I think everyone needs either STEM training, or practice in some real-world domain with feedback such as auto repair, farming, plumbing, or business management, not because it's easy, but because:
- it introduces people to the idea that some questions have correct and incorrect answers which can be found and distinguished from each other, and to how to know whether your question is one which might have an answer, or is even a question
- it teaches them to seek trade-offs and optimizations rather than perfection
- it is the only way people can ever recognize how sloppily they reason in the absence of empirical feedback
- it is the only way people can grasp how feedback within a system, and interactions with other systems, and noise within the system, will interfere with analytic reasoning. STEM isn't best characterized as analytical reasoning; it's best characterized as those disciplines in which physical feedback hits analytical reasoning.
I liked your comment and I agree with the importance of STEM to shape critical thinking. Lately the rise of the "long-termist bayesian rationalist" sphere makes me wonder if STEM thinking is being overstretched and overvalued and we are disregarding "humanity"and treating outselves as "meat machines". Do you have any thoughts on this?
I do, & explaining them is one thing I want to do with my substack, which I guess I need to start writing soon bcoz I have a bunch of subscribers.
It all began because I wanted to write better My Little Pony fan-fiction. I'd given up reading printed fiction because after about 1990, I didn't connect with it. By chance I discovered there was a community of literary authors who'd taken refuge from Marxist "relevance", Nazi "authenticity", and post-modern irony, within the MLP community, where they still wrote stories about the light and the dark of ordinary life.
I wrote MLP stories until I got to the place where the hardest part of writing stories was knowing what stories were worth telling. Why do people write stories, and why do they read them?
So I studied the history of Western literature and literary criticism. I found literary fashions came in waves, & the dogma of each wave was mostly that everything the last wave had said was good, was bad; & vice-versa. Most dogmas said either that the purpose of art and literature was to ask questions, or that it was to give answers.
Some cultures left stories behind in pictures or sculptures, so I began studying that history. The art of question-answering cultures looked similar. Usually, their pictures were drawn in primary colors, with no shadows, shading, or perspective. Brown is the easiest color to make, yet was the least-used. Figures were outlined with thick black lines. Physical space did not exist; artists crammed all the figures together. Important people were drawn big; unimportant people were drawn small, if at all, and were all identical. Artists did not draw ugly people, sick people, or children, and drew women only when necessary. Everything was abstract, made with a few lines and curves, drawn with little skill. Everything had a clear propagandistic purpose.
Cultures who asked questions were less common. They drew realistic forms in scenes from a definite point of view. Each person in a scene looked different, even the low-ranking.
Stranger still, many cultures had the two artistic traditions existing side-by-side. All religious and political drawings, such as the tomb statues of Egyptian pharaohs and the rock bas-relief carvings made in the public spaces in Neo-Assyrian palaces, are of the abstract "answers" school. Drawings of only personal interest, such as those in the tombs of Egyptian scribes, and in the private royal spaces in Neo-Assyrian palaces, are of the realistic "questions" school. Anyone looking at the abstract "answers" art would think the culture's artisans had little skill; but one glance at the naturalistic "answers" school would prove that wrong.
Once I realized that abstract art was intentionally "bad", I realized these abstractions expressed the culture's metaphysical beliefs while retelling its foundational myths, while the realistic art, free from religious & political propaganda obligations, showed the world as it really was. In Europe, the culture of ready-made answers, of boxes that every person had to fit into or be driven out, was formalized in a grand, brilliant, unified theory of art, religion, philosophy, science, and politics, by Plato.
My substack is called "Two Cultures" because I believe that Western history was driven by conflict between these two cultures: Rationalists, who want the world to be simple, orderly, and perfect; and Empiricists, who accept that it is complicated, messy, and can't be perfected.
Rationalists impose their ideology on nature, especially human nature. They demand either a strict command hierarchy, or complete unity of intent, so are usually totalitarian, like Sparta, Plato's "Republic", the Catholic Church, Calvinists, Jacobins, communists, fascists, and progressives. They believe the Universe consists of objects which each fall into one of a small number of distinct categories: earth/air/water/fire, human/animal, good/bad. Each category has a Form (Plato); each instance of it is somehow spawned by that Form. *No thing can ever turn into another kind of thing.* The only way to turn an oligarchy into a democracy is to destroy the oligarchy and create a democracy in its place. If a thing is already like its Form, any change will probably make it worse. This is why Rationalists are either radicals who want a revolution rather than gradual change (the Left), or conservatives who want to keep things as they are (the Right).
Empiricists search for truth in nature. They've learned the secrets of *machines*, things of many parts which together produce behavior qualitatively different from anything any one part can produce on its own. They are animated without *anima*. An empiricist political system, like that of Athens, early medieval Venice, Ganondagan, or the US, is a machine for integrating the desires of many different people into something none of them desires most, but which is a compromise between them all. Empiricists tinker with their political systems forever, believing they can always be improved, rather than throwing the dice on the off-chance that some theoretical Rationalist system is in fact perfect (even though it's always failed catastrophically before).
Now I can get back to your question: Are we disregarding "humanity"and treating ourselves as "meat machines"?
My questions to you are, 1) What is this "humanity", and 2) Why do you seem to think treating ourselves as meat machines would be a bad thing?
But we know the answers. "Humanity" is the Form (or "essence", or "spirit", or "nature") of the human, which in Rationalist philosophy is always assumed to be perfect. Being a meat machine would be bad because machines are an assembly of parts without a soul (Latin anima, Greek psyche) to *anima*te them.
Your question comes from 2400 years of Western philosophy presuming that value, spirit, life, agency, moral status, all require validation by some supernatural power. Anything untouched by God's magic finger is dead and valueless. If it nonetheless moves, and claims to live and have value, it's an offense to the gods.
But when the first materialists invented atomism, and with it the proposition that we are meat machines, they didn't see their theory as disenchanting the Universe. Just the opposite: Leucippus, Democritus, and Lucretius saw materialism as *enchanting* the Universe. Their gods were cruel, unjust, and capricious hedonists. Being given a purpose by a Greek or Roman god, as Sisyphus, Orestes, and Aeneas were, was at best a terrible ordeal, and often an eternal torment. The classical worldview was that humans are ruled by generally brutal and uncaring human rulers, ruled by brutal and uncaring gods, ruled by the uncaring fates. Being made of atoms in the void, interacting by logical rules, liberated you from the hierarchy of cruelty and injustice, and meant your life was governed by logic rather than caprice. In the classical world, it was the religious who were the nihilists, and the materialists who saw the world as awesome. See this chat with Gemini for a fuller explanation.
This view of the material world as more awe-inspiring than any divine creation was finally stomped out, deliberately, by the Roman Catholic Church in the High Middle Ages. At that time, Europe had 3 kinds of wonder: miracles (of God), magic (of Satan), and marvels (of man). There is, believe it or not, a medieval poem about a knight who had a robot assistant. It was neither magical nor miraculous, but a marvel, made by human ingenuity. See Le Goff, English translation Goldhammer 1988, The Medieval Imagination, part 1, "The Marvelous".
The Church rewrote the old tales about marvels to turn them into either miracles or magic, and erased the marvelous from Western thought. (Citation needed for that last sentence, but I can't find it now and honestly it's thin, just one sentence in one book with no sources.) The earliest instance I know of is St. Augustine interpreting marvels as miracles in the 4th century, described briefly, and without grasping the distinction between miracle and marvel, in Jones 1952, The Medieval Mind, p. 130-133.
The Christian God's "purpose" for nearly every human ever born is to live a few decades, then be tortured for eternity. The existentialist doctrine that we should find our own purposes is more reasonable and moral than looking to that god, or any other yet proposed by humans, for a purpose.
I want to re-introduce the marvelous. I'm not the first. Science fiction used to do so frequently. Jules Verne's submarines, Larry Niven's Ringworld. But they didn't have our advantage: that all the mysteries of epistemology needed to be solved to build working LLMs. We can dissolve all the arguments of the philosophers and linguists. We still cannot touch the noumena, but we can list its sufficient statistics to the 5th decimal place. People everywhere moan about life being disenchanted, when with a little effort they could, like Faust, *literally build a homunculus in their attic*.
The temptation to say "mere matter" comes from centuries of brutal religious punishment of any show of independence from the master plan of a brutal god. How much more plausible, and how much more amazing, for life to construct itself, over billions of years, and even build feelings and consciousness, from a few laws and particles, in a way we can understand! How grateful we should be that we can feel and desire at all, rather than horrified when our desires deviate from the dictums of ancient desert nomads!
One lone human is made from more atoms than there are stars in the Universe, and is many, many orders of magnitude more-complex than all the rest of the Universe outside of this planet put together. I believe a thorough calculation would show that most of the information comprising the entire Universe, can be found within your own body. We are marvelous meat machines, and should be proud of that.
It's harder to BS in STEM, but it can be done. At the moment, I can't think of many non-STEM fields where it's rare for people to BS their way to fame. Finance, small business management, law, medicine.
I have a much higher opinion of law than I used to. Law is basically collective philosophy with well-defined objectives and ground rules. It does what philosophy is supposed to do: decide what exists (e.g., the ontology of legal personhood), what our social values are, and how to coordinate in pursuing them. But grounded in reality, without metaphysics.
But I think BS rules today in all the humanities. The ones that resisted the longest were in pop culture: fiction, movies, music, other entertainment. Those all rely on making money, so they have to make something that some large audience will pay for. But they have their own parallel to BS: pandering, e.g., Marvel movies. The panderer is at least more-honest than the bloviating bullshitter.
Law is indeed an interesting subject. On the one hand it's entirely fabricated (a prime example of "intersubjective truth") but it also holds itself to a standard of internal consistency that the humanities do not. It's fair to say that persuasiveness holds value in the legal profession, but pure BS won't get you very far.
The emphasis on precedent raises an interesting question: From whence did our legal concepts arise in the first place? In the US it's mostly English Common Law, but the US founding fathers also made reference in their arguments to Scots Law, Maritime law, and even Roman Law. There was some ambiguity in how we chose from those traditions.
An interesting rabbit hole I learned about recently is Louisiana Civil Law, which in contrast to the other 49 states adopts certain features from Spanish and French Law regarding contracts, inheritance, and personal liability.
As I explained below, I do appreciate the STEM-to-Humanities pipeline idea, but first, consider this drawback of the idea: These broad discipline categories align with fundamentally different cognitive processing styles.
STEM disciplines fit a brain optimized for rule-governed, algorithmic ("fine-grained") information. A major advantage of STEM is domain modularity: you can rigorously analyze problems without needing reference to vast, disparate external fields. In contrast, the humanities attract and reward brains wired for long-distance associations, quasi-regularities, and diffuse pattern recognition across broad, interconnected domains of human experience.
This difference explains why STEM fields regularly support child prodigies and early-career breakthroughs, with math, physics, and computer science being the stand-out cases. We seldom see child prodigies in diplomacy, history, or literature. These disciplines require a breadth of contextual knowledge and human experience that can typically only be hard-won through decades of lived wisdom.
For these reasons, a general pipeline from STEM to the humanities won't work for most people. However, a one-way late-career transition from narrow/rule-governed fields to broad/integrative questions is a recognized pattern. Scholars can exploit early-life advantages in algorithmic domains, then pivot toward broad synthesis as their life experience and domain breadth accumulate.
Examples of this arc that I noticed as I was evolving as a young scientist.
Herb Simon: Moved from formal economics and decision models to broad questions of human cognition, artificial intelligence, and organizational behavior.
Christof Koch: Shifted from quantitative biophysics and cellular neurobiology to the expansive, integrative question of consciousness.
Steven Pinker: Began in technical psycholinguistics and visual cognition before expanding into broad historical, societal, and humanistic syntheses.
Alan Turing: Started in foundational logic and computation before pivoting to morphogenesis, artificial intelligence, and philosophy of mind.
Isaac Newton: Mastered formal physics and mathematics early, then devoted much of his later life to theology, history, and alchemy.
E.O. Wilson (Ant biology/sociobiology --> broad human nature, ethics, and biodiversity synthesis like Consilience).
Michael Polanyi (Physical chemistry --> philosophy of science, social theory, epistemological tacit knowledge).
(Personal note: E.O. Wilson was my sociobiology professor in 1981 and Steven Pinker wrote me letters of recommendation for graduate school.)
I'd be interested to hear what other examples of people transitioning from narrow/rule-governed domains to broad, integrative humanistic questions stand out to others here.
Having observed people for a while now, I find the fact that academia exists at all – that civilization somehow landed on an incentive structure that is even approximately truth seeking, in the objective sense – as nothing short of miraculous.
It's like the old parable of a dog walking on its hind legs: One shouldn't quibble about how well it does the trick, when the surprise is that it does it at all.
You're applying the MBA Fallacy - the assumption that skills and credibility transfer between work domains.
Speaking from work experience (not academia), the skills required and effective at design and execution positions are massively different from those in operational ones. The main reason we move people up from the bottom is that this ensures that once arrived at higher positions, they have enough residual knowledge of the lower ones to apply good judgment there. If you parachute people in from the top, they will be easily (and often willingly) bullshitted by operational claims.
Perhaps academia is different, but I really don't think so.
And yes, of course, the maintenance of stratified caste structures is always in play, everywhere.
Flippantly: make everyone study in math, logic, and analytic philosophy. If you can't do those well, and you progress up the ladder, you likely have high bullcrapology skills.
I would exclude logic, and would definitely exclude analytic philosophy, from that list. Teaching abstract logic without embedding it in a domain within which you can test your answers, such as computer programming or weather prediction, is worse than teaching nothing; it leads people to tremendously overestimate the reliability of their logical reasoning. Teaching people to reason by teaching them philosophy is like teaching them to dance by having them study cards with dance step footprints on them, made by people who also learned to dance by studying cards, but have never danced, only argued with other people about the cards.
When I was in grad school, I studied many of the same things studied by analytic philosophers today: modality, epistemological grounding, mereology, identity, the semantics of time, belief, intentionality, and more--but I studied them in the CS department. We read natural language texts, developed representations of all these things, and *implemented them in computer programs*. Then we ran the code to parse and represent English on more text, and watched how they failed, and failed, and failed. If you take a random text from the wild and test your theories on it in that way, most large paragraphs will require one PhD thesis and thousands more lines of code before you can parse it. We were a long way from being able to parse one full page of randomly chosen text by the time I gave up on symbolic AI around 1993. (Though we may return to it someday, to help unravel the tangled representations of deep neural networks.)
Artificial intelligence is just scientific analytical philosophy (or, as my first advisor, whose PhD is in analytical philosophy, liked to say, "philosophy done right"). But few philosophers had any interest in our findings. They didn't have the patience or the humility to find and fix their errors.
We can be more explicit about the dangers of logic. Logic arises twice, first as foundation logic, and later as applied logic.
Foundation logic is the logic of pure mathematics. There is a forced bivalence. If we have incomplete information we refuse to make the deduction. We never guess. The strength of this is that we can make long chains of deductions without losing certainty. It lets us boot up mathematics, probability, and statistics.
Real life has deadlines. A wrong decision is worse than a correct decision. But a missed deadline is worse than a wrong decision taken in time. We also have experience and live in a usefully inductive universe. Though we are forced to guess, we can invent Bayesian reasoning, to paper over the cracks created by incomplete information, and do much better than guessing blindly.
Or can we? Bayesian reasoning involves fuzzing the basic conjunctions, not the propositions. And the number of basic conjunctions grows exponentially with the number of propositions. Bayesian reasoning doesn't scale up to real life; we hit the problem of bounded rationality.
Applied logic to the rescue! We treat things that are very likely as certain, and avoid bogging down in the analysis of myriad events of very low probably. We live like military commanders, committing whole heartedly to the vigorous execution of our plans. Plans that are logical, albeit with a degree of approximation appropriate to the urgency of the situation (like playing chess with a clock; don't lose on time!). Plans that include reserves and a willingness to notice when things have gone wrong and deploy our reserves to rescue the situation.
The danger of logic is that we do *foundational logic* when we ought to be doing *applied logic*. We plunge in without reserves, and with no willingness to notice that things are not going to plan. Usually this works well, but when it works badly, it works terribly badly.
Bingo. I’ve always appreciated those in the humanities I’ve mentored/studied under, but that was a long time ago. Now it seems a noble field populated by lessor qualified “academics, not to mention students.
But how important is ability in math and logic (analytic philosophy might not be so easy to judge) for the historian, literary critic, etc.? The easier-to-judge STEM tasks need to have a close relationship with the harder-to-judge high-level humanities tasks, in order to be relevant.
Judging from the astonishing incompetence of many humanities professors at logical reasoning, I say it's very important. But I disagree with Robin's characterization of STEM task as easier than humanities tasks. No one "hits the ceiling" of STEM, reaching a competence level where they face no more challenges.
One thing missing from this is the human health span. We would like to promote from low positions to high positions based on whether the decisions that the candidate took in his low position have stood the test of time. How long can we wait? That depends on the human health span. We mustn't wait so long that our high position are exclusively filled by the old and tired.
This is relevant to how we talk about life-extension. Our discussions are usually too personal. We approve (or not) of life-extension based on an individual desire to live longer. But there are social reasons to approve of life-extension. Here we notice that if we lived longer our institutions could look at longer track records and have a more reliable basis for promotion.
I think there's something to this - encouraging the STEM to humanities transition probably is a good thing.
But I'm skeptical of the analogy to business management. In my limited (but first hand) experience, the best managers are not those who were skilled at specific non-managerial tasks, but those who, unprompted, found ways to improve the efficiency and effectiveness of those tasks. Most workers, even skilled ones, do as they've been trained and don't see or act on opportunities for improvement, unless asked by someone else who sees them. The few who see such opportunities and act on them without being prompted are managerial material.
However, I think academia is probably different and the analogy doesn't work. (But then I've never been an academic.)
I’m not sure. By the time the humanities person gets to be in influential positions, writing stuff that people take to be addressing harder and more important questions, aren’t they already being judged on the quality of their current output, which can be evaluated on their own?
Like your co-podcaster, Agnes Callard, did a lot of her early training on reading, interpreting, and commenting on historical figures. But now, she’s influential largely because people find what she writes on questions they care to be insightful and useful. Isn’t the fact that people read and reward what they find insightful already a sufficiently selective filter?
Or is your claim simply that academics are leaving money on the table? That would get even more insightful things if they had scholars follow a STEM to Humanities pipeline? If so, this is empirically testable; check if the those humanities scholars who came through STEM are more insightful on average (though you’d have to control for intelligence to see the training effects).
Traditionally the easy to hard sequence in the humanities depended on mastering various languages, hence requiring Latin and Ancient Greek for all university students. Given the dense structure of these languages and the logic needed to use, write, and speak them properly, they're probably as good a signal of objective feedback as things like calculus serve in STEM.
I would also add that even today, one of the ways a high school student can signal talent in the humanities, even if he or she isn't very strong in math, is to do well in Latin courses and get a high score on the national Latin exam.
I think everyone needs either STEM training, or practice in some real-world domain with feedback such as auto repair, farming, plumbing, or business management, not because it's easy, but because:
- it introduces people to the idea that some questions have correct and incorrect answers which can be found and distinguished from each other, and to how to know whether your question is one which might have an answer, or is even a question
- it teaches them to seek trade-offs and optimizations rather than perfection
- it is the only way people can ever recognize how sloppily they reason in the absence of empirical feedback
- it is the only way people can grasp how feedback within a system, and interactions with other systems, and noise within the system, will interfere with analytic reasoning. STEM isn't best characterized as analytical reasoning; it's best characterized as those disciplines in which physical feedback hits analytical reasoning.
You might be able to teach some of these things with a carefully thought-out strategy game.
I liked your comment and I agree with the importance of STEM to shape critical thinking. Lately the rise of the "long-termist bayesian rationalist" sphere makes me wonder if STEM thinking is being overstretched and overvalued and we are disregarding "humanity"and treating outselves as "meat machines". Do you have any thoughts on this?
I do, & explaining them is one thing I want to do with my substack, which I guess I need to start writing soon bcoz I have a bunch of subscribers.
It all began because I wanted to write better My Little Pony fan-fiction. I'd given up reading printed fiction because after about 1990, I didn't connect with it. By chance I discovered there was a community of literary authors who'd taken refuge from Marxist "relevance", Nazi "authenticity", and post-modern irony, within the MLP community, where they still wrote stories about the light and the dark of ordinary life.
I wrote MLP stories until I got to the place where the hardest part of writing stories was knowing what stories were worth telling. Why do people write stories, and why do they read them?
So I studied the history of Western literature and literary criticism. I found literary fashions came in waves, & the dogma of each wave was mostly that everything the last wave had said was good, was bad; & vice-versa. Most dogmas said either that the purpose of art and literature was to ask questions, or that it was to give answers.
Some cultures left stories behind in pictures or sculptures, so I began studying that history. The art of question-answering cultures looked similar. Usually, their pictures were drawn in primary colors, with no shadows, shading, or perspective. Brown is the easiest color to make, yet was the least-used. Figures were outlined with thick black lines. Physical space did not exist; artists crammed all the figures together. Important people were drawn big; unimportant people were drawn small, if at all, and were all identical. Artists did not draw ugly people, sick people, or children, and drew women only when necessary. Everything was abstract, made with a few lines and curves, drawn with little skill. Everything had a clear propagandistic purpose.
Cultures who asked questions were less common. They drew realistic forms in scenes from a definite point of view. Each person in a scene looked different, even the low-ranking.
Stranger still, many cultures had the two artistic traditions existing side-by-side. All religious and political drawings, such as the tomb statues of Egyptian pharaohs and the rock bas-relief carvings made in the public spaces in Neo-Assyrian palaces, are of the abstract "answers" school. Drawings of only personal interest, such as those in the tombs of Egyptian scribes, and in the private royal spaces in Neo-Assyrian palaces, are of the realistic "questions" school. Anyone looking at the abstract "answers" art would think the culture's artisans had little skill; but one glance at the naturalistic "answers" school would prove that wrong.
Once I realized that abstract art was intentionally "bad", I realized these abstractions expressed the culture's metaphysical beliefs while retelling its foundational myths, while the realistic art, free from religious & political propaganda obligations, showed the world as it really was. In Europe, the culture of ready-made answers, of boxes that every person had to fit into or be driven out, was formalized in a grand, brilliant, unified theory of art, religion, philosophy, science, and politics, by Plato.
My substack is called "Two Cultures" because I believe that Western history was driven by conflict between these two cultures: Rationalists, who want the world to be simple, orderly, and perfect; and Empiricists, who accept that it is complicated, messy, and can't be perfected.
Rationalists impose their ideology on nature, especially human nature. They demand either a strict command hierarchy, or complete unity of intent, so are usually totalitarian, like Sparta, Plato's "Republic", the Catholic Church, Calvinists, Jacobins, communists, fascists, and progressives. They believe the Universe consists of objects which each fall into one of a small number of distinct categories: earth/air/water/fire, human/animal, good/bad. Each category has a Form (Plato); each instance of it is somehow spawned by that Form. *No thing can ever turn into another kind of thing.* The only way to turn an oligarchy into a democracy is to destroy the oligarchy and create a democracy in its place. If a thing is already like its Form, any change will probably make it worse. This is why Rationalists are either radicals who want a revolution rather than gradual change (the Left), or conservatives who want to keep things as they are (the Right).
Empiricists search for truth in nature. They've learned the secrets of *machines*, things of many parts which together produce behavior qualitatively different from anything any one part can produce on its own. They are animated without *anima*. An empiricist political system, like that of Athens, early medieval Venice, Ganondagan, or the US, is a machine for integrating the desires of many different people into something none of them desires most, but which is a compromise between them all. Empiricists tinker with their political systems forever, believing they can always be improved, rather than throwing the dice on the off-chance that some theoretical Rationalist system is in fact perfect (even though it's always failed catastrophically before).
Now I can get back to your question: Are we disregarding "humanity"and treating ourselves as "meat machines"?
My questions to you are, 1) What is this "humanity", and 2) Why do you seem to think treating ourselves as meat machines would be a bad thing?
But we know the answers. "Humanity" is the Form (or "essence", or "spirit", or "nature") of the human, which in Rationalist philosophy is always assumed to be perfect. Being a meat machine would be bad because machines are an assembly of parts without a soul (Latin anima, Greek psyche) to *anima*te them.
Your question comes from 2400 years of Western philosophy presuming that value, spirit, life, agency, moral status, all require validation by some supernatural power. Anything untouched by God's magic finger is dead and valueless. If it nonetheless moves, and claims to live and have value, it's an offense to the gods.
But when the first materialists invented atomism, and with it the proposition that we are meat machines, they didn't see their theory as disenchanting the Universe. Just the opposite: Leucippus, Democritus, and Lucretius saw materialism as *enchanting* the Universe. Their gods were cruel, unjust, and capricious hedonists. Being given a purpose by a Greek or Roman god, as Sisyphus, Orestes, and Aeneas were, was at best a terrible ordeal, and often an eternal torment. The classical worldview was that humans are ruled by generally brutal and uncaring human rulers, ruled by brutal and uncaring gods, ruled by the uncaring fates. Being made of atoms in the void, interacting by logical rules, liberated you from the hierarchy of cruelty and injustice, and meant your life was governed by logic rather than caprice. In the classical world, it was the religious who were the nihilists, and the materialists who saw the world as awesome. See this chat with Gemini for a fuller explanation.
This view of the material world as more awe-inspiring than any divine creation was finally stomped out, deliberately, by the Roman Catholic Church in the High Middle Ages. At that time, Europe had 3 kinds of wonder: miracles (of God), magic (of Satan), and marvels (of man). There is, believe it or not, a medieval poem about a knight who had a robot assistant. It was neither magical nor miraculous, but a marvel, made by human ingenuity. See Le Goff, English translation Goldhammer 1988, The Medieval Imagination, part 1, "The Marvelous".
The Church rewrote the old tales about marvels to turn them into either miracles or magic, and erased the marvelous from Western thought. (Citation needed for that last sentence, but I can't find it now and honestly it's thin, just one sentence in one book with no sources.) The earliest instance I know of is St. Augustine interpreting marvels as miracles in the 4th century, described briefly, and without grasping the distinction between miracle and marvel, in Jones 1952, The Medieval Mind, p. 130-133.
The Christian God's "purpose" for nearly every human ever born is to live a few decades, then be tortured for eternity. The existentialist doctrine that we should find our own purposes is more reasonable and moral than looking to that god, or any other yet proposed by humans, for a purpose.
I want to re-introduce the marvelous. I'm not the first. Science fiction used to do so frequently. Jules Verne's submarines, Larry Niven's Ringworld. But they didn't have our advantage: that all the mysteries of epistemology needed to be solved to build working LLMs. We can dissolve all the arguments of the philosophers and linguists. We still cannot touch the noumena, but we can list its sufficient statistics to the 5th decimal place. People everywhere moan about life being disenchanted, when with a little effort they could, like Faust, *literally build a homunculus in their attic*.
The temptation to say "mere matter" comes from centuries of brutal religious punishment of any show of independence from the master plan of a brutal god. How much more plausible, and how much more amazing, for life to construct itself, over billions of years, and even build feelings and consciousness, from a few laws and particles, in a way we can understand! How grateful we should be that we can feel and desire at all, rather than horrified when our desires deviate from the dictums of ancient desert nomads!
One lone human is made from more atoms than there are stars in the Universe, and is many, many orders of magnitude more-complex than all the rest of the Universe outside of this planet put together. I believe a thorough calculation would show that most of the information comprising the entire Universe, can be found within your own body. We are marvelous meat machines, and should be proud of that.
In short, you want problem spaces where a person can't BS their way to a favorable outcome. STEM is good for that but plenty of other fields are too.
It's harder to BS in STEM, but it can be done. At the moment, I can't think of many non-STEM fields where it's rare for people to BS their way to fame. Finance, small business management, law, medicine.
I have a much higher opinion of law than I used to. Law is basically collective philosophy with well-defined objectives and ground rules. It does what philosophy is supposed to do: decide what exists (e.g., the ontology of legal personhood), what our social values are, and how to coordinate in pursuing them. But grounded in reality, without metaphysics.
But I think BS rules today in all the humanities. The ones that resisted the longest were in pop culture: fiction, movies, music, other entertainment. Those all rely on making money, so they have to make something that some large audience will pay for. But they have their own parallel to BS: pandering, e.g., Marvel movies. The panderer is at least more-honest than the bloviating bullshitter.
Law is indeed an interesting subject. On the one hand it's entirely fabricated (a prime example of "intersubjective truth") but it also holds itself to a standard of internal consistency that the humanities do not. It's fair to say that persuasiveness holds value in the legal profession, but pure BS won't get you very far.
The emphasis on precedent raises an interesting question: From whence did our legal concepts arise in the first place? In the US it's mostly English Common Law, but the US founding fathers also made reference in their arguments to Scots Law, Maritime law, and even Roman Law. There was some ambiguity in how we chose from those traditions.
An interesting rabbit hole I learned about recently is Louisiana Civil Law, which in contrast to the other 49 states adopts certain features from Spanish and French Law regarding contracts, inheritance, and personal liability.
As I explained below, I do appreciate the STEM-to-Humanities pipeline idea, but first, consider this drawback of the idea: These broad discipline categories align with fundamentally different cognitive processing styles.
STEM disciplines fit a brain optimized for rule-governed, algorithmic ("fine-grained") information. A major advantage of STEM is domain modularity: you can rigorously analyze problems without needing reference to vast, disparate external fields. In contrast, the humanities attract and reward brains wired for long-distance associations, quasi-regularities, and diffuse pattern recognition across broad, interconnected domains of human experience.
This difference explains why STEM fields regularly support child prodigies and early-career breakthroughs, with math, physics, and computer science being the stand-out cases. We seldom see child prodigies in diplomacy, history, or literature. These disciplines require a breadth of contextual knowledge and human experience that can typically only be hard-won through decades of lived wisdom.
For these reasons, a general pipeline from STEM to the humanities won't work for most people. However, a one-way late-career transition from narrow/rule-governed fields to broad/integrative questions is a recognized pattern. Scholars can exploit early-life advantages in algorithmic domains, then pivot toward broad synthesis as their life experience and domain breadth accumulate.
Examples of this arc that I noticed as I was evolving as a young scientist.
Herb Simon: Moved from formal economics and decision models to broad questions of human cognition, artificial intelligence, and organizational behavior.
Christof Koch: Shifted from quantitative biophysics and cellular neurobiology to the expansive, integrative question of consciousness.
Steven Pinker: Began in technical psycholinguistics and visual cognition before expanding into broad historical, societal, and humanistic syntheses.
Alan Turing: Started in foundational logic and computation before pivoting to morphogenesis, artificial intelligence, and philosophy of mind.
Isaac Newton: Mastered formal physics and mathematics early, then devoted much of his later life to theology, history, and alchemy.
E.O. Wilson (Ant biology/sociobiology --> broad human nature, ethics, and biodiversity synthesis like Consilience).
Michael Polanyi (Physical chemistry --> philosophy of science, social theory, epistemological tacit knowledge).
(Personal note: E.O. Wilson was my sociobiology professor in 1981 and Steven Pinker wrote me letters of recommendation for graduate school.)
I'd be interested to hear what other examples of people transitioning from narrow/rule-governed domains to broad, integrative humanistic questions stand out to others here.
Yes, there's already a recognizable pattern; I want it to be stronger. But of course probably most who start in STEM should stay in STEM.
Having observed people for a while now, I find the fact that academia exists at all – that civilization somehow landed on an incentive structure that is even approximately truth seeking, in the objective sense – as nothing short of miraculous.
It's like the old parable of a dog walking on its hind legs: One shouldn't quibble about how well it does the trick, when the surprise is that it does it at all.
You're applying the MBA Fallacy - the assumption that skills and credibility transfer between work domains.
Speaking from work experience (not academia), the skills required and effective at design and execution positions are massively different from those in operational ones. The main reason we move people up from the bottom is that this ensures that once arrived at higher positions, they have enough residual knowledge of the lower ones to apply good judgment there. If you parachute people in from the top, they will be easily (and often willingly) bullshitted by operational claims.
Perhaps academia is different, but I really don't think so.
And yes, of course, the maintenance of stratified caste structures is always in play, everywhere.
There are other plausible reasons for the usual career task order, besides "enough residual knowledge of the lower ones".
Flippantly: make everyone study in math, logic, and analytic philosophy. If you can't do those well, and you progress up the ladder, you likely have high bullcrapology skills.
I would exclude logic, and would definitely exclude analytic philosophy, from that list. Teaching abstract logic without embedding it in a domain within which you can test your answers, such as computer programming or weather prediction, is worse than teaching nothing; it leads people to tremendously overestimate the reliability of their logical reasoning. Teaching people to reason by teaching them philosophy is like teaching them to dance by having them study cards with dance step footprints on them, made by people who also learned to dance by studying cards, but have never danced, only argued with other people about the cards.
When I was in grad school, I studied many of the same things studied by analytic philosophers today: modality, epistemological grounding, mereology, identity, the semantics of time, belief, intentionality, and more--but I studied them in the CS department. We read natural language texts, developed representations of all these things, and *implemented them in computer programs*. Then we ran the code to parse and represent English on more text, and watched how they failed, and failed, and failed. If you take a random text from the wild and test your theories on it in that way, most large paragraphs will require one PhD thesis and thousands more lines of code before you can parse it. We were a long way from being able to parse one full page of randomly chosen text by the time I gave up on symbolic AI around 1993. (Though we may return to it someday, to help unravel the tangled representations of deep neural networks.)
Artificial intelligence is just scientific analytical philosophy (or, as my first advisor, whose PhD is in analytical philosophy, liked to say, "philosophy done right"). But few philosophers had any interest in our findings. They didn't have the patience or the humility to find and fix their errors.
We can be more explicit about the dangers of logic. Logic arises twice, first as foundation logic, and later as applied logic.
Foundation logic is the logic of pure mathematics. There is a forced bivalence. If we have incomplete information we refuse to make the deduction. We never guess. The strength of this is that we can make long chains of deductions without losing certainty. It lets us boot up mathematics, probability, and statistics.
Real life has deadlines. A wrong decision is worse than a correct decision. But a missed deadline is worse than a wrong decision taken in time. We also have experience and live in a usefully inductive universe. Though we are forced to guess, we can invent Bayesian reasoning, to paper over the cracks created by incomplete information, and do much better than guessing blindly.
Or can we? Bayesian reasoning involves fuzzing the basic conjunctions, not the propositions. And the number of basic conjunctions grows exponentially with the number of propositions. Bayesian reasoning doesn't scale up to real life; we hit the problem of bounded rationality.
Applied logic to the rescue! We treat things that are very likely as certain, and avoid bogging down in the analysis of myriad events of very low probably. We live like military commanders, committing whole heartedly to the vigorous execution of our plans. Plans that are logical, albeit with a degree of approximation appropriate to the urgency of the situation (like playing chess with a clock; don't lose on time!). Plans that include reserves and a willingness to notice when things have gone wrong and deploy our reserves to rescue the situation.
The danger of logic is that we do *foundational logic* when we ought to be doing *applied logic*. We plunge in without reserves, and with no willingness to notice that things are not going to plan. Usually this works well, but when it works badly, it works terribly badly.
Bingo. I’ve always appreciated those in the humanities I’ve mentored/studied under, but that was a long time ago. Now it seems a noble field populated by lessor qualified “academics, not to mention students.
But how important is ability in math and logic (analytic philosophy might not be so easy to judge) for the historian, literary critic, etc.? The easier-to-judge STEM tasks need to have a close relationship with the harder-to-judge high-level humanities tasks, in order to be relevant.
Judging from the astonishing incompetence of many humanities professors at logical reasoning, I say it's very important. But I disagree with Robin's characterization of STEM task as easier than humanities tasks. No one "hits the ceiling" of STEM, reaching a competence level where they face no more challenges.
I just meant that STEM tasks have many supporting structures to make them easier. Even so, each task can be arbitrarily hard in absolute terms.
Thanks for clarifying that. Sorry for misinterpreting you.
One thing missing from this is the human health span. We would like to promote from low positions to high positions based on whether the decisions that the candidate took in his low position have stood the test of time. How long can we wait? That depends on the human health span. We mustn't wait so long that our high position are exclusively filled by the old and tired.
This is relevant to how we talk about life-extension. Our discussions are usually too personal. We approve (or not) of life-extension based on an individual desire to live longer. But there are social reasons to approve of life-extension. Here we notice that if we lived longer our institutions could look at longer track records and have a more reliable basis for promotion.
As an undergraduate, we were supposed to take on above-my-paygrade topics like: What should Napoleon have done instead of invade Russia in 1812?
Then I went to MBA school and the topics were more like: What kind of personality types would buy gourmet dog food for their dog?
I rather like the latter kind of questions when I was a 21-year-old MBA student. (Of course, they don't let 21-year-olds go to MBA school anymore ...)
I think there's something to this - encouraging the STEM to humanities transition probably is a good thing.
But I'm skeptical of the analogy to business management. In my limited (but first hand) experience, the best managers are not those who were skilled at specific non-managerial tasks, but those who, unprompted, found ways to improve the efficiency and effectiveness of those tasks. Most workers, even skilled ones, do as they've been trained and don't see or act on opportunities for improvement, unless asked by someone else who sees them. The few who see such opportunities and act on them without being prompted are managerial material.
However, I think academia is probably different and the analogy doesn't work. (But then I've never been an academic.)
I’m not sure. By the time the humanities person gets to be in influential positions, writing stuff that people take to be addressing harder and more important questions, aren’t they already being judged on the quality of their current output, which can be evaluated on their own?
Like your co-podcaster, Agnes Callard, did a lot of her early training on reading, interpreting, and commenting on historical figures. But now, she’s influential largely because people find what she writes on questions they care to be insightful and useful. Isn’t the fact that people read and reward what they find insightful already a sufficiently selective filter?
Or is your claim simply that academics are leaving money on the table? That would get even more insightful things if they had scholars follow a STEM to Humanities pipeline? If so, this is empirically testable; check if the those humanities scholars who came through STEM are more insightful on average (though you’d have to control for intelligence to see the training effects).
It can be HARD to evaluate on its own the quality of current output.
Traditionally the easy to hard sequence in the humanities depended on mastering various languages, hence requiring Latin and Ancient Greek for all university students. Given the dense structure of these languages and the logic needed to use, write, and speak them properly, they're probably as good a signal of objective feedback as things like calculus serve in STEM.
I would also add that even today, one of the ways a high school student can signal talent in the humanities, even if he or she isn't very strong in math, is to do well in Latin courses and get a high score on the national Latin exam.