The Chicken in the Fridge

I was eating a chicken sandwich when I learned hens vocalize softly to their embryos for the last days of incubation, and the embryos vocalize back through the shell. The chicks recognize their mother's call by day eighteen, before they have hatched, before they have opened their eyes. I learned this some months ago and have continued eating chicken anyway.

There had been a line before this, and the line was no mammals. It came from a documentary about factory farming, specifically from a pig on a loading truck on the way to slaughter, shaking. Not running, not screaming, just shaking, with very human-like eyes that could only be read as abject, again human-like, terror. Mammals nurse their young. Mammals look at me with something I can read as a face. Fish okay, chicken okay, anything I could plausibly meet at a petting zoo off the menu. But I wonder why I drew that line and why it seems broken.

What broke it was looking up what else chickens do when raising their young. A few items, in escalating order of inconvenience.

After hatching, the mother teaches the chicks which seeds are safe by demonstration and alarm calls. Hens have distinct vocalizations for aerial predators versus ground predators, the way you might say watch out differently to a person about to be hit by a car than to a person about to be insulted at a wedding. Hens have shown measurable physiological stress when their chicks are exposed to mildly distressing stimuli that the chicks themselves seem fine with, which is a textbook definition of empathy that some psychology departments would accept on a Tuesday.

Whatever I had told myself the chicken vs. mammal line was about, what had actually been sorting the animals was my own emotional proximity to them. Which is to say preference. Which is to say comfort.

The sentence I would have to say, if I were going to say the true one, is: I draw the line where I can draw it without giving up things I really like. That sentence is unavailable to me as a public position because no one says that sentence about themselves and continues to be respected. Which is interesting, because the sentence is true of almost everyone's line, including the lines drawn by people who would have phrased it more flatteringly than I just did.

The noticing did not produce a change in what I eat. The gap between the noticing and the change is the entire territory this essay is going to walk around in for the next several thousand words.¹ I am not the writer who saw something clearly and was transformed.² I am the writer who saw something clearly and, finding the situation uncomfortable, went looking for company.

A bit

A Louis CK bit, from his 2010 special Hilarious:

I could trade my Infiniti for like a really good car, like a nice Ford Focus, and I'd get back like twenty thousand dollars, and I could save hundreds of people from dying of starvation with that money. Every day, I don't do that. Every day, I make them die with my car.

CK puts the car on a moral scale opposite the lives the difference would save, and picks the car. He indicts himself rather than the audience, and the audience laughs because they recognize themselves. The recognition is the whole product. If anyone in the audience traded their car the next morning, the bit would have been less funny.

The bit names a particular gap. On one side: what we would admit, if pressed, about the moral cost of our small comforts. On the other: what we are actually willing to give up to act on that admission. The gap is uncomfortable, and the comfortable way to handle it is to fill it with stories to tell ourselves. The stories take roughly the following forms;

Some of these stories are true. All of them are convenient. Let's call this the Infiniti package: the bundle of comforts you keep purchasing past the point where you can defend the cost, attached to whatever rationalization makes the keeping survivable. CK names a single one. Most of us are running several at once, and the chicken is one of mine.

The geography

I had been picturing where people draw their lines as a pyramid. The post-sharers across the wide base. The people who live in voluntary poverty and house refugees at the apex. The picture was convenient because it told me which direction was virtue, and that direction was up.

The picture is wrong. There is no up. The truer picture is two curves on a graph. The first curve is what we would admit matters, if pressed. The second is what we would actually give up to act on the admission. They start together and they peel apart. Where they peel apart is where the moral question lives.

Some recycle correctly. Some buy the slightly more expensive eggs and tip twenty when fifteen would do. Some skip the steak and have a small interior speech. Some donate one or two percent of their income, often to whichever charity their college roommate posted about. Some commit ten percent of their income, for life, to the most efficient charities they can identify. Some quit their jobs to do public-interest work for a quarter of the salary. Some live in voluntary poverty and put refugees in the spare room.

None of these positions is morally above any of the others.³ What is morally interesting, if anything is, is whether the person at any given position is candid about what's principle in their position and what's just the place they happened to settle.

That kind of candor is difficult. It requires looking at the gap between what someone's position implies and what they actually do, and most of us prefer not to.

The fog machine

The first thing we do with the knowledge of the gap is hide it behind the word complicated.

Some of this is fair. The world is genuinely entangled to a degree no individual can think about all the time. The cobalt in the phone I am typing this on may have been mined by a child in the Democratic Republic of the Congo, and verifying the supply chain from where I am sitting is approximately as feasible as personally inspecting the ocean for plastic. The avocado on the toast came from Michoacán, where the avocado trade and the local cartels have grown into each other. The bank that holds my checking account lends to companies doing things I would object to if I knew about them. If I tried to do an actual ethical audit of every product currently visible from where I am sitting, I would be approximately three percent through the room by tomorrow night and would not have eaten.⁴

Here is the move that is harder to look at. Complicated is also one of the most useful words in the English language if you don't want to do anything. It is the cousin of the Louis CK's Infiniti. Instead of I want the nicer car more than I want that kid alive, the sentence becomes the world is too complex to know what to do, so I will keep doing what I am doing. And complexity is, in many cases, manufactured. Lobbying is byzantine on purpose. Terms of service are unreadable on purpose. Environmental, Social, and Governance (ESG) ratings are gameable on purpose. The fog is real. The fog is also being pumped out of a machine that someone is paying to operate. A confused consumer is a compliant one.

This means a candid position has to hold three things at once.

• The world is genuinely too complex to fully audit.
• Most of us hide behind more complexity than is actually there.
• Some of the additional fog is being deliberately produced by people whose business models depend on our paralysis.

The interaction of those three is the moral weather of late modernity, and we do not get out of the weather by complaining about the weather.

What was supposed to absorb this

The thing that was supposed to absorb most of this for us is the institutional layer, the standing apparatus of agencies and rules we have collectively delegated to handle vigilance on our behalf, and this is one of the unsung achievements of the modern world, generally underappreciated by the people benefiting from it most.

For example, we do not have to read every ingredient list to confirm there is no lead, because someone, somewhere, in an office we have never visited, with a budget we have never seen, has been delegated the work of caring about that on our behalf. The FDA, the SEC, the building codes, the medical licensing boards, the labor laws. None of them are charismatic. None of them attract a fan base. We notice them in the news only when something has gone wrong, which is also the only time most of us learn the name of the agency in question. These institutions are compressed trust. They convert what would otherwise be a lifetime of moral and scientific vigilance for every individual into a standing apparatus that handles the vigilance at scale, freeing each of us to live what we then call a life.

When the institutions work, we don't notice them. When they fail, we discover how much of our daily peace of mind they were quietly underwriting.

The clearest example I know of an institution working is Frances Kelsey, an FDA medical officer who in 1960 declined to approve thalidomide for the U.S. market. Her counterparts in Europe approved it. Roughly ten thousand European children were born with severe limb deformities; about seventeen American children were affected, mostly from samples brought in through other channels. Kelsey was being slow on purpose, against considerable pressure from the drug's sponsor, who wrote to her superiors complaining about her caution. Most of us do not know her name. That is the system working as designed.

The catch is that institutions fail in three distinguishable ways, and only one of them is anyone's fault.

1. Intractability. They cannot monitor every transaction, just like an individual can't. They use proxies. They sample. They set thresholds and hope the thresholds track the thing. This is them drawing a no-mammals line for the same reason I drew one, which is that the alternative is incomprehensibly large and they are finite.
2. Slowness. Institutions are built for stability, which is a virtue when the world is stable and a vice when the world moves faster than the institutional clock. By the time the FDA writes a rule for a new compound, three more compounds have entered the market. By the time Congress passes a law about a technology, the technology has rerouted around it.⁵ The slowness reflects the same conservatism that made the institution trustworthy in the first place, now applied to a moving world that punishes it.
3. Capture, qualitatively different from slowness. The lawyers writing the regulations used to work for the firms the regulations are about and will work for them again next year. Capture is an institution being quietly redesigned to look the other way, on someone else's behalf, by people who took the job offer.

A concrete example, because the abstraction has been doing too much work. The FAA's certification of the Boeing 737 MAX. The agency had delegated more than ninety percent of the safety review to Boeing itself. 346 people died in the two crashes that followed.⁶

This is what capture looks like in the wild. The FAA officials involved are not movie villains. By all available accounts, they were people inside a structure designed to defer to the manufacturer's expertise, in a system where the manufacturer's expertise was paying their next salary. The 346 dead never appeared in the certification documents. They existed, if anywhere, in failure-mode probability tables that, if anyone was reading them, were being read at the company.

So when we say the institutions failed, this is the kind of thing we mean. And when the institutions fail, the cognitive load doesn't disappear. It re-lands on the people with the least leverage to fix it, who are then asked to fix it through their personal consumer choices. The people with the actual leverage, the ones who could rewrite a label, change a fiduciary duty, hire a lobbyist with different instructions, they operate inside frames that absolve them.

That asymmetry is the morally galling part. The whole arrangement was built so that the decisions and the consequences sit in different rooms, and the people in the consequences room are then handed a guide to ethical shopping. Which is like telling the people in the rubble to write better building codes.⁷

The moral question doesn't disappear when a company does something terrible. Companies cannot be moral in the rich human sense; they do not experience guilt or have changes of heart. But the systems they operate inside are choices, made by people.⁸ The question relocates to the people who built the system inside which the terrible thing became the rational outcome. They can be moral. They are also harder to identify, which is probably not a coincidence.

The most serious recent attempt

If institutions are the collective outsourcing of moral attention, what happens when someone tries to opt out of the outsourcing and handle the attention personally? A few years ago I would have said the most serious recent attempt at exactly that, applied to the chicken question or its analogues, was Effective Altruism (EA). I would probably still say it, although the saying has gotten harder.

EA is what happens when someone takes Singer's drowning child argument (the same argument as Louis CK's Infinity and child starvation bit)⁹ fully and tries to build framework around acting on it. The result is a movement whose members are unusually willing to do moral arithmetic in places where moral arithmetic has historically not been welcome, including the question of how much one should care about insects, and whether donating a kidney makes sense as a return on investment. It produced rigorous evaluations of which charities save the most lives per dollar (turns out: anti-malaria bednets, deworming, direct cash transfers, vitamin A supplementation are the most important; none of which are the charities that get galas thrown for them).

EA produced a pledge structure, Giving What We Can, that asked people to commit to ten percent of income for life. It produced a career-advising organization, 80,000 Hours,¹⁰ explicitly designed to nudge talented people toward jobs with high moral leverage rather than high salaries. By any reasonable accounting, the movement has saved tens of thousands of lives. That is not nothing.

The recent embarrassments are also real. The largest funder of the movement, Sam Bankman-Fried, turned out to be running the largest fraud in a generation, used EA logic explicitly as cover (I need to make as much money as possible so I can give it away), and was friendly with EA leadership in ways that look, in hindsight, like a failure to apply ordinary skepticism when there is enough money.

A wing of EA pivoted into longtermism, the argument that we should weight the welfare of vast hypothetical future populations of peoples heavily because there will be so many of them, and the conclusions sometimes got strange in ways that were suspiciously convenient for whatever the wealthy donors already wanted to fund.¹¹ The demandingness problem (if donating 10% of your income, why not 50%?) was waved off with appeals to sustainability that, on inspection, look like the same Infiniti dressed in better clothes.

What this suggests is that even maximum effort at the question doesn't escape the question. It relocates the line to a higher altitude, where the same dynamics resume. The lines are drawn in better-justified places. They are still drawn, and the rationalizations for where they are drawn look similar to the ones non-EAs use. This isn't a debunking of EA, although the implementation has obviously had problems. EA tried as hard as anyone, and what it ran into was human nature. The same patterns that show up in someone standing in front of their fridge show up at higher altitudes, just better dressed. Which is information.¹²

One of the things that information turns out to include is that ethical seriousness does not produce convergence. It produces, if anything, more articulate disagreement. The EAs disagree with each other about longtermism, about how much money is enough, about how much animal lives should count in the calculation. The chicken people disagree with the no-chicken people, and within each camp the disagreements run deeper than the cross-camp ones do. If you press on any moral question for long enough, what you find is not a hidden answer everyone secretly knows but a thicket of people who have all thought about it carefully and arrived at different places. Which raises a different question, which is what they say to each other when they meet.

Pound the table

Picture the chicken question argued out loud, between a person who has stopped eating chicken and a person who hasn't. It starts as a comparison of facts: parental care, factory farms, climate impact, whatever. It proceeds through a comparison of values: sentience, locality, consequence. Then somewhere in the third loop it becomes a comparison of the moral characters of the two people in the room. By the end, both are angrier than the disagreement seems to warrant. Neither has moved.

There is a piece of old courtroom wisdom for this, sometimes attributed to Carl Sandburg, more often anonymous: if the facts are on your side, pound the facts. If the law is on your side, pound the law. If neither is on your side, pound the table. It's a description of how lawyers escalate when an argument isn't going their way, and it's also one of the more useful three-line summaries of how disagreement works.

The proverb maps onto the three kinds of moral disagreement that exist, which are not interchangeable, and which most people fail to distinguish, including me, most of the time.

• Empirical. You think the policy reduces crime, I think it doesn't. We're disagreeing about a fact. The right move is to pound the facts, which is to say, look at evidence, and the evidence is in principle capable of resolving the disagreement, even if it doesn't always in practice. Empirical disagreement is the easiest to handle and the rarest variety in the disagreements that actually wreck relationships.

• Value. You think personal liberty matters more than collective welfare, I think the reverse. We're not disagreeing about a fact. We're weighting things differently. The move here is to pound the law, in the broader sense: to argue inside the framework you both share, about what that framework demands in this case. There's no evidence that resolves a value disagreement, but there's argument, because the framework is shared.

• Frame. You and I aren't disagreeing about a fact or a value. We're disagreeing about what the question even is. You think the conversation is about fairness; I think it's about freedom. You think it's about consequences; I think it's about character. We are looking at the same situation through different conceptual apparatuses, and the apparatuses are not the same thing translated into different words. They are actually structuring the perceived situation differently.

The third kind produces the experience of I genuinely cannot understand how this person can hold that view, because at the level of frame, the other person is not in the same conversation you think you're in.

Frame disagreements are where you pound the table, because the other two moves are unavailable to you. The facts aren't on your side because the other person is reading the same facts differently. The law isn't on your side because the framework isn't shared. So you pound the table, which is to say you raise your voice, you appeal to feeling, you walk out, you legislate around the other side, you stop returning calls. Most of what looks like irrationality in modern political life is people accurately registering that they are in a frame disagreement and that the other moves are not available, and pounding the table accordingly. This is one of the more useful things to keep in mind when watching people argue, because it tells you that the table-pounder is not necessarily failing to argue. They may have correctly identified that argument has run out.

Most of the worst disagreements in modern life are frame disagreements being mistaken for empirical or value disagreements. People marshal evidence and deeper values, and the evidence and values keep failing to land, because the actual disagreement is one level deeper than the marshaling can reach. The same thing is happening, in miniature, inside the chicken question. The person who agonizes about their chicken and the person who finds the agonizing absurd aren't really disagreeing about chickens. They're operating in different frames about what ethics is for. One frame says ethics is about minimizing harm to sentient creatures wherever they are. The other says ethics is about loyalty to the people and place you actually live with. Both frames are coherent. Both produce real moral behavior. The frames don't translate cleanly into each other, which is why the conversations between them tend to fail.

Once you can see frames, you can also see what zealotry actually is, which is something narrower and weirder than strong belief. A non-zealot has beliefs and also has a self that is larger than the beliefs. A zealot has fused with the belief. Defending the belief and defending the self are the same act. The chicken-question version is the militant vegan who can no longer be in the room with meat-eaters, or the militant carnivore who treats vegetarianism as a personal accusation. Both have stopped having a position and started being one.

Once someone has stopped having a position and started being one, every disagreement starts collapsing toward the same conclusion. A frame disagreement gets treated as a value disagreement: we share the framework, you're just weighting things wrong. When that fails, it gets treated as empirical: we share the values, you've got the facts wrong. When that fails too, the only explanation left is bad faith: you must be lying. By the end, the only available explanation for someone disagreeing with you is that they are evil. This is how a society moves from arguing to fighting, and it works the same way no matter which side is doing the reframing.

The opposite move is subtler than expanding the frame, which may not be possible in any deep sense. It is learning to recognize the fact of the frame, mine and other people's, as a feature of how minds work rather than as a moral position. Recognition is not agreement. It is noticing that the person across from me isn't failing to see what I see. They are seeing something else, and the something else is real to them in the way my seeing is real to me.

I do not know which side of that line I am on, with respect to the chicken or to anything else. The recognition I have just spent four paragraphs describing is the practice that holds the alternative open, and I have not been particularly good at it.

What is available

The map of positions is real. People really do draw their lines in different places, and the places really aren't ranked. We all own an Infiniti in one way or another. The institutions cannot save us, and we cannot replace them with personal vigilance, because there isn't enough of us. The complexity is genuine and is also being weaponized against us. The most serious attempts to act on all of this end up relocating the problem rather than solving it. The chicken question, in any version I have been able to construct, does not close.

Some company, briefly, in case it's useful: a few people who walked further into the question than I have.

Bernard Williams gave Singer's argument the cleanest philosophical body-blow it ever took. He did not argue that the math was wrong. He argued that the math, applied honestly, would erase the person doing it.

A man's wife and a stranger are drowning. He can save one. He saves his wife. Williams asks what we should make of a husband who, in that moment, pauses to construct a justification. My wife is closer to me, impartial morality permits some degree of bias toward those with whom one stands in special relations, therefore... The husband who reasoned his way to his wife, even arriving at the right answer, has had what Williams called one thought too many. The first thought, that's my wife, was supposed to be the only thought. The second one, even agreeing, was already the wrong shape of thought to be having.

A self trained to cross-check every human instinct against the full entanglement, every supply chain traced, every comfort audited, every distant harm tallied, is something else: a self quietly emptied of the loves the morality was supposed to be in service of. Williams' point applies far beyond drowning wives. It applies to the chicken question, to the bank account, to the avocado, to every ordinary preference the impartial framework would have us second-guess. A marriage, a hometown, a Sunday morning at the kitchen table, all of these are forms of caring more about some people and places than others. An ethics that demands the same accounting from each of them ends up dissolving the very things it was supposed to be defending.

I think Williams is right, and I think he is the reason the chicken question can't close. The cleanness I would need, to know whether eating chicken is okay, would cost me more than the chicken does.

The Quakers, who are the most underused example in contemporary discussions of ethics, partly because they don't make themselves easy to use. Three hundred and fifty years of slow collective discernment in silence. No charismatic founder, no hierarchy, no revealed text that resolves disputes. They sit together and wait until something becomes clear, and they don't act until it does. By this method, with no apparent strategic advantage, they ran the Underground Railroad and went on doing things at that scale for several centuries without becoming famous for it.¹³

The practice doesn't pretend to resolve frame disagreement through argument. It assumes argument has limits, and that something else, slower and more communal, can sometimes find what argument can't. A Quaker meeting working through whether to do something, war tax resistance, sheltering refugees, divestment, can take years. People sit. People speak when they feel called to. People notice when the room feels different. Sometimes nothing emerges and the question stays open. Sometimes a sense of the meeting forms, and the meeting acts. The chicken question, handed to a meeting, would not produce a position paper. It would produce, possibly, after enough time, a way of behaving that some of the people in the room would later recognize as having been clear all along.

The fridge

I started this essay trying to figure out where the line goes. I have not figured it out.¹⁴ The chicken question is still open. I am still standing in front of fridges with the door open longer than I need to. The line is drawn approximately where it was when I started, with the principle I told myself I was using replaced by no principle at all and a slightly more candid discomfort sitting in the principle's old chair.

There is one more move available, which is to notice that this essay is, on the geography it just sketched, not the most flattering act. Producing a thoughtful piece about why I cannot change my behavior is a substitute for changing my behavior. The essay does not redeem the fridge. It is, if I am being precise, what I am doing instead of not eating chicken.

The chicken is in there now. I will eat it tomorrow.

Footnotes

1. It is worth flagging early that humans have been eating things for roughly three hundred thousand years and almost none of those humans have written essays about it. The vast majority of people who have ever lived ate what was available, what they could afford, what their parents fed them, what their religion permitted, and what was placed in front of them by people they loved. To agonize about the ethics of one's grocery list is a luxury problem belonging to a particular sliver of human history, mostly the part with refrigerators and surplus calories and enough leisure time to interrogate one's preferences in writing. I am aware of this. The fact that the question is a luxury does not mean the question is not real, but it does mean I should not claim too much for the seriousness of the asking. Any honest version of this essay starts by admitting that most of the humans who have ever lived would, if shown the seven thousand words that follow, recognize them mainly as a description of how comfortable I am. ↩

2. A diet change also comes with a relabeling problem. At a certain level of dietary restriction, the restriction stops being something you do and becomes something you are. You are introduced as the vegetarian friend, due to dietary considerations. Your other qualities (sense of humor, conversational ability, capacity for loyalty, the things you presumably wanted to be known for) get filtered through the restriction whether you like it or not. People begin to forward you articles. Disagreements with you about unrelated topics become opportunities for the other side to pattern-match you to the more annoying members of your dietary tribe. Nobody likes to be relabeled. I still don't like it, but accept it for what it is, like gun ownership or wearing hemp. ↩

3. Every position on the scale also has a variant in which the person has not actually moved on the scale but has read about it, and is now telling you about it. This variant is recognizable by the direction their attention runs. They look downward, at people they are pretending not to notice are them five years ago. They do not look up. The asshole is not located at any particular level. The asshole is wherever the talking is happening. There is always someone more committed than them, doing it more quietly, and that person is the one they have decided to call insufferable, which is one of the more reliable signals available that you have located the asshole and that it is not the person they were pointing at. The author has noticed, with some interest, that he has just written this footnote, and this whole post. ↩

4. The actual logical extension of trying to do a full ethical audit of one's possessions is an interesting little wormhole. You'd need to research not only every product but every component of every product, and not only the components but the labor conditions and supply chains of the components, and not only those but the labor conditions of the people doing the auditing of the labor conditions, and not only those but whether the auditors are themselves captured by the industries they audit, at which point you have constructed a recursive moral debugging task that exceeds the lifespan of any single human. Most people abandon the project somewhere around the second nested level of the chocolate chip in a granola bar. I abandoned earlier than that. I had things to do. ↩

5. If they ever do. ↩

6. Beginning in 2005, the FAA expanded a program called Organization Designation Authorization, which allowed aircraft manufacturers to use their own employees to certify the safety of their aircraft on the agency's behalf. By the time Boeing developed the 737 MAX in the mid-2010s, more than ninety percent of the plane's safety certification was being conducted by Boeing employees rather than FAA employees. The MCAS system, which automatically pushed the plane's nose down to compensate for the new engines' position on the airframe, was certified through this process without requiring simulator-based pilot training, which would have raised costs for airlines and slowed deliveries. In October 2018, Lion Air Flight 610 crashed into the Java Sea, killing all 189 people aboard. In March 2019, Ethiopian Airlines Flight 302 crashed shortly after takeoff from Addis Ababa, killing all 157 people aboard. Both crashes were caused by MCAS triggering on faulty sensor data. The FAA officials who had overseen the certification had spent careers moving back and forth between Boeing and the agency. ↩

7. I am aware that the celebration of compressed trust in the preceding paragraphs reads very differently depending on whose town the factory closed in. There is a reader who has been to too many funerals for people under fifty, whose neighbors got prescriptions from doctors trusting an FDA approval that turned out to be wrong. That reader watches the agonizing about chicken with reasonable incredulity. The institutions I have been praising as one of modernity's underrated achievements have, in their direct experience, been systematically wrong about things that mattered. Both views are correct at the same time, which is the part the essay is having trouble holding. The institutional layer is a real achievement and is currently being hollowed out, and the hollowing is being felt unequally, and the people feeling it most are not, on the whole, the people writing the essays. Complicated, in their experience, has only ever been a one-way street, somehow always concluding with and therefore we will continue doing what was already convenient for the people already in charge. They are not wrong about that either. ↩

8. The Powell memo, properly Confidential Memorandum: Attack on American Free Enterprise System, August 23, 1971. Not particularly secret now. Worth reading once. The relevant fact about it, for our purposes, is not whether you find the politics agreeable but that the memo makes vivid that the current arrangement was a project, with goals, organized by specific people, executed over decades. The settlement is not weather. It was built. Things that are built can be unbuilt, by people willing to do equivalent work in another direction, over equivalent time. This is a depressing piece of information and also, on the other hand, the only kind of information that supports the possibility of any change at all. ↩

9. Peter Singer's argument, from his 1972 essay Famine, Affluence, and Morality: if you would wade into a shallow pond to save a drowning child even though it would ruin your good clothes, you have the same obligation to prevent a distant stranger from dying of starvation or preventable disease, since the moral structure is the same and only the geography differs. Most people accept the pond case immediately and find the extension uncomfortable. EA is what happens when someone refuses to let the discomfort end the argument. ↩

10. 80,000 Hours was founded in 2011 by Will MacAskill and Benjamin Todd. The name refers to the approximate number of working hours in a typical career, which is the unit the organization asks you to think strategically about. In its early years the recommendations were heavy on global health and direct-impact charity work. By the mid-2020s they had migrated substantially toward jobs reducing risks from artificial intelligence, which is the longtermist pivot the next paragraph is about. The largest funder by a wide margin has been Dustin Moskovitz (Facebook co-founder) and Cari Tuna's Coefficient Giving, which has put more than thirty million dollars into the organization through 2025. ↩

11. One strong form of the longtermist argument holds that humanity is approaching a transition, usually identified with the development of powerful AI, after which the values and structures embedded into the dominant systems will become very difficult to change. If that is right, whoever shapes that transition shapes the values that lock in for an extraordinarily long time, and therefore the highest-leverage moral act is to influence the shaping. This is the version of the argument that connects most directly to the current AI-safety focus. When the argument is taken seriously, the recommended actions tend to be funding AI-alignment research at well-credentialed Western institutes rather than funding bednet distribution in places where children are currently dying. The resulting funding patterns happened to align almost perfectly with what Silicon Valley billionaires were already inclined to fund. The two leading frontier labs reflect the lineage directly. OpenAI was founded in 2015 as a nonprofit framed around AI safety concerns. Anthropic was founded in 2021 by former OpenAI researchers and took early funding from a who's-who of EA donors (Dustin Moskovitz, Jaan Tallinn, Sam Bankman-Fried, who led the Series B before his fraud unraveled). Anthropic's president, Daniela Amodei, is married to Holden Karnofsky, who co-founded Open Philanthropy. ↩

12. The cosmopolitan ethics that produces EA, the math that says you should weigh the distant child equally with the local neighbor, has a cost the framework prefers not to itemize. The cost is that the local neighbor lives next door to people who have decided, on principled grounds, that her loneliness is mathematically less important than a child she will never meet. The framework can defend that priority on consequentialist grounds and the defense even works on its own terms. People can still tell when their suffering has been ranked. On inspection, the populist response is the math grasped exactly and rejected as the wrong math. The alternative on offer is itself an ethics, organized around the place you actually live and the people you actually know, that goes to the funeral of the neighbor who was rude to you, that tips the waitress more than the meal warranted because you could see her shoes, that raises kids in a town instead of dragging them across the country for a better job. Both frameworks produce real moral behavior. Each has costs the other side notices first. I do not have a synthesis. Pretending to one would itself be an Infiniti package. ↩

13. Their record also includes being among the first organized opponents of slavery in the West, founding major institutions for the mentally ill at a time when those people were being chained in basements, and earning the 1947 Nobel Peace Prize for humanitarian work in the Second World War and its aftermath. There are roughly 350,000 Quakers in the world today, which is to say the achievement-per-capita ratio of the tradition is one of the stranger facts about the modern moral landscape. ↩

14. What I do not know about chickens, after researching this for several months, is approximately the following: how a chicken experiences time, whether the cognitive science I have been citing is replicable, whether the scientists who produced the science eat chicken themselves, whether the suffering of one chicken is the same kind of thing as the suffering of the roughly nine billion killed annually for food in the United States or the roughly two thousand I have personally eaten in the last decade, and whether I would behave any differently if any of the answers came back unambiguous. Of the three numbers in that question, the two thousand is the one I find hardest to make abstract, and the hardest to make abstract is the one closest to me, which was Williams' point. The list of things I have not figured out is approximately the list of things I would have needed to figure out. ↩