This post is a resolution to a series of questions I’ve asked myself over the past six months. My answer to those questions is that there’s a dichotomy between two kinds of people:
Clouds make decisions that maximize utility, making rapid decisions and ignoring rules and norms to do so, sometimes to the detriment of those around them. Atlases tend to follow prosocial rules and norms, often benefitting society at their own expense.
I think society today, or at least the society I surround myself with, is increasingly composed of clouds. I think these people would be happier if they adopted the cloud mindset with the prosocial goals of the atlas mindset. This essay is a roundabout way of making that argument.
2.
Last December, while writing a screenplay at a café in Guatemala, I stumbled across Ishtar, the Sumerian goddess of love and fertility. I thought I knew what that meant. I read The Kane Chronicles as a kid and was very familiar with Tawaret, the corresponding Egyptian goddess of marriage and fertility. In mythology, Tawaret was a fierce protector of the home and of childbirth, acting as a mother herself. She was considered dangerous and had some traits of a predator, including lion paws and a crocodile tail, but used that strength for protection rather than harm.
I couldn’t have been more wrong. Ishtar is the goddess of fertility in the sexual sense rather than the motherly sense. She is also the goddess of war. She is capricious and vengeful, carrying out divine justice and retribution. In the Epic of Gilgamesh, she asks Gilgamesh to become her consort; when he rejects her, she sends the Bull of Heaven in an attempt to kill him. Ishtar’s traits are much closer to those of Sekhmet, a warrior goddess who was the manifestation of Ra’s divine power.
Ishtar and Tarawet are nearly polar opposites. Ishtar is passionate, vengeful, and makes decisions on a whim. She constantly strives for power and tries to kill those who oppose her. Tarawet, on the other hand, uses her power to protect the vulnerable and ensure they can flourish. If Ishtar is a goddess of death, Tarawet is a goddess of life.
3.
The screenplay I was writing in Guatemala was based on The Lottery in Babylon by Jorge Luis Borges, one of my favorite short stories. In Borges’s Babylon, a fictional Company introduces the Lottery, initially a drawing of lots that includes both positive and negative cash rewards. When those who earn negative rewards refuse to pay, the Company assigns them jail time and other creative punishments. Over time, the Lottery grows to be all-encompassing with no limits on its results. The story’s opening paragraph (translated by Andrew Hurley) captures the idea:
Like all the men of Babylon, I have been proconsul; like all, I have been a slave. I have known omnipotence, ignominy, imprisonment. Look here—my right hand has no index finger. Look here—through this gash in my cape you can see on my stomach a crimson tattoo… In the half-light of dawn, in a cellar, standing before a black altar, I have slit the throats of sacred bulls. Once, for an entire lunar year, I was declared invisible—I would cry out and no one would heed my call, I would steal bread and not be beheaded. I have known that thing the Greeks knew not—uncertainty.
Borges writes that early lotteries with only cash rewards failed because they “appealed not to all a [person’s] faculties, but only to [their] hopefulness.” A lottery that encompassed all of society was sure to add variety to their life that canceled out the fear of loss. Later in the story, Borges leaves doubt that the Company exists at all. It may be that the Lottery is sustained by individual people leaving slips for others with arbitrary commands, and their downstream effects are enough to sustain the Lottery forever. It makes no difference either way. The Company is not special; its legitimacy comes from the fact that people listen to its demands. Given that, the actions of individuals are just as legitimate as those of a Company because they will be followed either way.
The screenplay I mentioned was written for Hackuba 2, a week-long hackathon I organized in Antigua, Guatemala. I spent the hackathon learning photography and attempting to put together a screenplay based on The Lottery in Babylon. While I wrote, I had inspiration from Michael, who attempted to decide payment for every meal via Chwazi—randomly assigning the bill to someone in the group. In economic terms, loss aversion and declining marginal utility mean that this reduces average utility for the group. But for the several people who participated, that reduction was more than canceled out by the fun of the draw.
This is the appeal of gambling, lotteries, and the Lottery: net-zero or net-negative trades are offset by the thrill of the risks and rewards involved. Only a certain kind of person would participate in a Chwazi or the Lottery that Borges imagines.
4.
I recently wrote a post comparing “mission mode” and “fallback mode.” People operating in mission mode focus on getting a small number of things done as well as possible. The classic example is 1Day Sooner and similar campaigns that worked to produce a COVID vaccine as fast as scientifically possible. In contrast, people operating in fallback mode attempt to never drop a ball and to make sure no important item slips through the cracks. This is the most common mode that governments operate in, since they are asked to deal with an enormous number of important issues that matter for the daily lives of their citizens.
Importantly, fallback mode can’t work all the time. Some issues take too much time and effort to solve when not focused entirely on them. For example, the Abundance movement argues that American industry and infrastructure have stagnated because they have been neglected by normal government operations, leaving overly restrictive laws on the books for too long. Similarly, the Green New Deal (despite being old news by now) intended to solve America’s contribution to climate change by putting energy entirely towards producing a clean economy.
This is why successful startup founders and scientists often act in mission mode. These people have everything to gain and very little to lose, so they can afford to take some risks to make progress in specific fields. In contrast, established companies and governments generally can’t afford to disrupt their normal functioning to accomplish a single goal.
5.
There’s a shared theme in the world of startups, Silicon Valley, and rationality. That theme is agency. Agency can be defined in all sorts of ways, but its core definition is a willingness to take unusual actions to get what you want. In other words, “you can just do things.” An agentic person might build a nuclear fusor to learn mechanical engineering, or they might lie about being Jain to get off the Stanford meal plan. Its core requirement is simply thinking out of the box, but that often involves breaking rules and norms that seem to be harmful or at least unproductive.
Everyone in Silicon Valley is looking for founders with agency. It’s rare to find someone who genuinely believes they have the power to shape the world around them and does so effectively. These people are often high-risk, high-reward, which means a bet on them could pay off very, very well. The most successful companies come from good ideas that no one before has had the vision or talent to implement.
6.
I think the four dichotomies above—the Lottery, Ishtar, mission mode, and agency—map onto the same fundamental difference. To extract what that difference is, I’m going to frame it in the terms of a recent conversation I had at Stanford in Beijing.
My friend Darrius argues that people should behave based on obligations to other people in society. Parents are obligated to raise their kids; in return, children should give back to their parents when they grow older. People give gifts to others on their birthdays and on Christmas because that’s what they’re expected to do. In the workplace, people should do what they’re told to make the organization they work in run more smoothly. Those who follow these rules are considered dependable and hardworking, and they’re the people who make high-trust societies function. Deviating from one’s obligations makes the world a worse place.
In contrast, I think people should do things because they want to do them, and each of those “obligations” can be framed as a want. As Nate Soares argues in Replacing Guilt, this allows people to make decisions that better serve themselves, and it removes the unpleasant feeling of guilt that comes when you fail to meet an obligation. Parents raise their kids and children give back to their parents not because they should but because they love and care about each other. Gifts are common because gift-givers care about the people who receive them, making the giver happy vicariously. In the workplace, people should do what they’re told if they believe it works towards the long-term mission of the organization, and they should push back if they find a better way to accomplish that mission.
Consider someone who wants to go to the gym more often, rather than sitting on the couch eating ice cream. Darrius would say that he should go to the gym, and he probably would. If he ended up on the couch instead, he would feel guilty, with some internal sensation that he’s a bad person. If I were making the same decision, I would weigh the two outcomes. If I went to the gym, I would burn some calories and be more physically fit. If I stayed home, I would enjoy my TV and ice cream, but I’d be getting much less fit, which is something I’m not happy about. That framing helps me reasonably weigh the options and pick the outcome I prefer. In some cases—say, if I’ve been every day in the last two weeks—I would think I’m better off taking a break and staying home, and I wouldn’t feel guilty about picking that better option.
I’m going to use the term atlases for people who use the obligation mindset and clouds for people who use the want mindset. Real atlases are steady, dependable, and do their job well, while clouds are unpredictable and change between moments based on their environment. The following sections explain why I think these terms explain the kinds of people mentioned above.
7.
Darrius uses the atlas mindset because he argues that, like a real atlas, people who follow their obligations are dependable. In his mandatory military service as a Singaporean citizen, he would have been appreciated for exercising daily, being prepared for every drill and activity, and following orders from his superiors. Militaries train for that exact discipline because they understand that the organization runs more smoothly that way. For the same reason, large corporations rely on people doing simple but valuable jobs, like Walmart relying on employees to check out customers and restock shelves.
In contrast, my friend Justin uses the cloud mindset because he argues that the ability to be creative and to innovate is the driving force of society. Like a cloud, people are most valuable when they are unpredictable. Society prizes people who buck conventional wisdom and end up being correct—in other words, people who do what they want when they think it’s right. Successful startup founders, for example, build companies around new ideas that are genuinely useful to society. Every good, technology, and idea we have is downstream of someone who had the brilliance to invent it.
I follow the cloud mindset because I think it’s more useful for accomplishing a given goal. If you really want a certain outcome, the best way to achieve it is to take the actions that actually achieve that outcome. The obligations of the atlas mindset might be the best way to do that, but if they’re not, the mindset adds unnecessary pressure to do something that doesn’t make sense. Consider the gym example. It might be that only working out every other day is most sustainable for your long-term health, so the cloud mindset would choose that. Someone with the atlas mindset would try to go as often as possible until they inevitably fail, ruining the entire plan. New Year’s resolutions fail for exactly this reason; they only succeed when you choose a plan that happens to be sustainable.
In some cases, obligations do add extra force to push for a decision that leads to a better outcome. In the gym example, it’s often a good idea to exercise, but it’s easy to not think through the entire chain of events and simply decide to rest. In that case, guilt is a useful motivating factor that would lead to a better outcome. Still, people who follow the cloud mindset often find effective substitutes for guilt, like paying a friend $10 every time they deviate from their routine. Someone thinking clearly about their decisions can implement a system to help choose the best decision most of the time. Guilt may accomplish the same thing, but it’s not necessarily helpful.
That said, I believe the atlas mindset is legitimately useful when people can’t be trusted to make good decisions. In corporate and military contexts, for example, there are often managers or leaders who genuinely are more informed about the direction of the organization and can make good strategic decisions. In that case, the organization will do better if its low-level employees are disciplined and follow instructions rather than trying to act on their own without high-level coordination.
Some would argue that this should happen on a societal scale, and it does to some extent. Murder and terrorism are illegal because anyone who decides to commit those crimes will seriously damage society. The fundamental purpose of the law is to restrict some freedom of action in a way that, on average, benefits society. But not all social coordination is done by governments! Religion, too, imposes obligations on its members that tend to improve society. Christianity’s obligations to love thy neighbor and give to the poor are intended to produce a more stable and equal society that people would prefer to live in. The Church wants to stop selfish behaviors that improve life for an individual but worsen society on average.
More rigorously, that’s a basic prisoner’s dilemma. In a one-shot game without coordination, people making the most informed decision—i.e., people with the cloud mindset—will act selfishly and defect. But we don’t want everyone in society to defect; we want to coordinate people so that they cooperate as much as possible. The atlas mindset is enormously useful in this situation because people feel obligated to cooperate, which improves society on average. A society of atlases who all cooperate will obviously be better off than a society of clouds who all defect.
8.
My problem with agency is that it encourages people to defect in a world of cooperation. That works on a small scale, but I wouldn’t want to live in a world of agentic people.
The prisoner’s dilemma is rarely accurate because it’s a negative-sum game, while many real-world games are positive-sum. Startup apparatuses, for example, exist because they’re positive-sum. Accelerators like YC bear the cost of starting a company with the expectation that it will on average produce more value than that cost. But people aggressively trying to build startups will often take trades that benefit them but negatively affect society. Cluely, for example, was essentially a marketing campaign for its founder, and it was apparently quite effective given I’m still talking about it. But that campaign created a widespread distaste for AI that both makes people unhappy and makes genuinely useful companies harder to grow. YC would never fund Cluely because it actively hurts the other companies in its portfolio.
Many of the most agentic people I know achieve their personal goals—status, popularity, funding—using socially destructive behaviors. Lying and code-switching are quite common because they allow someone to extract the most value from conversations and friendships that they then leave behind, analogous to a one-shot prisoner’s dilemma. One old friend of mine told a few lies to build our relationship; by the time I realized, they had begun courting the people I’d introduced them to. Analogously, Sam Bankman-Fried tried to make money for EA purposes by misusing customer funds in a way he believed was harmless. It worked out well for him, but when he was eventually caught, he both lost that money and permanently damaged the movement he had been working to help. His decision was agentic, selfish, and poorly thought out.
Hyper-agency is the cloud mindset operating to its maximum extent. These people and companies operate effectively because they take advantage of a culture of trust. The prevalence of the atlas mindset means people can be assumed to be reliable, allowing those who are not to get away with more. If everyone operated like them, they would not succeed, but legitimately trustworthy people and ventures also wouldn’t succeed because there would be so many guardrails to stop bad actors. Unfortunately, that’s the world we can expect as agency and the cloud mindset spread.
9.
Operating a government or organization in mission mode is analogous to the cloud, or agentic, mindset. Operating in fallback mode is analogous to the atlas mindset. I think that’s the clearest of these analogies. Remember that I defined the cloud mindset as focused on a single objective (mission), while the atlas mindset is dependable and has absolute rules to reduce the risk of a bad outcome. I also noted that cloud-mindset startups operate in mission mode because they’re trying very hard to build something new, while established companies avoid agency and operate in fallback mode because they want to preserve what’s working.
Mission mode is incredibly effective in many situations, and my original post advocates for using it more in government. Importantly, though, it can’t be used in every situation. Mission mode works in cases like Operation Warp Speed and the creation of Taiwan’s chip industry because both have an incredibly high upside with very little downside. In the COVID example, a rapid response saves millions of lives; if it fails, the government just burns some money, which it does all the time. In Taiwan, the success of TSMC has built the island into a high-income economy integral to world supply chains; if it failed, the country would simply remain in the status quo.
Importantly, mission mode isn’t applicable in every situation, and I think fallback mode deserves more praise. You can’t use mission mode to improve Medicare because there’s very little upside and a massive downside. If your new method works, it saves people some time; if it doesn’t, millions of people have no access to affordable healthcare. Most bureaucratic inefficiency originates from dynamics like this, where some improvement is possible but not worth the risk. Laws on the books tend to stay on the books because they work well enough, but a reimagined system might not work at all, and not working is catastrophically bad for an established government.
I think some people who consider themselves agentic apply their mission-mode thinking to too many situations. Defaulting to ignoring rules and norms ignores the reasons those norms were put in place. Sometimes, the right outcomes are hard to think about properly, so society institutes obligations that correct our thinking into producing those outcomes. Agency and cloud-based thinking cause people to ignore those important corrections.
10.
I think the kind of people who participate in the Lottery in Babylon are those for whom losing means nothing—in other words, those with the confidence to use the cloud mindset. People with the atlas mindset want reliable happiness, which they usually already have, so they don’t care enough to partake in games of chance.
When we tried to Chwazi the lunch bill in Guatemala, those who chose to participate fit a certain mold. They all had plenty of money—they were the ones who ordered the most, and didn’t seem bothered by the differences in their respective spending. They were also the extroverts who seemed to enjoy the adrenaline of the game. All were male. On the other hand, we had several people who were clearly conscious of their spending, and they didn’t think twice about declining to participate.
These are the kinds of people who would sustain the Lottery in Babylon. The reason lotteries rarely make sense is that money has declining marginal utility—ten million dollars doesn’t make you a million times as happy as ten dollars. The kind of lottery Borges pictures also suffers from loss aversion, the human tendency to avoid the risk of loss. These effects add up to an asymmetry: gains add much less happiness than the happiness lost from an equivalent loss.
The people who participate in the lottery are those for whom the adrenaline and excitement outweighs the asymmetry of loss. They tend to first be people who have very little loss asymmetry, typically because they have no reason to fear loss; either it doesn’t matter to them or they’ve never needed to fear loss. With that kind of privilege, the Lottery’s loss of money, status, and material wealth feels like a game. Most people fall outside that category. Someone with only a thousand dollars to their name would strongly prefer not to chance losing it, because being broke has real consequences. Similarly, parents are risk-averse because losing their wealth likely has physical consequences for their lifestyle. There’s a reason young men are the demographic most prone to gambling.
The ability to be agentic involves a similar risk and thus comes from a similar position of privilege. Few people have the privilege to found a startup; it requires having enough wealth and support to be able to survive if the project fails. Those who can take on risk absolutely should if it has positive value! But when these people grow desensitized to risk, since it so rarely matters to them, the wrong things start to feel like a game. In 1994, a group of traders and economists tried their hand at the game of the market, founding a hedge fund called Long-Term Capital Management that made money off international arbitrages. After three years of exhilarating 40% year-over-year gains, the fund was wiped out in months by back-to-back financial crises. The Federal Reserve was forced to coordinate a $4 billion bailout to avoid more widespread economic damage. Of course, the people involved were never punished; two were awarded the Nobel Prize in Economics, and the rest simply took positions in new hedge funds.
11.
Recall Ishtar, the Babylonian goddess of love and war. I consider her the goddess of the Lottery and of the cloud mindset. She has so much power and is so insulated from the world that she can treat it as a game to play to her whims. She can ask men to love her, or go to war for her, and can send animals to kill them when she’s done. The gods can never lose their power, so Ishtar will never feel the weight of the damage she inflicts on the world.
Now recall Tarawet, the Egyptian goddess of fertility and motherhood. Tarawet is a protector. Like a real parent, she uses her power to keep her children safe and provide a stable life for them to grow. Historically, childbirth was enormously risky, with a genuine chance that both the child and mother could die if it went wrong. People would pray to Tarawet to take away that risk because it was unacceptable for a family just trying to survive together.
In many ways, the people I describe in this post consider themselves gods. People who call themselves agentic, people who engage in startups and missions to fix the world, people who indulge in risk for their own benefit—they are often in the same position as Ishtar. The privilege of material, social, and familial wealth separates people from the consequences of their actions. They defect in the societal prisoner’s dilemma because, as in the real prisoner’s dilemma, they never interact with the suffering they cause. They will be happy as long as they lack the compassion to care about the person in the other cell.
12.
The title of this essay comes from David Mitchell’s novel Cloud Atlas. Mitchell explains the title better than I can, so I’ll leave him to it.
In the title itself, Cloud Atlas, the cloud refers to the ever-changing manifestations of the Atlas, which is the fixed human nature that is always thus and ever shall be. So the book’s theme is predacity, the way individuals prey on individuals, groups on groups, nations on nations, tribes on tribes.
This piece is written with a tinge of bitterness because it describes many of my friends and former friends. They are the friends who prey on my other friends, and they found the groups that prey on groups, the companies that prey on companies. They do so dispassionately, calculating the best way to serve themselves and their friends at the expense of the rest of the world.
I’m a fan of the cloud mindset as a tool. I believe it fails in moments because it’s used to accomplish the wrong goals. My friends and former friends try with all their might to help themselves in a manner that ignores the harm they cause to themselves and those around them. If everyone did the same, we would live in the world of Ishtars, a world where no one can be trusted because they will turn their back on you at every chance.
I don’t want that. I hope that those who choose to use the cloud mindset do so with the people around them in mind. I’d like to live in the world I was raised in—a world of people who support each other without fail, of people who choose to cooperate in the prisoner’s dilemma, of people who choose to avoid the capricious Lottery and live happier lives together as a result.
You might choose not to do each of those things. You might choose to continue naively prioritizing yourself without sacrificing anything for the people around you. I sincerely hope that’s not the case. Because I care about you, stranger, and I think you’ll be better off if you choose to care about me.
1.
This post is a resolution to a series of questions I’ve asked myself over the past six months. My answer to those questions is that there’s a dichotomy between two kinds of people:
Clouds make decisions that maximize utility, making rapid decisions and ignoring rules and norms to do so, sometimes to the detriment of those around them. Atlases tend to follow prosocial rules and norms, often benefitting society at their own expense.
I think society today, or at least the society I surround myself with, is increasingly composed of clouds. I think these people would be happier if they adopted the cloud mindset with the prosocial goals of the atlas mindset. This essay is a roundabout way of making that argument.
2.
Last December, while writing a screenplay at a café in Guatemala, I stumbled across Ishtar, the Sumerian goddess of love and fertility. I thought I knew what that meant. I read The Kane Chronicles as a kid and was very familiar with Tawaret, the corresponding Egyptian goddess of marriage and fertility. In mythology, Tawaret was a fierce protector of the home and of childbirth, acting as a mother herself. She was considered dangerous and had some traits of a predator, including lion paws and a crocodile tail, but used that strength for protection rather than harm.
I couldn’t have been more wrong. Ishtar is the goddess of fertility in the sexual sense rather than the motherly sense. She is also the goddess of war. She is capricious and vengeful, carrying out divine justice and retribution. In the Epic of Gilgamesh, she asks Gilgamesh to become her consort; when he rejects her, she sends the Bull of Heaven in an attempt to kill him. Ishtar’s traits are much closer to those of Sekhmet, a warrior goddess who was the manifestation of Ra’s divine power.
Ishtar and Tarawet are nearly polar opposites. Ishtar is passionate, vengeful, and makes decisions on a whim. She constantly strives for power and tries to kill those who oppose her. Tarawet, on the other hand, uses her power to protect the vulnerable and ensure they can flourish. If Ishtar is a goddess of death, Tarawet is a goddess of life.
3.
The screenplay I was writing in Guatemala was based on The Lottery in Babylon by Jorge Luis Borges, one of my favorite short stories. In Borges’s Babylon, a fictional Company introduces the Lottery, initially a drawing of lots that includes both positive and negative cash rewards. When those who earn negative rewards refuse to pay, the Company assigns them jail time and other creative punishments. Over time, the Lottery grows to be all-encompassing with no limits on its results. The story’s opening paragraph (translated by Andrew Hurley) captures the idea:
Borges writes that early lotteries with only cash rewards failed because they “appealed not to all a [person’s] faculties, but only to [their] hopefulness.” A lottery that encompassed all of society was sure to add variety to their life that canceled out the fear of loss. Later in the story, Borges leaves doubt that the Company exists at all. It may be that the Lottery is sustained by individual people leaving slips for others with arbitrary commands, and their downstream effects are enough to sustain the Lottery forever. It makes no difference either way. The Company is not special; its legitimacy comes from the fact that people listen to its demands. Given that, the actions of individuals are just as legitimate as those of a Company because they will be followed either way.
The screenplay I mentioned was written for Hackuba 2, a week-long hackathon I organized in Antigua, Guatemala. I spent the hackathon learning photography and attempting to put together a screenplay based on The Lottery in Babylon. While I wrote, I had inspiration from Michael, who attempted to decide payment for every meal via Chwazi—randomly assigning the bill to someone in the group. In economic terms, loss aversion and declining marginal utility mean that this reduces average utility for the group. But for the several people who participated, that reduction was more than canceled out by the fun of the draw.
This is the appeal of gambling, lotteries, and the Lottery: net-zero or net-negative trades are offset by the thrill of the risks and rewards involved. Only a certain kind of person would participate in a Chwazi or the Lottery that Borges imagines.
4.
I recently wrote a post comparing “mission mode” and “fallback mode.” People operating in mission mode focus on getting a small number of things done as well as possible. The classic example is 1Day Sooner and similar campaigns that worked to produce a COVID vaccine as fast as scientifically possible. In contrast, people operating in fallback mode attempt to never drop a ball and to make sure no important item slips through the cracks. This is the most common mode that governments operate in, since they are asked to deal with an enormous number of important issues that matter for the daily lives of their citizens.
Importantly, fallback mode can’t work all the time. Some issues take too much time and effort to solve when not focused entirely on them. For example, the Abundance movement argues that American industry and infrastructure have stagnated because they have been neglected by normal government operations, leaving overly restrictive laws on the books for too long. Similarly, the Green New Deal (despite being old news by now) intended to solve America’s contribution to climate change by putting energy entirely towards producing a clean economy.
This is why successful startup founders and scientists often act in mission mode. These people have everything to gain and very little to lose, so they can afford to take some risks to make progress in specific fields. In contrast, established companies and governments generally can’t afford to disrupt their normal functioning to accomplish a single goal.
5.
There’s a shared theme in the world of startups, Silicon Valley, and rationality. That theme is agency. Agency can be defined in all sorts of ways, but its core definition is a willingness to take unusual actions to get what you want. In other words, “you can just do things.” An agentic person might build a nuclear fusor to learn mechanical engineering, or they might lie about being Jain to get off the Stanford meal plan. Its core requirement is simply thinking out of the box, but that often involves breaking rules and norms that seem to be harmful or at least unproductive.
Everyone in Silicon Valley is looking for founders with agency. It’s rare to find someone who genuinely believes they have the power to shape the world around them and does so effectively. These people are often high-risk, high-reward, which means a bet on them could pay off very, very well. The most successful companies come from good ideas that no one before has had the vision or talent to implement.
6.
I think the four dichotomies above—the Lottery, Ishtar, mission mode, and agency—map onto the same fundamental difference. To extract what that difference is, I’m going to frame it in the terms of a recent conversation I had at Stanford in Beijing.
My friend Darrius argues that people should behave based on obligations to other people in society. Parents are obligated to raise their kids; in return, children should give back to their parents when they grow older. People give gifts to others on their birthdays and on Christmas because that’s what they’re expected to do. In the workplace, people should do what they’re told to make the organization they work in run more smoothly. Those who follow these rules are considered dependable and hardworking, and they’re the people who make high-trust societies function. Deviating from one’s obligations makes the world a worse place.
In contrast, I think people should do things because they want to do them, and each of those “obligations” can be framed as a want. As Nate Soares argues in Replacing Guilt, this allows people to make decisions that better serve themselves, and it removes the unpleasant feeling of guilt that comes when you fail to meet an obligation. Parents raise their kids and children give back to their parents not because they should but because they love and care about each other. Gifts are common because gift-givers care about the people who receive them, making the giver happy vicariously. In the workplace, people should do what they’re told if they believe it works towards the long-term mission of the organization, and they should push back if they find a better way to accomplish that mission.
Consider someone who wants to go to the gym more often, rather than sitting on the couch eating ice cream. Darrius would say that he should go to the gym, and he probably would. If he ended up on the couch instead, he would feel guilty, with some internal sensation that he’s a bad person. If I were making the same decision, I would weigh the two outcomes. If I went to the gym, I would burn some calories and be more physically fit. If I stayed home, I would enjoy my TV and ice cream, but I’d be getting much less fit, which is something I’m not happy about. That framing helps me reasonably weigh the options and pick the outcome I prefer. In some cases—say, if I’ve been every day in the last two weeks—I would think I’m better off taking a break and staying home, and I wouldn’t feel guilty about picking that better option.
I’m going to use the term atlases for people who use the obligation mindset and clouds for people who use the want mindset. Real atlases are steady, dependable, and do their job well, while clouds are unpredictable and change between moments based on their environment. The following sections explain why I think these terms explain the kinds of people mentioned above.
7.
Darrius uses the atlas mindset because he argues that, like a real atlas, people who follow their obligations are dependable. In his mandatory military service as a Singaporean citizen, he would have been appreciated for exercising daily, being prepared for every drill and activity, and following orders from his superiors. Militaries train for that exact discipline because they understand that the organization runs more smoothly that way. For the same reason, large corporations rely on people doing simple but valuable jobs, like Walmart relying on employees to check out customers and restock shelves.
In contrast, my friend Justin uses the cloud mindset because he argues that the ability to be creative and to innovate is the driving force of society. Like a cloud, people are most valuable when they are unpredictable. Society prizes people who buck conventional wisdom and end up being correct—in other words, people who do what they want when they think it’s right. Successful startup founders, for example, build companies around new ideas that are genuinely useful to society. Every good, technology, and idea we have is downstream of someone who had the brilliance to invent it.
I follow the cloud mindset because I think it’s more useful for accomplishing a given goal. If you really want a certain outcome, the best way to achieve it is to take the actions that actually achieve that outcome. The obligations of the atlas mindset might be the best way to do that, but if they’re not, the mindset adds unnecessary pressure to do something that doesn’t make sense. Consider the gym example. It might be that only working out every other day is most sustainable for your long-term health, so the cloud mindset would choose that. Someone with the atlas mindset would try to go as often as possible until they inevitably fail, ruining the entire plan. New Year’s resolutions fail for exactly this reason; they only succeed when you choose a plan that happens to be sustainable.
In some cases, obligations do add extra force to push for a decision that leads to a better outcome. In the gym example, it’s often a good idea to exercise, but it’s easy to not think through the entire chain of events and simply decide to rest. In that case, guilt is a useful motivating factor that would lead to a better outcome. Still, people who follow the cloud mindset often find effective substitutes for guilt, like paying a friend $10 every time they deviate from their routine. Someone thinking clearly about their decisions can implement a system to help choose the best decision most of the time. Guilt may accomplish the same thing, but it’s not necessarily helpful.
That said, I believe the atlas mindset is legitimately useful when people can’t be trusted to make good decisions. In corporate and military contexts, for example, there are often managers or leaders who genuinely are more informed about the direction of the organization and can make good strategic decisions. In that case, the organization will do better if its low-level employees are disciplined and follow instructions rather than trying to act on their own without high-level coordination.
Some would argue that this should happen on a societal scale, and it does to some extent. Murder and terrorism are illegal because anyone who decides to commit those crimes will seriously damage society. The fundamental purpose of the law is to restrict some freedom of action in a way that, on average, benefits society. But not all social coordination is done by governments! Religion, too, imposes obligations on its members that tend to improve society. Christianity’s obligations to love thy neighbor and give to the poor are intended to produce a more stable and equal society that people would prefer to live in. The Church wants to stop selfish behaviors that improve life for an individual but worsen society on average.
More rigorously, that’s a basic prisoner’s dilemma. In a one-shot game without coordination, people making the most informed decision—i.e., people with the cloud mindset—will act selfishly and defect. But we don’t want everyone in society to defect; we want to coordinate people so that they cooperate as much as possible. The atlas mindset is enormously useful in this situation because people feel obligated to cooperate, which improves society on average. A society of atlases who all cooperate will obviously be better off than a society of clouds who all defect.
8.
My problem with agency is that it encourages people to defect in a world of cooperation. That works on a small scale, but I wouldn’t want to live in a world of agentic people.
The prisoner’s dilemma is rarely accurate because it’s a negative-sum game, while many real-world games are positive-sum. Startup apparatuses, for example, exist because they’re positive-sum. Accelerators like YC bear the cost of starting a company with the expectation that it will on average produce more value than that cost. But people aggressively trying to build startups will often take trades that benefit them but negatively affect society. Cluely, for example, was essentially a marketing campaign for its founder, and it was apparently quite effective given I’m still talking about it. But that campaign created a widespread distaste for AI that both makes people unhappy and makes genuinely useful companies harder to grow. YC would never fund Cluely because it actively hurts the other companies in its portfolio.
Many of the most agentic people I know achieve their personal goals—status, popularity, funding—using socially destructive behaviors. Lying and code-switching are quite common because they allow someone to extract the most value from conversations and friendships that they then leave behind, analogous to a one-shot prisoner’s dilemma. One old friend of mine told a few lies to build our relationship; by the time I realized, they had begun courting the people I’d introduced them to. Analogously, Sam Bankman-Fried tried to make money for EA purposes by misusing customer funds in a way he believed was harmless. It worked out well for him, but when he was eventually caught, he both lost that money and permanently damaged the movement he had been working to help. His decision was agentic, selfish, and poorly thought out.
Hyper-agency is the cloud mindset operating to its maximum extent. These people and companies operate effectively because they take advantage of a culture of trust. The prevalence of the atlas mindset means people can be assumed to be reliable, allowing those who are not to get away with more. If everyone operated like them, they would not succeed, but legitimately trustworthy people and ventures also wouldn’t succeed because there would be so many guardrails to stop bad actors. Unfortunately, that’s the world we can expect as agency and the cloud mindset spread.
9.
Operating a government or organization in mission mode is analogous to the cloud, or agentic, mindset. Operating in fallback mode is analogous to the atlas mindset. I think that’s the clearest of these analogies. Remember that I defined the cloud mindset as focused on a single objective (mission), while the atlas mindset is dependable and has absolute rules to reduce the risk of a bad outcome. I also noted that cloud-mindset startups operate in mission mode because they’re trying very hard to build something new, while established companies avoid agency and operate in fallback mode because they want to preserve what’s working.
Mission mode is incredibly effective in many situations, and my original post advocates for using it more in government. Importantly, though, it can’t be used in every situation. Mission mode works in cases like Operation Warp Speed and the creation of Taiwan’s chip industry because both have an incredibly high upside with very little downside. In the COVID example, a rapid response saves millions of lives; if it fails, the government just burns some money, which it does all the time. In Taiwan, the success of TSMC has built the island into a high-income economy integral to world supply chains; if it failed, the country would simply remain in the status quo.
Importantly, mission mode isn’t applicable in every situation, and I think fallback mode deserves more praise. You can’t use mission mode to improve Medicare because there’s very little upside and a massive downside. If your new method works, it saves people some time; if it doesn’t, millions of people have no access to affordable healthcare. Most bureaucratic inefficiency originates from dynamics like this, where some improvement is possible but not worth the risk. Laws on the books tend to stay on the books because they work well enough, but a reimagined system might not work at all, and not working is catastrophically bad for an established government.
I think some people who consider themselves agentic apply their mission-mode thinking to too many situations. Defaulting to ignoring rules and norms ignores the reasons those norms were put in place. Sometimes, the right outcomes are hard to think about properly, so society institutes obligations that correct our thinking into producing those outcomes. Agency and cloud-based thinking cause people to ignore those important corrections.
10.
I think the kind of people who participate in the Lottery in Babylon are those for whom losing means nothing—in other words, those with the confidence to use the cloud mindset. People with the atlas mindset want reliable happiness, which they usually already have, so they don’t care enough to partake in games of chance.
When we tried to Chwazi the lunch bill in Guatemala, those who chose to participate fit a certain mold. They all had plenty of money—they were the ones who ordered the most, and didn’t seem bothered by the differences in their respective spending. They were also the extroverts who seemed to enjoy the adrenaline of the game. All were male. On the other hand, we had several people who were clearly conscious of their spending, and they didn’t think twice about declining to participate.
These are the kinds of people who would sustain the Lottery in Babylon. The reason lotteries rarely make sense is that money has declining marginal utility—ten million dollars doesn’t make you a million times as happy as ten dollars. The kind of lottery Borges pictures also suffers from loss aversion, the human tendency to avoid the risk of loss. These effects add up to an asymmetry: gains add much less happiness than the happiness lost from an equivalent loss.
The people who participate in the lottery are those for whom the adrenaline and excitement outweighs the asymmetry of loss. They tend to first be people who have very little loss asymmetry, typically because they have no reason to fear loss; either it doesn’t matter to them or they’ve never needed to fear loss. With that kind of privilege, the Lottery’s loss of money, status, and material wealth feels like a game. Most people fall outside that category. Someone with only a thousand dollars to their name would strongly prefer not to chance losing it, because being broke has real consequences. Similarly, parents are risk-averse because losing their wealth likely has physical consequences for their lifestyle. There’s a reason young men are the demographic most prone to gambling.
The ability to be agentic involves a similar risk and thus comes from a similar position of privilege. Few people have the privilege to found a startup; it requires having enough wealth and support to be able to survive if the project fails. Those who can take on risk absolutely should if it has positive value! But when these people grow desensitized to risk, since it so rarely matters to them, the wrong things start to feel like a game. In 1994, a group of traders and economists tried their hand at the game of the market, founding a hedge fund called Long-Term Capital Management that made money off international arbitrages. After three years of exhilarating 40% year-over-year gains, the fund was wiped out in months by back-to-back financial crises. The Federal Reserve was forced to coordinate a $4 billion bailout to avoid more widespread economic damage. Of course, the people involved were never punished; two were awarded the Nobel Prize in Economics, and the rest simply took positions in new hedge funds.
11.
Recall Ishtar, the Babylonian goddess of love and war. I consider her the goddess of the Lottery and of the cloud mindset. She has so much power and is so insulated from the world that she can treat it as a game to play to her whims. She can ask men to love her, or go to war for her, and can send animals to kill them when she’s done. The gods can never lose their power, so Ishtar will never feel the weight of the damage she inflicts on the world.
Now recall Tarawet, the Egyptian goddess of fertility and motherhood. Tarawet is a protector. Like a real parent, she uses her power to keep her children safe and provide a stable life for them to grow. Historically, childbirth was enormously risky, with a genuine chance that both the child and mother could die if it went wrong. People would pray to Tarawet to take away that risk because it was unacceptable for a family just trying to survive together.
In many ways, the people I describe in this post consider themselves gods. People who call themselves agentic, people who engage in startups and missions to fix the world, people who indulge in risk for their own benefit—they are often in the same position as Ishtar. The privilege of material, social, and familial wealth separates people from the consequences of their actions. They defect in the societal prisoner’s dilemma because, as in the real prisoner’s dilemma, they never interact with the suffering they cause. They will be happy as long as they lack the compassion to care about the person in the other cell.
12.
The title of this essay comes from David Mitchell’s novel Cloud Atlas. Mitchell explains the title better than I can, so I’ll leave him to it.
This piece is written with a tinge of bitterness because it describes many of my friends and former friends. They are the friends who prey on my other friends, and they found the groups that prey on groups, the companies that prey on companies. They do so dispassionately, calculating the best way to serve themselves and their friends at the expense of the rest of the world.
I’m a fan of the cloud mindset as a tool. I believe it fails in moments because it’s used to accomplish the wrong goals. My friends and former friends try with all their might to help themselves in a manner that ignores the harm they cause to themselves and those around them. If everyone did the same, we would live in the world of Ishtars, a world where no one can be trusted because they will turn their back on you at every chance.
I don’t want that. I hope that those who choose to use the cloud mindset do so with the people around them in mind. I’d like to live in the world I was raised in—a world of people who support each other without fail, of people who choose to cooperate in the prisoner’s dilemma, of people who choose to avoid the capricious Lottery and live happier lives together as a result.
You might choose not to do each of those things. You might choose to continue naively prioritizing yourself without sacrificing anything for the people around you. I sincerely hope that’s not the case. Because I care about you, stranger, and I think you’ll be better off if you choose to care about me.