What this covers
Robin Hanson presents a sweeping account of human motivation, arguing that signaling—the drive to impress others and display loyalty to groups—explains the shape of many institutions we take for granted. The conversation ranges across schooling, medicine, charity, art, and conversation itself, with Hanson contending that stated motives (learning, healing, helping) function as cover for deeper social purposes. He is careful to distinguish between signaling as a fact of behavior and signaling as a problem: the issue is not that we signal, but that we signal to audiences who lack the knowledge to judge what actually works, producing waste and institutional dysfunction.
The discussion moves across several distinct territories. Hanson examines how loyalty signaling—showing you belong to a group—accounts for roughly half of all signaling yet remains the hardest kind for people to admit. He applies the framework to medical spending, arguing that much healthcare consumption reflects signaling about empathy rather than genuine health benefit, and proposes a practical heuristic: consume medicine only if you would pay for it out of pocket. On institutional innovation, he contends that we fail to adopt social improvements we could easily design, precisely because our resistance to change protects the hidden signaling functions those institutions serve. He also explores prediction markets as tools for revealing what people actually expect to happen rather than what they claim, and ventures into longer-term futures through the Age of Em scenario, where brain emulations might reshape civilization. Throughout, the conversation emphasizes understanding Levels of Explanation—separating proximate causes from deeper social ones—as essential to seeing institutions as they truly function.
Hanson argues that most human behavior—well over 90% in rich societies—is driven by hidden signaling motives that diverge from our stated motives, and that social scientists and policymakers must understand this 'elephant in the brain' to design better institutions, with the core problem being not signaling itself but its negative externalities and uninformed audiences.
- We say we go to school to learn or to the doctor to get well, but the real motive is often selfish signaling and showing off
- The waste in signaling comes from negative externalities and coordination failures, not from the pleasure of showing off
- Audiences are too ignorant, so we impress them with things that don't actually work (e.g. ineffective medicine)
Stated motives function as cover stories masking the actual reasons people behave as they do.
- Although we treat conversation as if it is about imparting information and finding useful things, more plausibly it is about showing off your backpack of tools and skills in context.
“although we like to talk about conversation as if it was about imparting information and finding out useful things, more plausibly it’s about showing off your backpack of tools and skills in context.”
- People can be classified as onion people, who have layer after layer with no clear core, versus egg people, who have a shell that once broken reveals what is inside; introverts pursue the egg strategy, showing 'this is who I am, not much is hidden,' enabling a stronger trust bond precisely because they are not concealing more.
“Onion people have layer after layer after layer. You peel it back, and there’s still more layers. You don’t really know what’s underneath. Whereas egg people, there’s a shell, and you get through it, and you see what’s on the inside.”
- Stated motives function as excuses precisely because they are sometimes true—like 'the dog ate my homework' working because dogs sometimes eat homework; we sometimes do get healthy from the doctor, learn in school, and help via charity, so the claim is not that these motives are fake but that they are more claimed than they actually operate.
“I use the example of “the dog ate my homework.” That works as an excuse because dogs sometimes eat homework.”
- Many people favor prediction markets as a political signal—taking the stance that they are in favor of truth and accuracy and against the liars and hypocrites—even if they don't actually want to look at the answers or trade in them, so support for prediction markets can itself function as a signaling stance about the world.
“people want to take the stance that they’re in favor of truth and accuracy, and they’re against all those people out there who are lying and hypocrites. So they might want to take the stance of favoring prediction markets even if they don’t actually want to look at the answers”
- The cynic's conundrum is that a cynic assigns low motives to others' behavior, yet the most plausible explanation for why someone behaves cynically—especially in a whiny, complaining voice—is that they themselves have low motives and are somewhat failures; this is why the authors deliberately avoided a cynical tone and chose to be positive about humans.
“The cynic’s conundrum is that the cynic is assigning low motives to other people’s behavior, and the conundrum is that, when you ask why is the cynic behaving as a cynic... the most plausible explanation is that they have low motives for their behavior.”
- Just as the downward-sloping demand curve rules nearly 100 percent of activity yet explains only a small fraction of what's going on, signaling is by one metric very large and significant but is only one layer of explanation; explanations run along a chain from proximate to distal causes, and the book focuses on relatively distal social explanations of why institutions like schools and hospitals are the way they are.
“There are local proximate causes, and there are distal far causes, and there’s a whole chain in between. You have to identify where along that chain you’re trying to talk about.”
- Hidden meanings are pervasive across fiction, nonfiction, and conversation because people constantly operate on more than one level; like an actor who cannot perform a scene of two lovers simply professing love until they find the character's hidden motive, audiences and authors always expect multiple layers of meaning.
“hidden meanings are all over the place in fiction, nonfiction, even talks like this. People constantly have more than one level.”
- When people think about signaling they default to signaling ability—being smart, conscientious, energetic, healthy—and neglect signaling loyalty; in fact roughly half of all signaling is loyalty signaling, showing a group that you are one of them, but we are especially reluctant to admit this kind because it is even less flattering than admitting we signal ability.
“people tend to neglect signaling loyalty. I’d say roughly half of signaling is signaling loyalty.”
- The over-90-percent signaling figure depends on a broad definition; if signaling is defined narrowly as conscious signaling—doing something you don't want to do while pretending you do, or acting with a true motivation different from the stated one—the share would probably drop to 30 percent or less.
“It would probably drop to 30 percent or less.”
- People are especially willing to express moral superiority when they are asserting the superiority not of themselves individually but of the group they belong to; expressing moral superiority is a way of showing loyalty to one's group ('we are right, they are wrong'), whereas expressing it to those immediately next to you is rare and unwise.
“people are especially willing to express moral superiority when they’re expressing the superiority, not of themselves individually, but of the group of people they’re within together.”
Evolution wired humans to pursue costly displays that audiences mistake for genuine value or ability.
- Evolution created behavioral responses keyed to basic functional cues—is there a potential mate, is this expensive, is this difficult—and although the modern environment provides novel stimuli, we map them onto these evolved instincts, so signaling behavior persists even in environments evolution never anticipated.
“I actually think evolution did a pretty decent job of noticing basic functional things in the environment. Is there a potential mate? Is this expensive? Is this cheap? Is this difficult to do?”
- People like art for relatively mundane reasons but we like to think we like it for grand reasons; art is roughly impressive things that are hard to do and that don't seem to have much other function, and our evolved cues for detecting that are pretty good.
“People like art for relatively mundane reasons, but we like to think we like it for grand reasons.”
- In a rich society like ours, well over 90 percent of human behavior can ultimately be traced back to some kind of signaling.
“In a rich society like ours, well over 90 percent.”
- Because there is a strong human tendency to collect signs of 'I'm smarter than you,' a group defined by rationality tends in practice to just collect markers like knowing Bayes's theorem or overconfidence, then show them off in any context—mirroring how popular interest in prediction markets has centered on identifying who among us is wiser rather than on the institution itself.
“there’s such a strong human tendency to want to collect these signs of “I’m smarter than you” that a group that defines itself by rationality ends up, in practice, just collecting “I know Bayes’s theorem,” “I know about overconfidence,””
- Watching sports does little for Hanson because his core values center on insight and being part of a process producing more insight, and there is just not much insight being generated in sports.
“Iput a priority on discovering things intellectually that I could explain to other people, that I could transfer to other people.”
- Lines like 'come up and see my etchings' work despite being transparent because transparent isn't the same as clear; the recipient isn't sold yet and wants an out that lets them pretend they never intended to go anywhere near what was suggested—just as a paper bag around alcohol gives police a fake excuse not to arrest—so the indirect line provides plausible deniability until the very last moment of commitment.
“Transparent isn’t the same as clear. In our book we give the example of the person drinking alcohol in public in a paper bag.”
- A person should get a relatively minimal health plan that might cut medical spending by a third to a half compared to typical levels, facing a price on the margin, and apply the heuristic: would I pay for this out of pocket if I had to? If not, skip it unless there is an especially strong reason to think it's one of the genuine exceptions.
“You should get a relatively minimal plan of the sort that might cut back medical spending by a third or a half compared to what most people do.”
- When people give to charity they act as if trying to help, but what they are really doing is showing they feel empathy and vulnerability to nearby pain; because their audience hardly knows what actually helps, givers don't pay much attention to effectiveness—so more effective altruists in the world would create an informed audience that pressures givers toward genuinely helpful actions.
“What they’re really doing is trying to show that they feel empathy, that they feel a vulnerability to pain around them”
- Someone who commits suicide is following the human package of feelings and is not necessarily sending any particular signal.
“They are following the human package of feelings. They aren’t necessarily sending any particular signal.”
Ignorant audiences unable to judge real impact enable wasteful signaling and underinvestment in effective action.
- The main problem with signaling is not the breadth of the audience but how poorly informed it is: because our audiences are too ignorant, we can impress them with things that are stupid; if audiences knew which actions actually worked, we would try harder to impress them with effective things that made a real difference.
“It’s less about the scope of the audience and more about how well informed they are. I’d say our main problem is our audiences are too ignorant. We’re impressing people who can be impressed by things that are stupid.”
- In medicine we show that we care about others by pushing them to get and buy medicine, but because neither we nor the recipients nor the audience know which medicines actually work, we don't much care whether they work when we push people to consume more medicine.
“in medicine, we’re trying to show off that we care about other people by pushing them to get medicine and buying medicine for them. We don’t actually know which medicines work very well, and we don’t much care because the people we’re helping don’t know and the audience doesn’t know.”
- Private organizations resist prediction markets because, although people claim to value information and analysis, much corporate activity is really coalition politics, and prediction markets destroy the favorite excuse for failed projects—that everything was fine until something unforeseeable struck—by revealing that people knew early the project would fail, thereby exposing managers to accountability.
“a lot of corporate activity is about creating coalitions and getting support for your coalition and undermining other rival coalitions. That’s what people are actually doing, but they want to give lip service to the information thing.”
- The RAND Health Insurance Experiment, still some of our best data, showed that when people paid full price for medicine they consumed less than when medicine was free, and that the extra free medicine apparently produced no net health benefits—suggesting a useful metric is to get only the medicine you would pay for out of your own pocket.
“TheRAND Health Insurance Experiment, which is still some of our best data, showed that when people paid full price for medicine, they got less medicine than when they had free medicine. That extra free medicine, apparently, had no net health benefits.”
- Combinatorial prediction markets combine many individual questions so that ten questions with ten answers each yield ten billion possible combined questions; using Bayesian-network-based methods, these can be made computable and accurate, allowing people to effectively manage all of those billions of numbers so they stay reasonable.
“if you have 10 questions, each of whom has 10 answers, you can make 10 to the 10 possible questions, which is 10 billion.”
- One of the biggest positive externalities our world neglects is innovation; all sorts of innovation have hugely insufficient incentives, and promoting it would be very valuable.
“innovation is one of the things in our world that we just have a huge lack of incentives for. All sorts of innovation, I think that would be great to promote.”
- When we get tools that help us do something, like remembering phone numbers, we get worse at that task because we offload it onto the tool, but we get better at other things by reallocating our mental resources—so better social tools would not make us globally stupider, just shift our investment to tasks lacking tools.
“when we have tools to help us do something, like remember phone numbers, we get worse at that thing because we offload the task onto the tool. Then, we get better at other things.”
Prediction markets and better institutional mechanisms would replace emotion-driven signaling with accurate information.
- Hanson's stronger claim is not that ordinary people should acknowledge hidden motives—they may be happily fine ignoring the elephant—but that policy analysts and social scientists who claim to understand the social world well enough to recommend changes should understand hidden motives, so they can choose institutions that produce better outcomes; the fault with signaling lies in the negative externalities and inefficient equilibria, not in the pleasure of showing off.
“policy analysts and social scientists who claim that they understand the social world well enough to make recommendations for changes—theyshould understand the elephant in the brain.”
- Unlike the hard sciences where people eagerly pursue innovation despite cost and risk, social science offers many big powerful potential innovations—new mechanisms and institutions—but these are almost never actually applied, which is precisely why such improvements are easy to find: we don't make the improvements we could.
“The reason why it’s so easy to find big improvements in social science is we almost never actually apply them. We don’t actually make the improvements that we could.”
- Groups forming around a goal work better or worse depending on how well the goal can be verified by the group; effective altruists can commit to giving 10 percent of income and show tax returns—an easily monitored goal—whereas a rationality community struggles because there is no easy way to tell whether members are actually being rational.
“Groups that form around a goal can work better or worse, depending on how well the goal can be verified by the group.”
- The book is not a complaint about hypocrisy; since 90 percent of behavior is signaling, the authors accept it will continue. Pretty much every other living creature is far less admirable and interesting than humans—humans aren't what they pretend to be, but what they actually are is spectacular.
“Humans aren’t what they pretend to be. [laughs] But what they actually are is spectacular.”
- If we had far more prediction markets, the rational way to decide what to believe across a wide range of topics would simply be to look up the prediction market price on any particular question and believe its answer.
“If we had a lot more prediction markets, then you could just look at the prediction market on any particular question and just believe its answer. That would be the usual way to decide what to believe on a very wide range of topics”
- The objection that prediction markets would just teach better signaling and waste resources would imply all technological progress is waste; but since we are already mostly maxed out on signaling effort, new technology lets us get more from the same effort—just as signaling via a faster car still gets us places faster, signaling via prediction markets still yields more accurate answers on topics we care about.
“That argument suggests that all technological progress is a waste because anytime we can do anything better, we’ll just do it in the service of signaling, which is all a waste.”
- Like cost accounting, prediction markets and futarchy face a multiple-equilibria problem: in a world where nobody uses them, proposing to do so sends a bad signal ('people are bullshitting here, we need to find who'), while in a world where everybody uses them, refusing to do so sends an equally bad signal—so adoption depends on which equilibrium prevails, and accurate-forecasting organizations could eventually outcompete more emotionally indulgent ones.
“In a world where everybody else does cost accounting, you say, “Let’s not do cost accounting here.” That will be heard as saying, “Could we steal and just not talk about it?””
Social scientists gain insight by noticing where behavior contradicts intuitive theory or personal expectations.
- Hanson prioritizes discovering things intellectually that he can explain and transfer to other people, as part of a collective enterprise where people work together, discover, and tell each other so all know more over time; personal experiences yielding insight that can't be communicated get low priority because they aren't part of that grand cumulative enterprise.
“Iput a priority on discovering things intellectually that I could explain to other people, that I could transfer to other people.”
- The first thing to understand about any social phenomenon is the middle of the distribution—the typical, modal case—because if you get that wrong you are way off; only after roughly getting the center right does it make sense to study the variance and the dimensions along which it varies, and social science still has the basics of the typical case badly wrong.
“The first thing to understand about anything is the middle of the distribution. If you’ve got that wrong, you’re just way off. You need to get that roughly right before you can start looking at the variance in the distribution”
- Being socially unskilled gives an advantage in social science because socially skilled people intuitively do the right things without explicit thought and never notice that their chalkboard theories don't fit their own behavior, whereas a nerd goes through the social world puzzled, explicitly noticing where people's behavior deviates from what they say—giving them an edge in digging into those puzzles.
“when you’re really socially skilled and you move about in the social world, you just intuitively do all the right things, and you don’t think explicitly about it. You don’t really notice that your theories... don’t actually fit your behavior”
- To think and trade better, specialize: many people feel an obligation to have a defensible opinion about everything, but you and the world are better off if you focus on knowing some things deeply, merely practice thinking about others without confidence, and never bet on what you don't actually know.
“Many people, especially in ordinary conversations, and even in financial markets, they feel this obligation to have an opinion about everything.”
- From a decision-theory standpoint, having many versions of yourself (whether Em copies or many-worlds branches) does not justify taking greater risks, because you still care about how many copies exist—the more the better—so killing off many through stupid risks is still bad, just as ordinary decision trees rapidly branch into billions of future copies of you without warranting reckless behavior toward the distant future.
“If there are many versions of you, you still care about how many there are, and the more the better, so if you take stupid risks that kill a lot off, that’s bad.”
- Subconscious capacities to be hypocritical and engage in clever social manipulation are a separate cognitive capacity from explicit conscious reasoning, which explains why Machiavellianism does not correlate with measures of IQ; some people are simply better at being Machiavellian through superior social skills, but that need not track with high IQ.
“I think it fits the idea that we have all these subconscious capacities to be hypocritical and to engage cleverly socially. Those aren’t connected that much to our explicit conscious reasoning.”
Signaling explains spending patterns in rich societies but not the existence of resources themselves.
- Just as humanity's vastness and power relative to other animals is not primarily explained by signaling but by whatever makes humans special, the existence of an Age of Em economy would not be explained by signaling; signaling explains how rich societies spend their resources, not the existence of those resources themselves.
“the fact that humanity is vast and powerful compared to all the other animals on the earth is not primarily explained by signaling.”
- As CCTV and facial recognition spread, behavior will trend back toward the historical norm in which people were almost never alone and almost always watched—reversing the few hundred years of privacy (private rooms, silent reading) we have grown accustomed to—though this represents a different kind of signaling rather than more or less, since we are already nearly maxed out on signaling.
“We know something about what people used to be like when they lived in small villages... and they were basically almost never alone.”
- Ignoring the Age of Em, futarchy has at least a 30 percent chance of achieving some substantial scope of application within a hundred years; the potential is large enough to be worth trials and investigations, though governing nations is not the place to start initial trials despite being the most exciting framing.
“I think it’s at least a 30 percent chance that it will have some substantial scope in a hundred years.”
- In the Age of Em scenario, brain emulations become feasible and cheap and take over the economy; though Ems experience themselves in virtual reality, they are made of real computers, energy, and structures, so their economy would grow far faster than ours and quickly become visible on a cosmic scale—unless it destroys itself.
“That economy would grow much faster than ours, so it would even more quickly become visible on a cosmic scale unless it destroys itself.”
Early life success requires making what others want rather than purely following personal interest.
- Taxing signaling and subsidizing nonsignaling behavior doesn't make sense as policy because almost everything we do has a big signaling component, so subsidizing nonsignaling would require something like paying people to be alone with nothing to communicate or produce.
“Almost everything we’re doing has a big signaling part. It really would be a very difficult task to subsidize then. You’d have to basically pay people to be alone for an hour or something, every day perhaps, with nothing they can use to communicate or use to generate something else.”
- Hanson's one rule of life is that early in life you are a seller, not a buyer—meaning you should focus on making things that other people want rather than merely pursuing your own interests; he himself only gained this career sense late, going back for his PhD at 34 after years of selfishly following whatever interested him.
“Early in life, you’re a seller, not a buyer.”
Scientific method is a myth; success reflects idiosyncratic field details rather than coherent principles.
- There is no single 'science thing' that explains the success of physics and other fields; what is called the scientific method is a myth—just a name for a collection of idiosyncratic details across different areas that have very little coherent in common.
“What I found is, there is no such thing. It’s a myth. It’s just a name for a whole collection of idiosyncratic detail in different areas that has very little in common, or coherent.”
Book avoids moral outrage and political faction to present hidden motives as empirical facts.
- The book deliberately avoids a moral-outrage voice and avoids connecting to political factions because the first priority is to convince readers of the facts of these hidden motives; a strong moral tone would make readers think the authors are taking sides in other battles and refuse to be persuaded by the arguments.
“we thought if we took on a really strong moral tone or connected this to political factions or something, you wouldn’t be persuaded by our arguments. You would think that we are taking sides in some other battles.”
View quakes peak mid-life from competing effects of learning and exhausted obvious intellectual insights.
- View quake books peak somewhere in the middle of life because of two contrary effects: knowing more makes each new field easier to learn and connect (scale and scope economies), but the easy obvious wins get used up first (diminishing returns); the very young can't have view quakes at all because a view quake requires an existing theory or expectation that can then be surprised or overturned.
“The more you know, the more you can learn. The more you know about many different fields, the more intersections you could make”
Philosophy's primary value is inoculating readers against poor philosophical ideas circulating in society.
- Learning philosophy is useful mainly because it inoculates you against other philosophy; since there is a lot of philosophy loose in the world that you will be exposed to later, it may be in your interest to be exposed to it on purpose early so that you are inoculated.
“I do think learning philosophy is useful mainly because it inoculates you against other philosophy, and there is a lot of philosophy loose in the world.”
Quantum computing is overrated because it enables only narrow applications like factoring and simulation.
- Quantum computing is overrated because, despite futurists' eagerness for big changes, you actually can't do much more with quantum computers than ordinary ones—just a few things like factoring numbers (forcing cryptography changes) and quantum simulations that hardly anyone does—so it doesn't make that much difference.
“once you look at it, you realize you actually can’t do that much with quantum computers that’s more than you can do with ordinary computers.”
Great Filter likely lies ahead of us if evolutionary steps before humans required little difficulty.
- Anything giving us evidence that the steps before our point on the evolutionary path were easy is evidence that the great filter lies harder ahead of us, which is bad news for our future; the comforting hypothesis is that the very first step (origin of life) was the hardest, which would imply the only other galactic life shares our origin point.
“the problem is that anything that gives us evidence that things before us on the path are easier is evidence suggesting it’s harder farther on the path, which is bad news for our future.”