At the link, I speculate on doom, hardware, human jobs, the jagged edge (via a Joshua Gans working paper), and the Manhattan Project. The fun thing about being 6 years late to a seminal paper is that you can consider how its predictions are doing.
Sutton draws from decades of AI history to argue that researchers have learned a “bitter” truth. Researchers repeatedly assume that computers will make the next advance in intelligence by relying on specialized human expertise. Recent history shows that methods that scale with computation outperform those reliant on human expertise. For example, in computer chess, brute-force search on specialized hardware triumphed over knowledge-based approaches. Sutton warns that researchers resist learning this lesson because building in knowledge feels satisfying, but true breakthroughs come from computation’s relentless scaling.
The article has been up for a week and some intelligent comments have already come in. Folks are pointing out that I might be underrating the models’ ability to improve themselves going forward.
Second, with the frontier AI labs driving toward automating AI research the direct human involvement in developing such algorithms/architectures may be much less than it seems that you’re positing.
If that commenter is correct, there will be less need for humans than I said.
Also, Jim Caton over on LinkedIn (James, are we all there now?) pointed out that more efficient models might not need more hardware. If the AIs figure out ways to make themselves more efficient, then is “scaling” even going to be the right word anymore for improvement? The fun thing about writing about AI is that you will probably be wrong within weeks.
Between the time I proposed this to Econlog and publication, Ilya Sutskever suggested on Dwarkesh that “We’re moving from the age of scaling to the age of research“.
Yesterday’s super bowl was fun for a variety of reasons, but your 147th favorite economist was especially happy to see that markets continue to keep things interesting. The NFL was a “only teams with elite quarterbacks can win” league…until it wasn’t. After Brady, Manning, Brees, and Maholmes winning two decades of Super Bowls, we have back to back years of decidedly average quarterbacks winning (within-NFL average, to be clear. These are all objectively incredible athletes). How did this happen? Is it tactical evolution, flattening talent pools, institutional constraints, or markets updating? The answer is, of course, all of the above, but updating markets is the mechanistic straw that stirs the drink.
The NFL is a salary capped, which means each team can only spend so much money on total player salaries. As teams placed greater and greater value on quarterbacks, a larger share of their of their salary pool was dedicated accordingly. These markets are effectively auctions, which means eventually the winner’s curse kicks in, with the winner of the player auction being whoever overvalues the player the most. Iterate for enough seasons, and you eventually arrive at a point where the very best quarterbacks are cursed with their own contracts, condemned to work with ever decreasing quality teammates. Combine that with a little market and tactical awareness, and smart teams will start building their teams and tactics around the players and positions that market undervalues. And that (combined with rookie salary constraints), is how you arrive at a Super Bowl with the 18th and 28th salary ranked quarterbacks.
Whenever a market identifies an undervalued asset (i.e. quarterbacks 25 years ago) there will, overtime, be an update. Within that market updating, however, is a collective learning-as-imitation that eventually results in some amount of overshooting via the winners curse. This overshoot, of course, may only last seconds, as market pressure pushes towards equilibrium. In markets like long term sports contracts or 12 year aged whiskey, that overshoot can be considerable, as mistakes are calcified by contracts and high fixed cost capital.
What does this predict? In a market like NFL labor, I’d expect a cycle over time in the distribution of salaries, iterating between skewed top-heavy “star” rosters and depth-oriented evenly distributed rosters. At some point a high value position or subset of stars are identified and distproportionately committed to, but the success of those rosters eventually leads to over-committment, so much so that the advantage tilts towards teams that spread their resources wider across a larger number of players undervalued teams whose fixed pie of resources are overcommitted to a small number of players. That’s how you get the 2025 Eagles and 2026 Seahawks as super bowl champions.
I wonder when it will cycle back and what the currently undervalued position will be?
The title of this post, “everyone take copies,” comes from a conversation between the human subjects in an experiment in our lab, on which the paper is based. The experiment was studying how and when people take resources from one another.
Here’s a tip that doesn’t require any piracy. For those of you who are tired of the subscription economy fees, I think it’s safe to say in 2026 that anyone in the United States can find a local thrift store or annual rummage sale with oodles of nearly-free media. DVDs for a dollar. Used books for a dollar. Basically you are paying the transaction costs – the media itself is free. (I typed that dash myself, not AI!)
“Buying” a movie to stream on Amazon Prime can run over $20. Buying a used DVD is usually less than $10.
A village charges a boy with watching the flock and raising the alarm if wolves show up. The boy decides to have a little fun and shout out false alarms, much to the chagrin of the villagers. Then an actual wolf shows up, the boy shouts his warning, but the villagers are proper Bayesians who, having learned from their mistakes, ignore the boy. The wolves have a field day, eating the flock, the boy, and his entire village.
I may have augmented Aesop’s classic fable with that last bit.
The boy is certainly a crushing failure at his job, but here’s the thing: the village is equally foolish, if not more so. The boy revealed his type, he’s bad at his job, but the village failed to react accordingly. They updated their beliefs but not their institutions. “We were good Bayesians” will look great on their tombstones.
They had three options.
A) Update their belief about the boy and ignore him.
This is what they did and look where that got them. Nine out of ten wolves agree that Good Bayesians are nutritious and delicious.
B) Update their beliefs about the boy, but continue to check on the flock when the boy raises the alarm.
They should have weighted their responses. Much like Pascal taking religion seriously because eternal torment was such a big punishment, you have to weight you expected probability of truth in the alarm against the scale of the downside if it is true. You can’t risk being wrong when it comes to existential threats.
C) Update their beliefs about the boy and immediately replace him with someone more reliable.
It’s all fine and good to be right about the boy being a lying jerk but that doesn’t fix your problem. You need to replace him with someone who can reliably do the job.
So this is a post about fascism. Some think that fascism is already here, others dismiss this as alarmism, others splititng the difference claiming that we are in some state of semi- or quasi-fascism. Within the claims that it is all alarmism, what I hear are the echoes of villagers annoyed by 50 years of claims that conservative politics were riddled with fascism, that Republicans were fascists, that everything they didn’t like was neoliberalism, fascism, or neoliberal fascism. Get called a wolf enough times and you might stop believing that wolves even exist.
Even if I am sympathetic, that doesn’t get you off the hook. It hasn’t been fascism for 50 years will look pretty on your tombstone.
Let’s return to our options
A) Don’t believe the people who have been shouting about fascism for years, but take seriously new voices raising the alarm.
B) Find a set of people who, exogenous to current events, you would and do trust and take their warnings seriously.
C) Don’t believe anyone who shouts fascism, because shouting fascism is itself evidence they are non-serious people.
D) Start monitoring the world yourself
Both A) and B) are sensible choices! If you’ve Bayesian updated yourself into not trusting claims of fascism from wide swaths of the commentariat, political leaders, and broader public, that’s fine, but you’ve got to find someone you trust. And if that leads you to a null set, then D) you’re going to have to do it yourself. Good luck with that. It takes a lot of time, expertise, and discipline not to end up the fascism-equivalent of an anti-vaxxer who “did their own research.”
Because let me tell you, C) is the route to perdition in all things Bayesian. Once your beliefs are mired in a recursive loop of confirmation bias, it’s all downhill. Every day will be just a little dumber than the one before. And that’s the real Orwellian curse of fascism.
This post is just some thoughts about perspective. I apologize for any lack of organization.
My academic influences include North, Weingast, Coase, Hayek, the field of Public Choice, and others. I’m not an ‘adherent’ to any school of thought. Those guys just provided some insights that I find myself often using.
What lessons did they teach? Plenty. When I see the world of firms, governments, and other institutions, I maintain a sharp distinction between intention and outcome. Any given policy that’s enacted is probably not the welfare maximizing one, but rather must keep special interests relatively happy. So, the presence of special interests is a given and doesn’t get me riled up. When I see an imperfect policy outcome, I think about who had to be enticed to vote for it. We live in a world where ‘first bests’ aren’t usually on the table.
Historically, or in lower income countries, I think about violence. Their rules and laws are not operating in a vacuum of peaceful consent. There is always the threat of violence. Laws are enforced (or not) conditional on whether and what type of violence that may result. All of the ideal legislation is irrelevant if theft and fraud are the lay of the land.
I think about institutional evolution with both internal and external pressures. I’m a bit worried about the persistence of the US republic, or at least worried for its pro-growth policies. I’m not worried about China in the long run. I don’t think they have the institutions that get them to ‘high income’ status. I do think that they are a tactical concern in the short run and that the government does/will have access to great volumes of resources in the medium run. That’s a bit of a concern. But like I said, I’m not super worried in the long run.
Pushing beyond the despair and doomerism of “Nothing matters”, the question has never been is there a price for lying in politics, but rather what is the price of lying in politics. Note that “in politics” is doing a lot of heavy lifting here. In day to day life, the price of lying is the threat to your reputation. A reputaton for being untrustworthy is always very costly in the long run. But politics, however, has different layers across which the price of lying is heterogeneous. And yes, there are contexts where that price can go negative.
Put simply, what is the cost here? Is Greg Bovino, head of US Border Patrol, worried about his reputation? Is he worried about future personal legal liability? Is he worried about maintaining cooperative alignment across the administration and within the ranks of the Border Patrol and ICE? There’s a saying in politics – the worst thing you can do is tell the truth at the wrong time. But that’s more relevant to “lying by omission”, about simply abstaining from speaking on a subject so that you are not forced to choose between lying and paying a high political cost. This is different. I’m picking on this one person in the administration because Alex Pretti was summarily executed in the street in cold blood by a thicket of federal agents for the apparent crime of being in attendance and trying to help a woman while she was being pepper sprayed, but it is the subsequent lying that I am concerned with here. It follows a pattern that continues to darkly fascinate me.
Q: "Was Alex Pretti armed when he was shot?"Bovino: "The investigation is going to uncover all those facts…I wasn't there wrestling that assaultive subject that was assaulting Border Patrol agents."
Rather than simply “say nothing”, this administration has committed to the broad tactic of stating things that are factually, obviously untrue. That, more important, it is highly likely they know are untrue. That’s not something we’ve seen a lot of before. Politicians were known for being “slick” and “slippery”. For bending the truth, torturing the facts, or managing to fill entire press conferences without saying or committing to anything of substance. This administration, as I’ve said before, is different.
I see two likely explanations:
The price of lying is zero because no one believes anything anymore. The truth is subjective and siloed.
The price of lying is negative because constant and consistent commitment to the party can only be demonstrated by bearing the personal cost of telling obvious lies. In doing so you maintain the group, save yourself from being purged, and everyone in the group lives to fight another day. The net of which is a negative price for lying.
So what is it? Are we through the lol-nothing-matters looking glass, or are we witnessing an administration circle the wagons and solidify their committment to one another by blatantly lying on national television? I’m (perhaps obviously) of the belief that everything matters, that lying does have a cost, but the need for unity is so strong within this administration and it is, in fact, the lying that is holding it together. Until, of course, it doesn’t. Remember the most important lesson of The Folk Theorem – you can sustain cooperation in the Prisoner’s Dilemma, but only until you learn when the game is going to end. Then all bets are off.
“Who serves in secret police forces? Throughout history, units such as Hitler’s Gestapo, Stalin’s NKVD, or Assad’s Air Force Intelligence Directorate have been at the core of state repression. Secret police agents surveil, torture, and even kill potential enemies within the elite and society at large. Why would anyone do such dirty work for the regime? Are these people sadistic psychopaths, sectarian fanatics, or forced by the regime to terrorize the population? While this may be the case for some individuals, we believe that the typical profile of secret police agents is shaped by the logic of bureaucratic careers.”
The details and history in the paper are illuminating. The economic logic is simple, but it remains fascinating to be reminded of how far the reinforcing incentives of shame, power, and labor market demand can go when trying to understand the world. To recap the obvious
For some the opportunity for cruelty is benefit and others a cost, no doubt heterogeneous across context for many (but not all). The selection effects into ICE officers is obvious.
Shame selects as well. The larger the fraction of the American public that view ICE behavior as shameful and cruel, the fewer and more specific the individuals who will select in.
Labor demand for individuals is heterogeneous in multiple dimension, but it always weaker for those who are broadly incompetent.
Combine those three and you get what we are observing: those with the weakest opportunities in the labor market are selecting into ICE service because they face the lowest opportunity cost. If there is a positive correlation between enjoying cruelty and weak labor market opportunties (which I am willing to believe there is. Few enjoy working with ill-adjusted, cruel people), then the broad incompetence selected into ICE ranks will be stronger. If being ill-adjusted and cruel limits the scale of your social network, leaving you isolated and lonely, then the expected shame of ICE services is lower, selecting for still greater cruelty within officers. Through this mechanism cruelty and incompetence don’t just correlate, they reinforce, until you are left with a very specific set of individuals exercising violent discretion.
To be clear this isn’t a complex or profound model. The individual insights are obvious, but it remains useful to consider them within the framework of a toy model because they emphasize how mutually-reinforcing incentives can create shocking institutional outcomes.
I’m writing because I am catching up on the backlog of The Answer is Transaction Costs (TAITC), a podcast hosted by Michael Munger. Specifically, in an episode published August 27, 2024, a listener writes asking about what seems to be the extremely costly practice of interviewing college applicants prior to acceptance.
As it turns out, I work at a private university that enacted an interview policy in a quasi-random way and the university president gave me permission to share.
Initially, my university did not interview standard applicants. Our aid packages were poorly designed because applicants tend to look similar on paper. There was a pooling equilibrium at the application stage. As a result, we accepted a high proportion and offered some generous aid packages to students who were not good mission fits and we neglected some who were. Aid packages are scarce resources, and we didn’t have enough information to economize on them well.
The situation was impossible for the admissions team. The amount of aid that they could award was endogenous to the number of applicant deposits because student attendance drives revenue. But, the deposits were endogenous to the aid packages offered! There was a separating equilibrium where some good students attended along with some students who were a poor fit and were over-awarded aid. The latter attended one or two semesters before departing the university, harming retention and revenues. Great but under-awarded students tended not to attend our university. Student morale was also low due to poor fits and their friends leaving.
The confluence of politics, recent interest in agent-based computational modeling, and Pluribus have convinced me now is the time to write about the “Cooperative Corridor”. At one point I thought about making this the theme of a book, but my research has become overwhelmingly about criminal justice, so it got permanently sidelined. But hey, a blog post floating in the primordial ether of the internet is better than a book that never actually gets written.
It’s cooperation all the way down
Economic policy discussions are riddled with “Theories of Everything”. Two of my favorites are the “Housing” and “Insurance” theories of everything. Housing concerns such huge fractions of household wealth, expenditures, and risk exposure that the political climate at any moment in time can be reduced to what policy or leader voters think is the most expedient route to paying their mortgage or lowering their rent. Similarly, the decision making of economic agents can, through a surprisingly modest number of logical contortions, always be reduced to efforts to acquire, produce, or exchange insurance against risk. These aren’t “monocausal” theories of history so much as attempts to distill a conversation to a one or two variable model. They’re rhetorical tools as much as anything.
My mental model of the world is that it is cooperation all the way down. Everything humans do within the social space i.e. external to themselves, is about coping with obstacles to cooperating with others. It is a fundamental truth that humans are, relative to most other species, useless on our own. There are whole genres of “survival” reality television predicated on this concept. If you drop a human sans tools or support in the wilderness, they will likely die within a matter of days. This makes for bad television, so they are typically equipped with a fundamental tool (e.g. firestarting flint, steel knife, cooking pot, composite bow, etc) after months of planning and training for this specific moment (along with a crew trained to intervene if/when the individual is on the precipice of actual death). Even then, it is considered quite the achievement to survive 30 days, by the end of which even the most accomplished are teetering on entering the great beyond. No, I’m afraid there is no way around the fact that humans are squishy, nutritious, and desperately in need of each other. Loneliness is death.
Counterintuitive as it may be, this absolute and unqualified dependence on others doesn’t make cooperation with others all that much easier. This is the lesson of the Prisoner’s Dilemma, that our cooperation and coordination isn’t pre-ordained by need or even optimality. Within a given singular moment it is often in each of our’s best interest to defect on the other, serving our own interests at their expense.
Which isn’t to say that we don’t overcome the Prisoner’s Dilemma every day, constantly, without even thinking about it. Our lived experience, hell, our very survival, is evidence that we have manifested myriad ways to cooperate with others despite our immediate incentives. What distinguishes the different spaces within which we carry out our lives is the manner in which we facilitate these daily acts of cooperation.
Kin
The first and fundamental way to solve the prisoner’s dilemma is to change the payoffs so that each player’s dominant strategy is no longer to defect but instead to cooperate. If you look at the payoff matrix below, the classic problem is that no matter what one player does (Cooperate or Defect), the optimal self-interested response is always to Defect. Before we get into strategies to elicit cooperation, we should start with the most obvious mechanism to evade the dilemma: to care about the outcome experienced by the other. Yes, strong pro-social preferences can eliminate the Prisoner’s Dilemma, but that is a big assumption amongst strangers. Among kin, however, it’s much easier. Family has always been the first and foremost solution. Parents don’t have a prisoner’s dilemma with their children. It doesn’t take a large leap of imagination to see how kin relationships would help familial groups coordinate hunting and foraging or il Cosa Nostra ensuring no one squeals to the cops.
Kinship remains the first solution, but it doesn’t scale. Blood relations dilute fast. I’m confident my brother won’t defect on me. My third-cousin twice removed? Not so much. The reality is that family can only take you so far. If you want to achieve cooperation at scale, if you want to achieve something like the wealth and grandeur of the modern world, you’re going to need strategies and institutions.
Strategies
There are many, if not countless, ways to support cooperation among non-kin. Rather than give an entire course in game theory, I’ll instead just enumerate a few core strategies.
Tit-for-Tat = always copy your opponent’s previous strategy
Grim Trigger = always cooperate until your opponent defects, then never cooperate again
Walk Away = always cooperate, but migrate away from prior defectors to minimize future interaction
The Prisoner’s Dilemma is far, far easier to solve amongst players who can reasonably expect to interact again in the future. The logic underlying all of these strategies is commonly known as The Folk Theorem, which is the broad observation that all cooperation games are far easier to solve, with a multitude of cooperation solutions, if there is i) repeated interaction and ii) an indeterminate end point of future cooperation.
Strategies can facilitate cooperation with strangers, which means we can achieve far greater scale. But not as much as we observe in the modern world, with millions of people contributing to the survival of strangers over vast landscapes and across oceans. For that we’re going to need institutions.
Institutions
Leviathan is simply Thomas Hobbes’ framework for how government solves the Prisoner’s Dilemma. We concentrate power and authority within a singular institution that we happily allow to coerce us into cooperation on the understanding that our fellow citizens will be coerced into cooperating as well. That coercion can force cooperation at scales not previously achievable. It can build roads and raise armies. This scale of cooperation is the wellspring for both some of the greatest human achievements and our absolutely darkest and most heinous sins. Sometimes both at same time.
Governments can achieve tremendous scale, but there remain limits. My mental framing has always been that individual strategies scale linearly (4 people is twice as good as 2 people) and governments scale geometrically (i.e. an infantry’s power is always thrice its number). Geometric scaling is better, but governments always eventually run into the limits of their reach. Coercion becomes clumsy and sclerotic at scale. There’s a reason there has never been a global government, why empires collapse.
Markets can achieve scale unthinkable by governments because their reach is untethered to geography. Markets are networks. They scale exponentially. They solve the prisoner’s dilemma through repeated interaction and reputation. The information contained in prices supports search and discovery processes that both support forming new relationships while also creating sufficient uncertainty about future interactions. Cooperation is a dominant strategy. This scale of cooperation, of course, is not without critical limitations. Absent coercion there is no hope for uniformity or unanimity. No completeness. Public goods requiring uniform commitment or sacrifice are never possible within markets. The welfare of individuals outside of individual acts of cooperation (i.e. externalities) is not weighed in the balance.
There are other institutions that solve the prisoner’s dilemma. Religions, military units, sororities…the list goes forever. This article is already going to be too long, so I’ll start getting to the point. Much of the fundamental disagreement within politics and society at large is what comprises our preferred balance of institutions for supporting and maintaining cooperation, who we want to cooperate with, and the myths we want to tell ourselves about who we are or aren’t dependent on.
The Cooperative Corridor
Wealth depends on cooperation at scale. Wealth brings health and prosperity, but it also brings power. The “cooperation game” might be the common or important game, but it isn’t the only game being played. Wealth can be brought to bear by one individual on another to extract their resources. This is colloquially referred to as “being a jerk”. Perhaps more importantly, groups can bring their wealth to bear to extract the resources from another group. This is colloquially referred to as “warfare”.
Governments are an excellent mechanism for warfare. All due respect to the mercenary armies of history (Landsknechts, Condottieri, etc.), but markets are not well-suited to coordinate attack and defense. Which isn’t to say markets aren’t necessary inputs to warfare. This is, in fact, the rub: governments are good at coordinating resources in warfare, but markets are far better at generating those resources. A pure government society may defeat a pure market society in a war game, but a government-controlled society whose resources are produced via market-coordinated cooperation dominates any society dominated by a singular institution.
This all adds up to what I refer to as the Cooperative Corridor. A society of individuals needs to cooperate to grow and thrive. A culture of cooperation can be exploited, however, by both individuals who take advantage of cooperative members and aggressive (extractive) rival groups. Institutions and individual strategies have to converge on a solution that threads this needle. One answer might appear to be to simply cooperate with fellow in-group members while not cooperating with out-group individuals. This is no doubt the origin of so many bigotries—the belief that you can solve the paradox of cooperation by explicitly defining out-group individuals. Throw in the explicit purging of prior members who fail to cooperate, and you’ve got what might seem a viable cultural solution. The thing about bigotry, besides being morally repugnant, is that it doesn’t scale. The in-group will, by definition, always be smaller than the out-group. Bigotry is a trap. Your group will never benefit from the economies of scale as much as other groups that manage to foster cooperation between as many individuals as possible, including those outside the group.
[SPOILERS AHEAD if you haven’t watched through Episode 6]
You’ve been warned, so here’s the spoilers. An RNA code was received through space, spread across the human species, and now all but a handful of humans are part of a collective hive mind whose consciousnesses have been fully merged. That’s the basic part. The bit that is relevant to our discussion is the revelation that members of the hive mind 1) Can’t harm any other living creature. Literally. They cannot harvest crops, let alone eat meat. 2) They cannot be aggressive towards other creatures, cannot lie to them, cannot it seems even rival them for resources. 3) The human race is going to experience mass starvation as a result of this. Billions will die.
In other words, a cooperation strategy has emerged that spreads biologically at a scale it cannot support. It is also highly vulnerable to predation. If a rival species were to emerge in parallel, it would undermine, exploit, enslave, and eventually destroy it. The whole story borders on a parable of how a species like Homo sapiens could destroy and replace a rival like Homo neanderthalensis.
Cultural strategies are selected within corridors of success. Too independent, you die alone. Too cooperative, you die exploited. Too bigoted, you are overwhelmed by the wealth and power of more cosmopolitan rivals. Too cosmopolitan, you starve to death for failure to produce and consume resources. Don’t make the mistake of thinking the “corridor of success” is narrow or even remotely symmetric, though. On the “infinitely bigoted” to “infinitely cosmopolitan” parameter space, a society is likely to dominate it’s more bigoted rivals with almost any value less than “infinitely cosmopolitan.” So long as members of society are willing to harvest and consume legumes, you’re probably going to be fine (no, this isn’t a screed against vegetarianism, which is highly scalable. Veganism, conversely does have a much higher hurdle to get over…). So long as a group is willing to defend itself from violent expropriation by outsiders, they’re probably going to be fine. Only a sociopathic fool would see empathy as an inherent societal weakness. Empathy, in the long run, is how you win.
How this relates to political arguments
I almost wrote “current political arguments”, but I tend to think disagreements about institutions of cooperation are pretty much all of politics and comparative governance. We’re arguing about instititutions of in-group, out-group, and collective cooperation when we argue about the merits of property rights, regulation, immigration, trade, annexing territory, war. When we confront racism, nationalism, and bigotry, we we are fighting against forces that want to shrink the sphere of cooperation and leverage the resources of the collective to expropriate resources of those confined or exiled to the out-group. These are very old arguments.
The good news is that inclusiveness and cosmopolitanism are economically dominant. They will always produce more resources. But being economically and morally superior doesn’t mean they are necessarily going to prevail. The world is a complex and chaotic system. The pull towards entropy is unrelenting. And, in the case of cultural institutions and human cooperation, the purely entropic state is a Hobbesian jungle of independent and isolated familial tribes living short, brutish lives. Avoiding such outcomes requires active resistance.
Economist Writing Every Day 