics could potentially study the literal language behind DNA. There, you will have the union of coding and biological evolution, both of which you identified as cybernetics-relevant fields.
Other interesting field for cybernetics is obviosly the economy, but it should be economics mixed with geology, meteorology and virtually all of Earth sciences; and social sciences too, particularly sociology, psychology, political science, and military science. Basically that mix would become Asimov's psicohistory, or Cockshott's cybercommunism, or Turchin's cliodynamics.
The financial sector of the economy refers to the management of uncertainty, balancing the known unknowns and unknown unknowns of the economy and taking decisions over that. In that sense, finance should be close to the ICTs sector.
Politics and war refers to the management of uncertainty over social behavior, so both are good realms for applying cybernetics too.
This is something that bothered quite a bit when I was studying philosophy. It’s been great to discover Yuk Hui’s works, although I came to them rather late.
I should say also that there is a reasonably strong cyberneticist tradition in contemporary philosophy of mind; sometimes its influence is explicit, as in enactivism, and sometimes it is less so.
Oddly specific avoidance of Land but on the off chance he hasn't come up in your research, his 90-10s work on cybernetics is imo a clear offshoot of process philosophy (cladistically deleuzian of course)
heh I almost started including Land in the final portion but felt that would have to be its own post...ironically it was Land that first led me to cybernetics, I was reading about the CCRU in my early 20s bc it was edgy, but didn't understand it at all...I looked up what CCRU stood for, and vividly remember reading the wikipedia page about cybernetics as an entry point to all of this.
Here is an attempt: cybernetics, particularly Stafford Beer, was adopted by the Allende government in Chile. The US decided to destroy the Allende government, therefore all of its experiments had to be crushed, therefore cybernetics had to go on the truck to the stadium. Or get in the helicopter.
Given that the CIA has been documented as creating the American abstract expressionist art movement after ww2 for the very deliberate purpose of discouraging descriptive artists from describing poor people or other bad things, suppressing 'cybernetics' is not a far-fetched theory.
Maybe there are antimemes that have a purpose, for example, what if there are phenomena that grow when not talked about or described in any symbolic way, and viceversa. For example: the fertility rate in teenagers plummets directly with sex education, while in conservatives societies, talking about sex is taboo and they have high fertility rates.
The version I was taught when I learned math econ is that cybernetics was a catchy name for what became, less excitingly, control theory, similar to catastrophe theory and differential geometry. That doesn't include ideas like those of Stafford Beer.
More precisely, Ashby’s and Wiener’s cybernetics fit into the historical development of control theory, both in the classical 1940s frequency-domain framework of Bode and Nyquist (used by Wiener) and in the modern 1960s state-space differential equations framework developed independently by Kalman and Pontryagin (and anticipated by Ashby in the 1940s).
Belated additional data point in support of the argument:
I am a dissertating sociologist originally interested in modeling people as economically satiable. It seems plausible that at least some people can have enough goods and services and any more will not increase their utility. The assumption of non-satiation (https://en.wikipedia.org/wiki/Local_nonsatiation) is crucial for many different foundational economic theories so an alternative assumption could result in usefully different models of economic action.
In any case, I naively assumed that this would be an intuitive pitch. When *I* think about consuming, I think about eating, and people are very familiar with the idea of being full in that context. To my bafflement, I struggled to interest any of my peers and professors. No doubt some of the problem had to do with me being a fresh grad student unaccustomed to rhetorical norms but more and more I am appreciating how dominant modes of thought make it hard for social scientists to think about set points.
Maximization is deeply baked into our null hypotheses about the world. Popular constructs like human/social/cultural capital collapse swaths of human activity into a stock of stuff that people always want more of. Regression models, no matter how fancy, are always hunting for relationships of "more A" --> "more B". Control systems and set points, by contrast, are a nightmare or just impossible to identify in regression models. If a car is maintaining a constant speed good luck using a regression model to identify any correlation between "height of accelerator" and "velocity" despite the causal relationship. The fact that you have little to no variation on velocity makes it a dead variable for OLS. In fact, the more powerful the control system, the worse OLS methods will do at finding it in observational data. Andrew Abbott has a nice Sociological Theory paper, "Transcending General Linear Reality", on how linear regression has stunted sociologists' imagination and it touches on some cybernetics-adjacent ideas.
Anyway, thanks for the post! It gave me several nice ideas and things to read.
Awesome post. My sense is that the cybernetic perspective fails to ever take hold because the things we’re trying to study are just really fucking complicated. You end up sounding like a madman, or gesturing toward a general vibe that readers may or may not identify with. For me, the conclusion of every analysis ends up being “this is super complicated and these complications are not widely understood or appreciated”.
Not to say one can’t go farther than that, but it’s simply really hard and takes monumental effort to either articulate the mechanistic breakdown of interacting processes, or to write about them well enough to sound intelligent.
There’s just not a lot of room in intellectual meme space to move our understanding of a complicated topic from 3% to 4%… one is always incentivized to reduce the problem to such a degree that you can say you understand 90-100%, having lost sufficient perspective to actually say anything interesting.
You’ve at least managed to articulate this phenomenon super well, as the mimetic environment of intellectualism is simply inhospitable to certain important ways of thinking.
yes -- yes, that's pretty much exactly right. Which is why I'm increasingly sympathetic to ppl out here sounding like madmen
Way to stick the landing. Your teaser got my hopes up, and you delivered. You have convinced me to stop talking about pragmatism–such an awkward term– as a source for cybernetics and talk instead about process philosophy.
Since you centered on Bergson, I have to plug Emily Herring's fantastic "Herald of a Restless World: How Henri Bergson Brought Philosophy to the People."
Your gesture to Dewey and James (he of the stream of consciousness) leads me to recommend Charles Peirce to this genealogy. It always feels like he is on the verge of blowing up, like Bergson is lately. Thanks to how well Pierce's semiotics help make sense of generative AI, it may be happening.
I'm seeing some mentions of Pierce coming up in contemporary social science methodology -- felt like he was less of the lineage in this piece, or at least that he hated being associated w Bergson. I havne't read Herring's book but a good friend did and told me about it, so parts of it might be in here
Perice hated being associated with anyone, except maybe James, because James was always so nice to him.
I'm halfway through Leif Weatherby's Language Machines, and am thinking about writing a review centered on his positioning of Peirce. Two other recent books that bring Peirce to bear on generative AI are Paul Kockelman's "Last Words Large Language Models and the AI Apocalypse" and David Gunkel's "AI for Communication."
If you or anyone digging through your comments has other recent examples of a Peirce revival, especially Peirce helping make sense of the outputs of LLMs, I'd be grateful for the pointer.
Ok, now I really need to read Language Machines...
and, not about LLMs, but here is Peirce's "IBE" applied in by methodologists in a top poli sci journal...I understand that sociology has also seen a revival of interest in pragmatist phil of science https://www.journals.uchicago.edu/doi/abs/10.1086/734280
Pragmatism was an awkward term because you were misusing it... Jamesian Pragmatism is a philosophy of the mind of the continental kind. It's subject is the human spirit's inquiry into whether, when, and which things matter. It's not a methodology, and is a really strange choice as a 'source for cybernetics'. If you were to go continental, Freud would have made a lot more sense. Cybernetics has the best chance of making sense (a long shot in any case to be honest) when it focuses tightly on recursion and feedback loops. Basically, you make a system Alive by giving it the Compulsion to Repeat that Freud outlined in Beyond the Pleasure Principle.
I would say pragmatism is a method for avoiding interminable fights over the definition of terms:
The pragmatic method is primarily a method of settling metaphysical disputes that otherwise might be interminable. It is to try to interpret each notion by tracing its respective practical consequences.
This is one of the more pragmatist definitions of pragmatism I've heard! Phrased in terms of what it's for rather than what it technically means. And maybe in some cases, this includes what it needs to be distinguished from. Sometimes I like to think of pragmatism a bit like a genealogy of the future, as opposed to genealogy of the past.
Also, fights over definitions cant be 'interminable' (to my first point, what an *extremely* Latinised way to say 'endless'). Either we end up agreeing or the thing falls out of use ... plus, that dialectic teaches us about the world. If all the things we might use pragmatism for, dialogue-murder for the sake of using methodologies instead of our bodies to think may well be the least in the spirit of its original thinkers
To be clear, “interminable” is James’s word, chosen for that particular definition in “What Pragmatism Means” though he did use “endless” to make similar points elsewhere.
I remember. But James was writing for readers who knew Latin. I have a whole thing about this. Anyway, I say what James did: its a method not a methodology. I just think he was being weird with the 'settling disputes' bit because it can be hard even for him to find the right words. Disputes was an odd choice for talking metaphysics. But like I said, all this is karma for not picking Freud instead at the start
Since you answered my mostly Greek and Saxon English with nearly only Latin English, we may as well be talking in two languages. Still, I'll try answer that since I know both.
Yes pragmatism is a method, but its not a methodology. Its a razor for thinking about meaning. I cant think of any metaphysical disputes either James said they wanted to settle. If anything what I like about pragmatism is its willingness to say about some of those disputes: ya know what, maybe this doesn't need settling, actually.
A scattered response to this wonderful essay:
1. The most provocative - perhaps Cybernetics can't be grounded in Western Philosophy at all, but most naturally fits inside a combination of Buddhist/Daoist philosophical ideas.
2. There's plenty of Cybernetic writing, however, it's on LinkedIn rather than Substack and not exactly a genre you want to emulate.
3. To echo Russell about Plato: the few people who read Whitehead (and now Deleuze/Stengers/Latour) don't read Weiner (or von Neumann or Rene Thom) and vice versa. Unlike AI, where cognitive scientists, philosophers and technologists read each other a lot.
4. Continuing the previous point, Cybernetics lacks vivid thought experiments like Searle's Chinese Room, and instead has indecipherable diagrams.
Your point 4 is really what I'm trying to get at in the essay -- but the question is *why* Western philosophy finds those thought experiments so compelling. Perhaps I should've gone further down this road, but I think that certain games are the proper cybernetic analogue to thought experiments.
Point 1 deserves a much longer response, but yes this resonance (and even explicit intellectual cross-pollination) came up a lot in my reading. Yuk Hui makes the connection with Daoism and certain strands of Confucianism very clear, and I'm struck by how many of the early cybernetic thinkers had some exposure to "the East." Wiener visited China and Japan in the 1930s, and both Stafford Beer and Ross Ashby were stationed in India during WW2.
There's a now forgotten (makes your point!) literature on Systems/Cybernetic theory and Buddhism. Joanna Macy's "Mutual causality in Buddhism and general systems theory" is the best known example of that genre. Then again, there's a 'Buddhism and X' literature for whatever is the intellectual fad of the day.
Did you read Richard Seaford? He wrote about the birth of both Indian and Greek Philosophies. He argues that coinage gave way to philosophy in Greece and in India because it involves the changing of a substance (gold, silver, or both mixed) to turn it into blocks that have a fixed meaning on economical value.
I hadn't-- very interesting!
Ok, so… I am doing causal discovery on physical systems (mostly astronomical). The main objection I hear when presenting my work is that while DAGs capture static (and acyclic, by construction) causation relations, dynamics is made up of feedback loops. Coupled systems of ODEs are already good at modeling those: are you sure we need DAGs? Meanwhile my wife, who is a regional economist, just presented a conference paper on the causal determinants of innovation based, again, on causal discovery applied to firm survey and patent data. The main objection she received is that “DAGs maybe are not for every discipline” and “how will you sell this?”. To each their own kind of antimemetics I guess…
oh yeah it's fun seeing statistical methodology across different disciplines....they have such different intuitions!
But, fwiw, my position is that causal inference (whether DAGs or potential outcomes or whatever) is an anti-cybernetic perspective. For better and for worse, depending on what we're trying to do
As I am facing concrete challenges when applying these methods to physical systems where causality often is indeed cyclical, I am very interested in your take
yeah so the tension is -- rigorous causal inference is the only serious way to get at one element of the causal chain. It is in that sense necessary...but this means throwing out the feedback/recursion that actually makes the thing tick.
Most of these problems, at least the ones I've encountered, are just beyond our ability to understand with the degree of rigor we've come to demand
Well yeah we need exogenous variables to be able to identify causal relationships via IV/experiments, but nothing is really exogenous in a closed system where everything interacts. If this is the kind of tension you are getting at, is there a systematic exploration of it in the literature?
Yes that’s the tension—and no I don’t really think people know how to deal with it. Most of our problems are unanswerable with the degree of rigor we’ve come to demand
You may have mentioned him in other posts, but I think its time for a rediscovery of Anthony Wilden who put cybernetics at the centre of his thinking and writing
Honestly no I hadn't heard of him -- but he fits right in here, thanks for the pointer!
There's a bridge between code and biology: the protoscience of biosemiotics.
Biosemiotics could potentially study the literal language behind DNA. There, you will have the union of coding and biological evolution, both of which you identified as cybernetics-relevant fields.
Other interesting field for cybernetics is obviosly the economy, but it should be economics mixed with geology, meteorology and virtually all of Earth sciences; and social sciences too, particularly sociology, psychology, political science, and military science. Basically that mix would become Asimov's psicohistory, or Cockshott's cybercommunism, or Turchin's cliodynamics.
The financial sector of the economy refers to the management of uncertainty, balancing the known unknowns and unknown unknowns of the economy and taking decisions over that. In that sense, finance should be close to the ICTs sector.
Politics and war refers to the management of uncertainty over social behavior, so both are good realms for applying cybernetics too.
Not a cybernetician and barely a blogger any more, but I have some pages on the subject, this one on the split between cybernetics and AI might be interesting https://hyperphor.com/ammdi/Marvin-Minsky%E2%88%95on-Cybernetics
This is something that bothered quite a bit when I was studying philosophy. It’s been great to discover Yuk Hui’s works, although I came to them rather late.
I should say also that there is a reasonably strong cyberneticist tradition in contemporary philosophy of mind; sometimes its influence is explicit, as in enactivism, and sometimes it is less so.
Oddly specific avoidance of Land but on the off chance he hasn't come up in your research, his 90-10s work on cybernetics is imo a clear offshoot of process philosophy (cladistically deleuzian of course)
heh I almost started including Land in the final portion but felt that would have to be its own post...ironically it was Land that first led me to cybernetics, I was reading about the CCRU in my early 20s bc it was edgy, but didn't understand it at all...I looked up what CCRU stood for, and vividly remember reading the wikipedia page about cybernetics as an entry point to all of this.
Many such cases, I look forward to the post
I am famously always promising to blog and then not blogging, but today feeling very happy that via Dan Davies, today I found your blog.
“why nobody writes about cybernetics”
I’d like to think that I write about both cybernetics and process philosophy
absolutely! sorry should've included you from the start, fixed
Here is an attempt: cybernetics, particularly Stafford Beer, was adopted by the Allende government in Chile. The US decided to destroy the Allende government, therefore all of its experiments had to be crushed, therefore cybernetics had to go on the truck to the stadium. Or get in the helicopter.
Given that the CIA has been documented as creating the American abstract expressionist art movement after ww2 for the very deliberate purpose of discouraging descriptive artists from describing poor people or other bad things, suppressing 'cybernetics' is not a far-fetched theory.
https://www.independent.co.uk/news/world/modern-art-was-cia-weapon-1578808.html
Cheers!
Maybe there are antimemes that have a purpose, for example, what if there are phenomena that grow when not talked about or described in any symbolic way, and viceversa. For example: the fertility rate in teenagers plummets directly with sex education, while in conservatives societies, talking about sex is taboo and they have high fertility rates.
The version I was taught when I learned math econ is that cybernetics was a catchy name for what became, less excitingly, control theory, similar to catastrophe theory and differential geometry. That doesn't include ideas like those of Stafford Beer.
More precisely, Ashby’s and Wiener’s cybernetics fit into the historical development of control theory, both in the classical 1940s frequency-domain framework of Bode and Nyquist (used by Wiener) and in the modern 1960s state-space differential equations framework developed independently by Kalman and Pontryagin (and anticipated by Ashby in the 1940s).
Belated additional data point in support of the argument:
I am a dissertating sociologist originally interested in modeling people as economically satiable. It seems plausible that at least some people can have enough goods and services and any more will not increase their utility. The assumption of non-satiation (https://en.wikipedia.org/wiki/Local_nonsatiation) is crucial for many different foundational economic theories so an alternative assumption could result in usefully different models of economic action.
In any case, I naively assumed that this would be an intuitive pitch. When *I* think about consuming, I think about eating, and people are very familiar with the idea of being full in that context. To my bafflement, I struggled to interest any of my peers and professors. No doubt some of the problem had to do with me being a fresh grad student unaccustomed to rhetorical norms but more and more I am appreciating how dominant modes of thought make it hard for social scientists to think about set points.
Maximization is deeply baked into our null hypotheses about the world. Popular constructs like human/social/cultural capital collapse swaths of human activity into a stock of stuff that people always want more of. Regression models, no matter how fancy, are always hunting for relationships of "more A" --> "more B". Control systems and set points, by contrast, are a nightmare or just impossible to identify in regression models. If a car is maintaining a constant speed good luck using a regression model to identify any correlation between "height of accelerator" and "velocity" despite the causal relationship. The fact that you have little to no variation on velocity makes it a dead variable for OLS. In fact, the more powerful the control system, the worse OLS methods will do at finding it in observational data. Andrew Abbott has a nice Sociological Theory paper, "Transcending General Linear Reality", on how linear regression has stunted sociologists' imagination and it touches on some cybernetics-adjacent ideas.
Anyway, thanks for the post! It gave me several nice ideas and things to read.