One important insight I associate with Foucault is that power is much more multifarious than we often like to think. Basically no one is truly powerless and no one is all-powerful. There are always complex webs and conflicts of power. The phrase of “speaking truth to power” often, as you note, pretends that the speakers are relatively low power - though speaking truth does constitute one kind of power, and there are often others.
But I think this undermines even the case that in historical societies, power was mostly illegitimate and extractive. That may have been true of formal society-level power, but even in traditional societies, the power of parents over children, and the different forms of power of children over parents, and various other forms of power in all other social relationships, likely weren’t as illegitimate and extractive.
One of the issues is, I think, that a lot of the truths people want to tell to power are not really epistemic truths but more moral truths. Crucially this difference gets elided and then leads to the appearance of motivated reasoning which then discredits the analysis. It tells us that such moral truths have become embedded as priors within various progressive domains of discourse. Not necessarily wrong but often unexamined. Then we're talking at cross purposes.
I've often bristled at the way in which academics dismiss claims that the academy in some sense "brainwashes" students, moving them toward left-wing positions, and this is basically the reason why. As you write, "such moral truths have become embedded as priors within various progressive domains of discourse." In the process of being acculturated into a given academic discipline, when many moral "truths" have been set up as pre-discursive, there is definitely a degree to which people are being taught to see such things as true even if professors don't see that as their explicit mission or intend to alter the political views of their students. This is less nefarious than the kinds of hysterical claims of brainwashing common on the right, but arguably it is more insidious because it is more difficult to detect, for professors and for students.
I think more so than any conscious and unconscious conditioning, the primary driver of the politicisation derives from the way choosing to attend and exist in the academy affects one's political outlook.
When one looks at the average political views of people by sector, universities are simply not special in that regard. It is assumed that a nefarious, deliberate and calculated process is happening in universities, yet occupations like construction or truck driver exhibit just as much deviation from the norm without people believing there is some grand conspiracy making that happen.
Well I don't have anything against promoting progressive values in general but if we embed them as assumptions within the domain of knowledge as if they're facts that cannot be challenged we often fail to argue for those values explicitly and teach students they can't question them. The less transparent those assumptions are to us the more likely it is that we'll end up unable to engage with those who don't share those prior commitments and dismiss them as manifestly disingenuous for not recognising the truth.
Great stuff. This explains a frustration I have with much of the current left, which is that instead of analyzing new technologies and sources of power with a fresh (though still leftist) perspective, they want to shoehorn everything into familiar abuses of power we've seen in the past. This is most obvious with AI. Capabilities deniers want to say it's just like crypto or the housing bubble. People who claim datacenters use a ton of water want datacenters to have exactly the same problems as oil and gas drilling.
To your second point, there's also a huge motivated reasoning component to this. A core belief of leftists is that workers have the real power over the economy, if only they could organize. The possibility that AI will replace workers is so threatening to this core belief that many would rather say AI is useless and being foisted on workers than develop new claims for a new era.
The way many of my fellow progressives/leftists discuss AI drives me insane. I'm more of the "AI as normal technology" camp than a singularity doomer (at least insofar as the immediate future), but there's still tons of useful analysis that can be gleaned from that angle.
I honestly think most of it is they played around with ChatGPT when it was new, decided it was overhyped, and have refused to update that opinion for three years
Dan, great stuff. I've been running a series at Sacred Cow BBQ that converges on a lot of this from a different door, and, so here's two pushbacks I'd offer as friendly extensions/amendments rather than disagreements.
The motivated-reasoning section is exactly where my next piece is going. The empirical literature you're gesturing at (Taber & Lodge, Kahan on identity-protective cognition, Mercier & Sperber) is the cognitive layer of what's also a structural story, and an angle I've hit hard in some of my own academic research. Smart/sophisticated people are more motivated and more capable of building identity defenses and strategies.
The posture of speaking truth to power is self-flattering, yes, but the deeper reason it survives is that the disciplines that adopted it lack what Gad Saad (of all people) recently called an "autocorrective mechanism of reality." The harder sciences, e.g., Engineering, and even the "harder" social sciences like business get reality-corrected all the time: build a bad bridge, predict a market wrong, you hear about it. Humanities and parts of social science don't. Approval-optimization without reality-correction will always select for skillful rationalization over accuracy. Also, that's the disciplinary version of the RLHF problem in AI alignment.
The other push: I'd be (only slightly) gentler on the original constructivist turn, even granting the horrible places where it ended up. The Frankfurt school and standpoint folks weren't wrong that institutions distort knowledge production. They were wrong about the target (positivism wasn't the problem; approval-optimization was), and they built a meta-theoretical apparatus that ended up immunizing their own descendants against exactly the critique you're making here. That's the irony I'd add to your self-exemption section: the same critique-of-power tradition that should have made academia self-aware about its own ideological power produced the tools that now prevent it from being applied inward (which is the argument Part 2 of my series ran in detail: https://kylesaunders.substack.com/p/the-fifty-year-war-or-reality-bats).
"Isolated demands for rigor as bravery." Amen. Strong piece.
Love your insights as always. But your choice of Curtis Yarvin as a conservative intellectual seems very stretched, and your only example is a single comment he made. I follow lots of serious conservative intellectuals and I have never even run into Yarvin.
Further, contrary to the self-justifying comment from Stephen Pinker (who I respect) that the far left doesn't dominate the universities is a recent poll showing that not only have conservatives been chased out of academia since the 1980s, so have self-identified moderates.
And finally, it isn't just the obvious woke classes that are the problem, it is the tenor of all classes to take on a left wing bias. For example, at many "elite" universities, you can get a graduate degree in English literature without ever reading Shakespeare! (you know, patriarchy, racism, etc.)
I saw someone remark that the postmodern politics of speaking truth to power has curdled (on the right and left) into speaking power to truth, and that seemed an extremely pithy summary of the basic problem.
I'm probably overly literal, but doesn't "speaking truth to power" include agreeing with the powerful when they are right? If I say, "Hey, Mr Landlord, I used to think rent control was good for tenants, but I did more research, and it turns out that you were right. Rent control hurts tenants in the long run," am I not speaking truth (as I now understand it) to power?
The expression only becomes bad epistemology if you assume in advance that the powerful *always* distort the truth to maintain their power. But the ideological critiques of Marxists and Post-Modernists go even further. They suggest that the very idea of truth is a myth imposed on us by the powerful to maintain control. You can't oppose truth to power because truth flows out of power. The role of the intellectual is not to "speak truth" but to confront power by whatever means necessary.
Thus, I'd prefer not to attribute our epistemological disorders to this expression b/c at least it holds intellectuals accountable to the truth.
PS - I was curious about the origins of the expression and asked Claude about it. Apparently, it comes from the Quakers. They used it as the title of a 1955 anti-war pamphlet (https://quaker.org/legacy/sttp.html). I haven't read all of it, but regardless of whether you agree with their thesis, the introduction suggests that it is a serious intellectual argument. They try to prove rather than assume that the preconceptions of America's Cold Warriors are wrong.
When I find myself agreeing too much with what Dan writes, I need to write a sufficient push back since the comments are also mostly agreeing with you.
I have no major disagreements with your argument, particularly your claim that:
“An ethos of speaking truth to power has three problems: it prejudges what inquiry is supposed to discover; it licenses motivated reasoning under a heroic self-image; and it exempts intellectual elites from the suspicion they direct at others.”
That said, allow me to play devil’s advocate on a few points.
First, regarding the concern that an ethos of “speaking truth to power” prejudges what inquiry is supposed to discover: is this necessarily a flaw? Could it instead sometimes function as a heuristic shaped by experience? You worry that people operating within this ethos often see patterns of oppression or injustice where none exist. But many of these people likely have the opposite fear: failing to recognize genuine patterns of injustice that are actually there. Examples that you give, ranging from Wells to Solzhenitsyn, are indeed people who seem to be driven more by the fear of false negatives than false positives.
In that sense, perhaps the disagreement is not over whether bias exists, but over different kinds of errors each individual is more willing to tolerate.
Second, on motivated reasoning: if motivated reasoning appears to be ubiquitous across intellectual and political life, perhaps the issue is not the existence of motivation itself, but whether it can be harnessed for better purposes. While I agree that the heroic self-image slop can pollute our epistemic reservoirs, there are also genuine cases of moral courage. People are often driven by impure motives, yet still manage to produce socially valuable outcomes. So perhaps the relevant question is not whether motivations are (or even can be) pure, but whether the resulting practices remain open to criticism, correction, and accountability.
Third, I strongly agree with your point about self-exemption. However, I'm not sure you would agree that perhaps Western intelligentsia should spend more time interrogating how their own professions, institutions, and social environments contribute to the very conditions they critique when attempting to “speak truth to power". It can be as unproductive as the norm it tries to replace.
My main question, then, is this: even if “speaking truth to power” constitutes bad epistemology in some sense, what would the "true dissidents fighting clear cases of oppression and exploitation" do? Are those people also practicing bad epistemology then? Would it be fair to act only after their claims have been validated?
It seems that “speaking truth to power” is less an epistemology than a stance adopted under conditions of asymmetric risk. Now, whether or not those asymmetric risks exist for everyone, I'm not sure. As you pointed out, Western intelligentsia might not be under those conditions.
As always, thank you for the thought-provoking post. Let me know what you think.
Awesome post, Dan! I broadly agree that the kind of general attitude you describe tends to be epistemically noxious, especially so when it excludes self-suspicion. (In my own research, I'm developing the idea that suspicious hypotheses are just one hypothesis among many that ought to be considered in each case, rather than as the default one you should already believe.) But it's tough to apply the general claim about the badness of pathological suspicion to particular case studies—since, as you note, the topics at issue are complex and ought to be considered on a case-by-case basis.
While I don't agree with it, what you describe as "the left's" take on AI strikes me as reasonable, and a lot of good evidence has been marshaled in favor of it. I don't see why it's excluded from being one of the views within the space of "reasonable disagreement," within the "profound expert disagreements," or merited by the "ambiguous, contested evidence across many fields." While I completely agree that some left intellectuals on this topic are exactly as you describe, the evidence you point to for the existence of motivated reasoning on this topic is far from decisive, since the individuals that make up the group presumably hold the same belief for different reasons and with varying degrees of precision, thoughtfulness, and sophistication. Dismissing the left's take on etiological grounds risks running afoul of some of the same pathologies you intend to diagnose; alas, such is walking the tightrope between trust and suspicion.
Why do you think the populist left’s take on AI are reasonable? It seems to me that many on the left go farther than "AI as normal technology" and end at "AI as complete scam", which seems very foolish to me. The comparisons to crypto/NFTs are particularly bad.
My personal lower bound estimate is that AI will be at least as impactful as smartphones, social media, or even the internet itself.
Hey Sean, I guess a lot rides on what we think "the left's" take on AI is. What I mean by a view on a difficult/contested topic being reasonable is just that there's some good evidence in favor of it—not that the view is true or even that it's the view most strongly supported by the overall balance of evidence. This makes reasonability fairly minimal.
I didn't take Dan's version of the left's view to be "AI is a complete scam." I'd agree that that's unreasonable, though I'm not sure what that take is even saying. That said, that doesn't strike me as representative of the left's views on AI. A more common—and more reasonable— nearby view is that it's a bubble.
The take that AI is like crypto/NFTs was indeed very common among the left when LLMs first hit the scene, and imo that was reasonable (but false) back then. It seems most on the left no longer believe this—and updating one's view in response to new evidence is a hallmark of reasonability.
Even the stochastic parrot stuff Dan linked to is reasonable, imo—and perhaps even true. The article Dan links doesn't even dispute that LLMs are stochastic parrots; it simply disputes the significance of that claim. If people make an inference such as "stochastic parrot > therefore it's useless and its impact will be negligible," then I'd agree that's a terrible inference. But again, I just don't see that sort of inference very often on the left. Instead, it makes most sense to ask in what contexts people invoke the idea of the stochastic parrot. Here's one: Why does AI consistently make up papers whenever I ask it to write me an intro ethics syllabus? Because it's a stochastic parrot: it doesn't have deep knowledge of ethics; it's just predicting what should be said next. And people on the left—for instance, many left-leaning professors—marshal this point to argue not that AI is useless, but that AI hasn't reached expert-level ability in their field.
That same article links to a speech Cory Doctorow gave, where he refers to AI as a "spicy autocomplete machine." But it's a speech, and in context it sounds more like he's just doing what speechwriters do: making a joke. Throughout the rest of his speech, and in the rest of his written work (of which I've only read some), his views seem fairly reasonable, and he's clear that he thinks the impacts of AI will be significant. In other words, joking that it's a "spicy autocomplete machine" does not mean he isn't taking AI seriously.
One of the things that make LLMs interesting is that they can exhibit such capabilities having been trained only on language. Now, of course, they're trained on other data streams as well, but fundamentally what they do is "next-token prediction", and their abilities flow from that. That's interesting! There are obvious and consequential failure modes like the ones you highlight (LLMs are relentless bullshitters). But that doesn't make them useless per se.
I recently came across this paper that distinguishes between "emergent capabilities" and "emergent intelligence", and thought it an interesting and useful way to decompose what we see in language models' performance.
I wish that there was a better term, or a more elaborate explanation of "speaking truth to power". Because it seems to me depending on the framing it can be good or bad. The way you have framed it the ethos causes one to prejudge who/what has power and it relies on a simple heuristic that power and truth systematically collide.
But, I don't see it that way - I think an ethos of attempting to push back on hegemonic ideas is good *in the long run*. OFC there are intellectually lazy and misguided folks who assume they know where power is before actually investigating the system, but the problem isn't the ethos described - it's the method they use.
Let's use AI as an example. I am 100% with you - AI is huge, and the people/organizations who wield AI will have immense power over society. So if someone says "it's all a bubble/scam to get more money" they aren't speaking truth to power. They are speaking misunderstandings that allow the powerful to accumulate more power. So if you wanted to say "The ethos of assuming you know the truth and who has power without actual empirical investigation of those facts is bad" I am 100% with you. But what about *actually* speaking *truth* to real power.
I will attempt to explain why I think it is a good ethos to have with a few key examples
#1) Marx - It is clear upon reading Marx's work in the 21st century that he made quite a few empirical mistakes. But let's place him in a historical context. He (and Engels) were observing the birth of the industrial revolution, especially as it related to Dickensian Britain. The immiseration of the working class in the 19th century with declining life expectancy and quality of life (despite rapidly increasing GDP/productivity) is the historical reality. Marx was reacting to what he saw as a gross injustice, formulating an entirely new frame of looking at power structures. Without his reaction to the ascendant power of a capitalist oligarchy we may never have seen the labor movement in the early 20th century take off.
#2) Scientific progress - We often see ideas that are entrenched become more and more sclerotic as they do not explain the phenomena we see in reality. I am not saying that 100% of people should doubt the scientific paradigms of the day. However a world where a committed 10-20% of people are working to disprove the core ideas of the day is more likely to get us out of a local minima and push us into a more accurate paradigm. On a population level we want some people to be innovators/disruptors, even if that group is only a fraction of the whole.
#3) Wars of position - I hope you've read Gramsci because I think his idea of a war of position is actually incredibly useful to analyze large scale societal changes. If you live in a society (like antebellum southern US) then the dominant structure is white supremacy, enslavement and a rich slave owning class dominating the economic/political world. If you have only so much time/effort in your life to produce work it seems to me that it is entirely reasonable/beneficial to fighting ruthlessly against that regime. If I am going to try and convince my fellow citizens that slavery is wrong I am going to make my argument as persuasive as possible. If someone comes in and says, "Well as much as you hate slavery, what about the incredible amounts of cotton we are producing? Think about how much that cotton is used in textiles to help clothe the cold worldwide. If we start producing less cotton that may cause harm to some others." If I spend my time analyzing the world-wide utilitarian calculation of abolishing slavery it will not only reduce the amoutn of time/effort I have left for advocating for abolition, it will weaken the argument. And if you have CORRECTLY identified Hegemonic regime that is in power making the world a worse place, it is often strategically sound to fight that hegemonic power on all fronts, rather than concede small empirically true points on their side. Why would I be willing (as an intellectual) to see myself as a soldier in a war of position againts an evil regime, instead of as a noble truth teller. Because, according to you "almost everything you believe [in] is based on information you acquired from others—from the claims, gossip, reports, books, remarks, opinion pieces, teaching, images, video clips, and so on that other people communicated to you." So I want to make sure the information I pass onto others can direct them away from evil.
-------
Finally, you live in Britain rather than USA so I wanted to just take issue with one statement of yours: "Western intellectual elites live in liberal democracies characterised by universal rights, formal equality, the rule of law, pluralism, constitutional limits on state power"
I live in the USA. We do not have universal rights (women in southern states no longer have bodily autonomy, non-whites can be racially profiled and targeted by law enforcement, observers in my city have been shot, killed and the government refuses to investigate the ICE officers who killed them). We do not have formal equality (Black citizens in the south are losing all representation in federal government). The rule of law is decaying fast. Our Supreme court has literally interpreted the constitution to say that any act of the president can not be illegal so long as it is deemed "official", an arbitrary determination. So while I do not believe we yet live in a totalitarian regime, the sanguine assumption that powerful regimes have been permanently defeated by a liberal regime does not reflect what I see in my country.
Once you have accurately determined what is truth and where the power lies, encouraging intellectuals and citizens to push back will *over time* lead to a more copacetic equilibrium.
Thanks Dan. One thing I kept thinking as I listened to your piece was about simplicity. Human intentionality and social interactions are inherently immensely complex. This means that nothing about social change can be simple - the diagnosis of a problem, its causes, solutions and methods to achieve those solutions. The one simple rule of thumb I would adopt is that if anyone is telling you that any of those things are simple then they are almost certainly wrong.
I think it has always been the case that many of the people willing to speak truth to power have been contrarian jackasses with a tendency to conspiratorial thinking and lacking an accurate self-image. I think being reasonable mitigates against telling the truth to people in power. The reasonable man adapts himself to the world and all that.
Also Rogan's point about punk rock is telling because Johnny Ramone was famously a Reaganite Republican. Although probably an outlier among punk rock in the 80s, but the rebel aesthetic has never been exclusively left wing.
Anyway as much as things have certainly changed, I'm not sure the fundamentals are as transformed as you imagine (I think you are failing to appreciate how much of this stuff existed in previous times).
Agreed: ever was it thus. A quarter of a century ago, the Big Bad Thing was genetic modification (despite scientific consensus being that it's not inherently dangerous). Anyone noting that "Flavr Savr" tomatoes might reduce food waste, or that Golden Rice could prevent blindness in vitamin A-deficient children were routinely castigated for being Monsanto shills.
This article sits oddly with the rest of your work, which highlights the savvy of individuals within their information environment.
Other comments have noted how the act of speaking truth to power is largely a social, moral activity and not an epistemic endeavor, but it is also important to note that power differentials greatly shape people's ability to reach truths of the more epistemic variety.
Specific knowledge and data are essential to navigate complex social issues, but those things are also embedded in power structures and have an opportunity cost which can render them inaccessible or invisible to most. For this reason, most then bring a partial view when they try to speak to power. Individuals who do that often improve the information environment by freely sharing what someone in their position is thinking and feeling, which enables others to reach more informed conclusions, so it can have informational benefits too.
This was a thought-provoking article but it needed more specific examples. You asserted a lot of things about “intellectuals” and the “modern left” and “many elites” that felt true, but we need actual examples to provide fact-based proof beyond just rhetoric that “feels” true. Kind of undermines your own thesis here.
One important insight I associate with Foucault is that power is much more multifarious than we often like to think. Basically no one is truly powerless and no one is all-powerful. There are always complex webs and conflicts of power. The phrase of “speaking truth to power” often, as you note, pretends that the speakers are relatively low power - though speaking truth does constitute one kind of power, and there are often others.
But I think this undermines even the case that in historical societies, power was mostly illegitimate and extractive. That may have been true of formal society-level power, but even in traditional societies, the power of parents over children, and the different forms of power of children over parents, and various other forms of power in all other social relationships, likely weren’t as illegitimate and extractive.
One of the issues is, I think, that a lot of the truths people want to tell to power are not really epistemic truths but more moral truths. Crucially this difference gets elided and then leads to the appearance of motivated reasoning which then discredits the analysis. It tells us that such moral truths have become embedded as priors within various progressive domains of discourse. Not necessarily wrong but often unexamined. Then we're talking at cross purposes.
I've often bristled at the way in which academics dismiss claims that the academy in some sense "brainwashes" students, moving them toward left-wing positions, and this is basically the reason why. As you write, "such moral truths have become embedded as priors within various progressive domains of discourse." In the process of being acculturated into a given academic discipline, when many moral "truths" have been set up as pre-discursive, there is definitely a degree to which people are being taught to see such things as true even if professors don't see that as their explicit mission or intend to alter the political views of their students. This is less nefarious than the kinds of hysterical claims of brainwashing common on the right, but arguably it is more insidious because it is more difficult to detect, for professors and for students.
I think more so than any conscious and unconscious conditioning, the primary driver of the politicisation derives from the way choosing to attend and exist in the academy affects one's political outlook.
When one looks at the average political views of people by sector, universities are simply not special in that regard. It is assumed that a nefarious, deliberate and calculated process is happening in universities, yet occupations like construction or truck driver exhibit just as much deviation from the norm without people believing there is some grand conspiracy making that happen.
A far more obvious conclusion is that the qualities of our daily existence significantly affects our political beliefs. https://www.voronoiapp.com/demographics/Political-Leaning-by-Occupation-3374
Well I don't have anything against promoting progressive values in general but if we embed them as assumptions within the domain of knowledge as if they're facts that cannot be challenged we often fail to argue for those values explicitly and teach students they can't question them. The less transparent those assumptions are to us the more likely it is that we'll end up unable to engage with those who don't share those prior commitments and dismiss them as manifestly disingenuous for not recognising the truth.
Great stuff. This explains a frustration I have with much of the current left, which is that instead of analyzing new technologies and sources of power with a fresh (though still leftist) perspective, they want to shoehorn everything into familiar abuses of power we've seen in the past. This is most obvious with AI. Capabilities deniers want to say it's just like crypto or the housing bubble. People who claim datacenters use a ton of water want datacenters to have exactly the same problems as oil and gas drilling.
To your second point, there's also a huge motivated reasoning component to this. A core belief of leftists is that workers have the real power over the economy, if only they could organize. The possibility that AI will replace workers is so threatening to this core belief that many would rather say AI is useless and being foisted on workers than develop new claims for a new era.
The way many of my fellow progressives/leftists discuss AI drives me insane. I'm more of the "AI as normal technology" camp than a singularity doomer (at least insofar as the immediate future), but there's still tons of useful analysis that can be gleaned from that angle.
I honestly think most of it is they played around with ChatGPT when it was new, decided it was overhyped, and have refused to update that opinion for three years
Dan, great stuff. I've been running a series at Sacred Cow BBQ that converges on a lot of this from a different door, and, so here's two pushbacks I'd offer as friendly extensions/amendments rather than disagreements.
The motivated-reasoning section is exactly where my next piece is going. The empirical literature you're gesturing at (Taber & Lodge, Kahan on identity-protective cognition, Mercier & Sperber) is the cognitive layer of what's also a structural story, and an angle I've hit hard in some of my own academic research. Smart/sophisticated people are more motivated and more capable of building identity defenses and strategies.
The posture of speaking truth to power is self-flattering, yes, but the deeper reason it survives is that the disciplines that adopted it lack what Gad Saad (of all people) recently called an "autocorrective mechanism of reality." The harder sciences, e.g., Engineering, and even the "harder" social sciences like business get reality-corrected all the time: build a bad bridge, predict a market wrong, you hear about it. Humanities and parts of social science don't. Approval-optimization without reality-correction will always select for skillful rationalization over accuracy. Also, that's the disciplinary version of the RLHF problem in AI alignment.
The other push: I'd be (only slightly) gentler on the original constructivist turn, even granting the horrible places where it ended up. The Frankfurt school and standpoint folks weren't wrong that institutions distort knowledge production. They were wrong about the target (positivism wasn't the problem; approval-optimization was), and they built a meta-theoretical apparatus that ended up immunizing their own descendants against exactly the critique you're making here. That's the irony I'd add to your self-exemption section: the same critique-of-power tradition that should have made academia self-aware about its own ideological power produced the tools that now prevent it from being applied inward (which is the argument Part 2 of my series ran in detail: https://kylesaunders.substack.com/p/the-fifty-year-war-or-reality-bats).
"Isolated demands for rigor as bravery." Amen. Strong piece.
The traditional speaker of truth to power was the Court Fool
Love your insights as always. But your choice of Curtis Yarvin as a conservative intellectual seems very stretched, and your only example is a single comment he made. I follow lots of serious conservative intellectuals and I have never even run into Yarvin.
Further, contrary to the self-justifying comment from Stephen Pinker (who I respect) that the far left doesn't dominate the universities is a recent poll showing that not only have conservatives been chased out of academia since the 1980s, so have self-identified moderates.
And finally, it isn't just the obvious woke classes that are the problem, it is the tenor of all classes to take on a left wing bias. For example, at many "elite" universities, you can get a graduate degree in English literature without ever reading Shakespeare! (you know, patriarchy, racism, etc.)
I saw someone remark that the postmodern politics of speaking truth to power has curdled (on the right and left) into speaking power to truth, and that seemed an extremely pithy summary of the basic problem.
I'm probably overly literal, but doesn't "speaking truth to power" include agreeing with the powerful when they are right? If I say, "Hey, Mr Landlord, I used to think rent control was good for tenants, but I did more research, and it turns out that you were right. Rent control hurts tenants in the long run," am I not speaking truth (as I now understand it) to power?
The expression only becomes bad epistemology if you assume in advance that the powerful *always* distort the truth to maintain their power. But the ideological critiques of Marxists and Post-Modernists go even further. They suggest that the very idea of truth is a myth imposed on us by the powerful to maintain control. You can't oppose truth to power because truth flows out of power. The role of the intellectual is not to "speak truth" but to confront power by whatever means necessary.
Thus, I'd prefer not to attribute our epistemological disorders to this expression b/c at least it holds intellectuals accountable to the truth.
PS - I was curious about the origins of the expression and asked Claude about it. Apparently, it comes from the Quakers. They used it as the title of a 1955 anti-war pamphlet (https://quaker.org/legacy/sttp.html). I haven't read all of it, but regardless of whether you agree with their thesis, the introduction suggests that it is a serious intellectual argument. They try to prove rather than assume that the preconceptions of America's Cold Warriors are wrong.
When I find myself agreeing too much with what Dan writes, I need to write a sufficient push back since the comments are also mostly agreeing with you.
I have no major disagreements with your argument, particularly your claim that:
“An ethos of speaking truth to power has three problems: it prejudges what inquiry is supposed to discover; it licenses motivated reasoning under a heroic self-image; and it exempts intellectual elites from the suspicion they direct at others.”
That said, allow me to play devil’s advocate on a few points.
First, regarding the concern that an ethos of “speaking truth to power” prejudges what inquiry is supposed to discover: is this necessarily a flaw? Could it instead sometimes function as a heuristic shaped by experience? You worry that people operating within this ethos often see patterns of oppression or injustice where none exist. But many of these people likely have the opposite fear: failing to recognize genuine patterns of injustice that are actually there. Examples that you give, ranging from Wells to Solzhenitsyn, are indeed people who seem to be driven more by the fear of false negatives than false positives.
In that sense, perhaps the disagreement is not over whether bias exists, but over different kinds of errors each individual is more willing to tolerate.
Second, on motivated reasoning: if motivated reasoning appears to be ubiquitous across intellectual and political life, perhaps the issue is not the existence of motivation itself, but whether it can be harnessed for better purposes. While I agree that the heroic self-image slop can pollute our epistemic reservoirs, there are also genuine cases of moral courage. People are often driven by impure motives, yet still manage to produce socially valuable outcomes. So perhaps the relevant question is not whether motivations are (or even can be) pure, but whether the resulting practices remain open to criticism, correction, and accountability.
Third, I strongly agree with your point about self-exemption. However, I'm not sure you would agree that perhaps Western intelligentsia should spend more time interrogating how their own professions, institutions, and social environments contribute to the very conditions they critique when attempting to “speak truth to power". It can be as unproductive as the norm it tries to replace.
My main question, then, is this: even if “speaking truth to power” constitutes bad epistemology in some sense, what would the "true dissidents fighting clear cases of oppression and exploitation" do? Are those people also practicing bad epistemology then? Would it be fair to act only after their claims have been validated?
It seems that “speaking truth to power” is less an epistemology than a stance adopted under conditions of asymmetric risk. Now, whether or not those asymmetric risks exist for everyone, I'm not sure. As you pointed out, Western intelligentsia might not be under those conditions.
As always, thank you for the thought-provoking post. Let me know what you think.
Awesome post, Dan! I broadly agree that the kind of general attitude you describe tends to be epistemically noxious, especially so when it excludes self-suspicion. (In my own research, I'm developing the idea that suspicious hypotheses are just one hypothesis among many that ought to be considered in each case, rather than as the default one you should already believe.) But it's tough to apply the general claim about the badness of pathological suspicion to particular case studies—since, as you note, the topics at issue are complex and ought to be considered on a case-by-case basis.
While I don't agree with it, what you describe as "the left's" take on AI strikes me as reasonable, and a lot of good evidence has been marshaled in favor of it. I don't see why it's excluded from being one of the views within the space of "reasonable disagreement," within the "profound expert disagreements," or merited by the "ambiguous, contested evidence across many fields." While I completely agree that some left intellectuals on this topic are exactly as you describe, the evidence you point to for the existence of motivated reasoning on this topic is far from decisive, since the individuals that make up the group presumably hold the same belief for different reasons and with varying degrees of precision, thoughtfulness, and sophistication. Dismissing the left's take on etiological grounds risks running afoul of some of the same pathologies you intend to diagnose; alas, such is walking the tightrope between trust and suspicion.
Why do you think the populist left’s take on AI are reasonable? It seems to me that many on the left go farther than "AI as normal technology" and end at "AI as complete scam", which seems very foolish to me. The comparisons to crypto/NFTs are particularly bad.
My personal lower bound estimate is that AI will be at least as impactful as smartphones, social media, or even the internet itself.
Hey Sean, I guess a lot rides on what we think "the left's" take on AI is. What I mean by a view on a difficult/contested topic being reasonable is just that there's some good evidence in favor of it—not that the view is true or even that it's the view most strongly supported by the overall balance of evidence. This makes reasonability fairly minimal.
I didn't take Dan's version of the left's view to be "AI is a complete scam." I'd agree that that's unreasonable, though I'm not sure what that take is even saying. That said, that doesn't strike me as representative of the left's views on AI. A more common—and more reasonable— nearby view is that it's a bubble.
The take that AI is like crypto/NFTs was indeed very common among the left when LLMs first hit the scene, and imo that was reasonable (but false) back then. It seems most on the left no longer believe this—and updating one's view in response to new evidence is a hallmark of reasonability.
Even the stochastic parrot stuff Dan linked to is reasonable, imo—and perhaps even true. The article Dan links doesn't even dispute that LLMs are stochastic parrots; it simply disputes the significance of that claim. If people make an inference such as "stochastic parrot > therefore it's useless and its impact will be negligible," then I'd agree that's a terrible inference. But again, I just don't see that sort of inference very often on the left. Instead, it makes most sense to ask in what contexts people invoke the idea of the stochastic parrot. Here's one: Why does AI consistently make up papers whenever I ask it to write me an intro ethics syllabus? Because it's a stochastic parrot: it doesn't have deep knowledge of ethics; it's just predicting what should be said next. And people on the left—for instance, many left-leaning professors—marshal this point to argue not that AI is useless, but that AI hasn't reached expert-level ability in their field.
That same article links to a speech Cory Doctorow gave, where he refers to AI as a "spicy autocomplete machine." But it's a speech, and in context it sounds more like he's just doing what speechwriters do: making a joke. Throughout the rest of his speech, and in the rest of his written work (of which I've only read some), his views seem fairly reasonable, and he's clear that he thinks the impacts of AI will be significant. In other words, joking that it's a "spicy autocomplete machine" does not mean he isn't taking AI seriously.
One of the things that make LLMs interesting is that they can exhibit such capabilities having been trained only on language. Now, of course, they're trained on other data streams as well, but fundamentally what they do is "next-token prediction", and their abilities flow from that. That's interesting! There are obvious and consequential failure modes like the ones you highlight (LLMs are relentless bullshitters). But that doesn't make them useless per se.
I recently came across this paper that distinguishes between "emergent capabilities" and "emergent intelligence", and thought it an interesting and useful way to decompose what we see in language models' performance.
https://royalsocietypublishing.org/rsta/article/384/2320/20250014/481676/Large-language-models-and-emergence-a-complex
Jesus what a post! Fuc***g amazing!
I wish that there was a better term, or a more elaborate explanation of "speaking truth to power". Because it seems to me depending on the framing it can be good or bad. The way you have framed it the ethos causes one to prejudge who/what has power and it relies on a simple heuristic that power and truth systematically collide.
But, I don't see it that way - I think an ethos of attempting to push back on hegemonic ideas is good *in the long run*. OFC there are intellectually lazy and misguided folks who assume they know where power is before actually investigating the system, but the problem isn't the ethos described - it's the method they use.
Let's use AI as an example. I am 100% with you - AI is huge, and the people/organizations who wield AI will have immense power over society. So if someone says "it's all a bubble/scam to get more money" they aren't speaking truth to power. They are speaking misunderstandings that allow the powerful to accumulate more power. So if you wanted to say "The ethos of assuming you know the truth and who has power without actual empirical investigation of those facts is bad" I am 100% with you. But what about *actually* speaking *truth* to real power.
I will attempt to explain why I think it is a good ethos to have with a few key examples
#1) Marx - It is clear upon reading Marx's work in the 21st century that he made quite a few empirical mistakes. But let's place him in a historical context. He (and Engels) were observing the birth of the industrial revolution, especially as it related to Dickensian Britain. The immiseration of the working class in the 19th century with declining life expectancy and quality of life (despite rapidly increasing GDP/productivity) is the historical reality. Marx was reacting to what he saw as a gross injustice, formulating an entirely new frame of looking at power structures. Without his reaction to the ascendant power of a capitalist oligarchy we may never have seen the labor movement in the early 20th century take off.
#2) Scientific progress - We often see ideas that are entrenched become more and more sclerotic as they do not explain the phenomena we see in reality. I am not saying that 100% of people should doubt the scientific paradigms of the day. However a world where a committed 10-20% of people are working to disprove the core ideas of the day is more likely to get us out of a local minima and push us into a more accurate paradigm. On a population level we want some people to be innovators/disruptors, even if that group is only a fraction of the whole.
#3) Wars of position - I hope you've read Gramsci because I think his idea of a war of position is actually incredibly useful to analyze large scale societal changes. If you live in a society (like antebellum southern US) then the dominant structure is white supremacy, enslavement and a rich slave owning class dominating the economic/political world. If you have only so much time/effort in your life to produce work it seems to me that it is entirely reasonable/beneficial to fighting ruthlessly against that regime. If I am going to try and convince my fellow citizens that slavery is wrong I am going to make my argument as persuasive as possible. If someone comes in and says, "Well as much as you hate slavery, what about the incredible amounts of cotton we are producing? Think about how much that cotton is used in textiles to help clothe the cold worldwide. If we start producing less cotton that may cause harm to some others." If I spend my time analyzing the world-wide utilitarian calculation of abolishing slavery it will not only reduce the amoutn of time/effort I have left for advocating for abolition, it will weaken the argument. And if you have CORRECTLY identified Hegemonic regime that is in power making the world a worse place, it is often strategically sound to fight that hegemonic power on all fronts, rather than concede small empirically true points on their side. Why would I be willing (as an intellectual) to see myself as a soldier in a war of position againts an evil regime, instead of as a noble truth teller. Because, according to you "almost everything you believe [in] is based on information you acquired from others—from the claims, gossip, reports, books, remarks, opinion pieces, teaching, images, video clips, and so on that other people communicated to you." So I want to make sure the information I pass onto others can direct them away from evil.
-------
Finally, you live in Britain rather than USA so I wanted to just take issue with one statement of yours: "Western intellectual elites live in liberal democracies characterised by universal rights, formal equality, the rule of law, pluralism, constitutional limits on state power"
I live in the USA. We do not have universal rights (women in southern states no longer have bodily autonomy, non-whites can be racially profiled and targeted by law enforcement, observers in my city have been shot, killed and the government refuses to investigate the ICE officers who killed them). We do not have formal equality (Black citizens in the south are losing all representation in federal government). The rule of law is decaying fast. Our Supreme court has literally interpreted the constitution to say that any act of the president can not be illegal so long as it is deemed "official", an arbitrary determination. So while I do not believe we yet live in a totalitarian regime, the sanguine assumption that powerful regimes have been permanently defeated by a liberal regime does not reflect what I see in my country.
Once you have accurately determined what is truth and where the power lies, encouraging intellectuals and citizens to push back will *over time* lead to a more copacetic equilibrium.
Thanks Dan. One thing I kept thinking as I listened to your piece was about simplicity. Human intentionality and social interactions are inherently immensely complex. This means that nothing about social change can be simple - the diagnosis of a problem, its causes, solutions and methods to achieve those solutions. The one simple rule of thumb I would adopt is that if anyone is telling you that any of those things are simple then they are almost certainly wrong.
J'accuse...!😜
I think it has always been the case that many of the people willing to speak truth to power have been contrarian jackasses with a tendency to conspiratorial thinking and lacking an accurate self-image. I think being reasonable mitigates against telling the truth to people in power. The reasonable man adapts himself to the world and all that.
Also Rogan's point about punk rock is telling because Johnny Ramone was famously a Reaganite Republican. Although probably an outlier among punk rock in the 80s, but the rebel aesthetic has never been exclusively left wing.
Anyway as much as things have certainly changed, I'm not sure the fundamentals are as transformed as you imagine (I think you are failing to appreciate how much of this stuff existed in previous times).
Agreed: ever was it thus. A quarter of a century ago, the Big Bad Thing was genetic modification (despite scientific consensus being that it's not inherently dangerous). Anyone noting that "Flavr Savr" tomatoes might reduce food waste, or that Golden Rice could prevent blindness in vitamin A-deficient children were routinely castigated for being Monsanto shills.
This article sits oddly with the rest of your work, which highlights the savvy of individuals within their information environment.
Other comments have noted how the act of speaking truth to power is largely a social, moral activity and not an epistemic endeavor, but it is also important to note that power differentials greatly shape people's ability to reach truths of the more epistemic variety.
Specific knowledge and data are essential to navigate complex social issues, but those things are also embedded in power structures and have an opportunity cost which can render them inaccessible or invisible to most. For this reason, most then bring a partial view when they try to speak to power. Individuals who do that often improve the information environment by freely sharing what someone in their position is thinking and feeling, which enables others to reach more informed conclusions, so it can have informational benefits too.
This was a thought-provoking article but it needed more specific examples. You asserted a lot of things about “intellectuals” and the “modern left” and “many elites” that felt true, but we need actual examples to provide fact-based proof beyond just rhetoric that “feels” true. Kind of undermines your own thesis here.