What I get out of effective altruism

This is the first post in a series on the core things I get out of the cluster of ideas that effective altruism (EA) offers as a package. EA has been thrust into the national spotlight, and I’ve been publicly associated with it for half a decade, after lurking on it from afar for the decade before. This is what I’d want to say to a smart friend who wanted to know the high-level deal with EA, beyond the sales pitch but without any jargon. This first post is a very high-level explanation of the big story EA tells about the world, and why I think it’s useful, but only as one mental tool among many. Among other things I want to explain to people what the underlying connection is between bed nets, animal welfare, and AI.

The effective altruist story of the world

The core story effective altruism tells about the world, which was already there in the drowning child thought experiment that a lot of the movement grew from, is that the ways we go about doing good, whether through our donations or our careers, are often strongly determined by what we believe will get us social status or otherwise help us get ahead. There’s good reason to think our higher-level thinking itself evolved to help us win social games, so when we set out to find “how to do good,” a lot of what we’re actually doing is picking up on subtle social vibes around different causes, trying to suss out what the “good” (often meaning high-status) people work on and believe in.

This is not inherently bad or useless. Status systems communicate a lot of useful information. Society often gives status to things because they really are useful, just not in ways that can be clearly articulated to every person affected by the push and pull of what we give status to. So social status behaves a lot like a market, with the behavior of billions of people together determining how much status different things get, without any central planner making the final decision. A lot of our inclination to do good is really an inclination to respond to these market incentives. And that’s good! It keeps society running, helps us live together, and often helps us respond to subtle human needs that can’t be clearly articulated through direct argument. When I say status is like a market, I mean both that markets are phenomenally efficient and that status isn’t always bad. The status we give our friends is not meaningless evolutionary noise, just like the money a local baker receives for a cupcake isn’t the full story of what happened in the act of making it.

But like any market, this one has deep failures and missed opportunities. Our sense of how to do good was built for the small bands of early humans we evolved in, who couldn’t meaningfully affect the world far away from them. We now live in a hyper-globalized world of over 8 billion people, and can affect almost any of them by donating to the right places. We also have unbelievable sci-fi control over the world, to the point that we could bathe the entire planet in nuclear fire if we wanted to. Among other things, we now farm hundreds of billions of animals every year. Despite our strong evolved intuition that the world will mostly stay the way it was when we grew up, the last 200 years have seen constant, previously unimaginable technological change, and the near future might bring new technologies that upend our safety at least as much as nuclear weapons did a mere 80 years ago. In this environment, our evolved intuitions around goodness and status are much blunter instruments than they used to be for figuring out where we could actually do the most good. To rely on them completely because they feel right and normal and good, and to never be willing to override what immediately feels virtuous or what your peers (or society more broadly) are encouraging you to do, seems at the very least like an unserious way of living, a way of not fully thinking through the situation we’re in. It seems like a form of playing pretend. EA is one of many ways I try to think more seriously about ethics and complicate my immediate sense of what’s good.

A metaphor I like for two types of thinking is serial vs. parallel processing. In serial processing, you follow logic from premises to conclusions, step by step. In parallel processing, you “vibe it out.” You pick up on subtle, inarticulable aspects of the world and respond to them. When you do a math or physics problem, you’re thinking serially. When you pick up on a social cue or sense you’re in danger, you’re thinking in parallel. Most of our thinking is actually an incredibly complex mix of both. Importantly, we didn’t really evolve to do consistent, thorough serial thinking without having it easily thrown off by the vibes of parallel processing. Thinking step by step is much more cognitively demanding. As Dennett says in Consciousness Explained:

Conscious human minds are more-or-less serial virtual machines implemented—inefficiently—on the parallel hardware that evolution has provided for us.

In the evolutionary environment, serial thinking about morality probably wasn’t very useful. Social cohesion was a matter of life and death, and relied on members of the group not randomly getting some idea in their heads that seemed logical and right but didn’t fit what the group needed from them. Our desire to look virtuous to our peers was heavily selected for, regardless of its effects on the wider world, which we had little ability to affect anyway. How would evolution even select for what “had good effects on the world” far beyond us?

Of course, our moral intuitions have also evolved alongside society. Modern people’s moral intuitions are wildly different from those of people thousands of years ago, and it’s not a coincidence that they now line up with a lot of very complex ways of behaving that happen to make a complex society function remarkably well. Our general respect for and good treatment of strangers, and the corresponding ability to trade with and benefit from people very different from us who live far away and well outside our immediate networks, is a relatively new and miraculous development in the history of humanity. But society has changed, and keeps changing, so much that even our newer moral intuitions often can’t tell apart the very different levels of impact we can have on the world. They’re still too blunt for our hyper-globalized, hyper-technological situation.

So to circle back to effective altruism, this is the core story it tells about the world:

Our thinking about ethics and about what we should do is often heavily structured around what will give us social status. This isn’t necessarily bad, because social status is like a complex market, which can both communicate a lot of useful information and incentivize a lot of pro-social behavior. But this market is likely to have gigantic failures, places where it drastically misjudges where we could do the most good, because our moral intuitions ultimately come from a very different situation, where we lived in small, close-knit bands in a world we mostly couldn’t affect, rather than in a world of 8 billion people where each of us can affect lives on the other side of the planet with a simple donation. While our moral intuitions and status systems have evolved to keep up with our changing situation, the world has changed so fast in the last few hundred years that there are likely massive gaps between our moral sense of what we can do and what we can actually do. There are lots of other serious belief systems and ways of investigating the world that have been phenomenally successful at finding truth (including what looks like moral truth), and we should defer to them (people into EA are quite deferential to scientific and philosophical consensus where it exists). But the new world we’re in is so strange and vast, and people are so averse to thinking step by step about doing good, that other moral market failures likely persist. We should try our best to find them, compare them, and act on the best ones.

Getting more concrete

It’s rare for people to actually try to think step by step about what they’re doing overall.

A lot of our core beliefs about ourselves and the world are surprisingly surface-level. One of the most humbling experiences is sitting down with a blank Google Doc and trying to type out a detailed explanation of some belief that’s important to you and guides a lot of your life, or what you’re ultimately trying to do with your career, or your life plan in general, or why you donate to the charities you do. Things you believed you “think about a lot” might turn out to be a few slogans you repeat in your head over and over while you’re at the grocery store. Our own minds are pretty opaque to us. We often think we’re doing deep, premise-to-conclusion serial thinking about the problems we care about, when in fact we’re just basking in the warm glow of status we get from turning simple slogans over in our heads. A lot of my own “serious thinking about economics” looks more like me offhandedly thinking “Markets communicate information. Hayek! I’m so smart” and feeling great about it while I wait in line at Trader Joe’s. My most popular post was basically a long meditation on how we do this when learning about the world, and how to avoid it.

This doesn’t make sense if you think of the world as a sequence of rational, thought-out decisions by actors carefully weighing every trade-off. It does make sense if you think of human society as mostly guided by markets for social status, where our behavior and thinking are strongly pulled toward whatever makes us feel we’re doing or believing the right thing, and careful serial thinking about what we’re doing is much more demanding and unpleasant. When we do think step by step, we regularly bump up against our own failures or lack of knowledge or seriousness, all things our brains are hard-wired to avoid.

But doing it also often yields a surprising amount of useful insight.

There are many ways of thinking serially about the world (my favorites being physics and analytic philosophy) that have been remarkably successful at overriding our longstanding moral and factual beliefs and getting us to better behavior or deeper moral truth. The specific goal of EA is to scout out the most important of these that are still left to find, the ones that seem robustly good after wrestling with all the reasons we have to be uncertain, and that merely require us to think step by step about how to live coherently with what we value in this new sci-fi world.

This is a good time to come back to the drowning child argument for giving much more to global charity. Here’s the argument as Peter Singer originally wrote it:

I begin with the assumption that suffering and death from lack of food, shelter, and medical care are bad. I think most people will agree about this, although one may reach the same view by different routes. I shall not argue for this view. People can hold all sorts of eccentric positions, and perhaps from some of them it would not follow that death by starvation is in itself bad. It is difficult, perhaps impossible, to refute such positions, and so for brevity I will henceforth take this assumption as accepted. Those who disagree need read no further.My next point is this: if it is in our power to prevent something bad from happening, without thereby sacrificing anything of comparable moral importance, we ought, morally, to do it. By “without sacrificing anything of comparable moral importance” I mean without causing anything else comparably bad to happen, or doing something that is wrong in itself, or failing to promote some moral good, comparable in significance to the bad thing that we can prevent. This principle seems almost as uncontroversial as the last one. It requires us only to prevent what is bad, and to promote what is good, and it requires this of us only when we can do it without sacrificing anything that is, from the moral point of view, comparably important. I could even, as far as the application of my argument to the Bengal emergency is concerned, qualify the point so as to make it: if it is in our power to prevent something very bad from happening, without thereby sacrificing anything morally significant, we ought, morally, to do it. An application of this principle would be as follows: if I am walking past a shallow pond and see a child drowning in it, I ought to wade in and pull the child out. This will mean getting my clothes muddy, but this is insignificant, while the death of the child would presumably be a very bad thing.The uncontroversial appearance of the principle just stated is deceptive. If it were acted upon, even in its qualified form, our lives, our society, and our world would be fundamentally changed. For the principle takes, firstly, no account of proximity or distance. It makes no moral difference whether the person I can help is a neighbor’s child ten yards from me or a Bengali whose name I shall never know, ten thousand miles away. Secondly, the principle makes no distinction between cases in which I am the only person who could possibly do anything and cases in which I am just one among millions in the same position.I do not think I need to say much in defense of the refusal to take proximity and distance into account. The fact that a person is physically near to us, so that we have personal contact with him, may make it more likely that we shall assist him, but this does not show that we ought to help him rather than another who happens to be further away. If we accept any principle of impartiality, universalizability, equality, or whatever, we cannot discriminate against someone merely because he is far away from us (or we are far away from him). Admittedly, it is possible that we are in a better position to judge what needs to be done to help a person near to us than one far away, and perhaps also to provide the assistance we judge to be necessary. If this were the case, it would be a reason for helping those near to us first. This may once have been a justification for being more concerned with the poor in one’s town than with famine victims in India. Unfortunately for those who like to keep their moral responsibilities limited, instant communication and swift transportation have changed the situation. From the moral point of view, the development of the world into a “global village” has made an important, though still unrecognized, difference to our moral situation. Expert observers and supervisors, sent out by famine relief organizations or permanently stationed in famine-prone areas, can direct our aid to a refugee in Bengal almost as effectively as we could get it to someone in our own block. There would seem, therefore, to be no possible justification for discriminating on geographical grounds.From Famine, Affluence, and Morality

This thought experiment is, among other things, a way to snap you awake into the hyper-globalized world we’re in, out of the still very strong moral illusion that we live in small hunter-gatherer bands and can only affect the people right in front of us. In reality, there are enormous numbers of invisible levers floating in front of each of us that we can pull at any time, each with some effect on other people’s lives, and some with unimaginably larger effects than others. These levers go undetected by the status games around us, and at times even by the more complex and nuanced ethical systems that have arisen to keep pace with modernity, because the world has just moved too fast for most people to intuitively sense how much real impact they can have now.

One of these levers is donating to specific malaria charities, which can reliably prevent one death from malaria for roughly every $5,000 donated.

This fact, that each of us could save a real person’s life for a meaningful but still small fraction of our annual income (the median American household makes $87,000 per year), seems at first like one interesting moral claim among many, but once you think through its implications, a lot of our behavior looks kind of absurd. For example, the ongoing Russian invasion of Ukraine has been an unimaginable catastrophe, with around 600,000 people killed since the war began. I don’t want to cheapen what’s happened, and to be clear, I’m not saying “the money spent on this war should have been donated to charity instead.” Instead, consider that the budget of the school district I taught in (Fairfax County, admittedly one of the largest in the country) is about $4 billion per year. 600,000 x $5,000 = $3 billion. So politicians in Fairfax County, Virginia could get together and say “Hey all, we’re going to double the taxes you pay into the school system for exactly one year, which will be a lot, and donate all of it to anti-malaria and similarly effective charities, but by doing this our one county will prevent an entire Ukraine war’s worth of deaths in a single year, just spread around the world.” If a single American county had the opportunity to prevent a major war’s worth of deaths with one policy change for one year, that would be major news! We wouldn’t treat it as one interesting decision among many. “Oh, this one random county could’ve prevented a whole war’s worth of deaths, but decided not to. They want to focus on infrastructure!” That would be pretty shocking.

And yet we all do have this collective opportunity, and it mostly goes undiscussed. When I bring it up, people often treat it as one interesting fact among many, go on with their days, and get morally fired up about things that don’t seem nearly as consequential, and that I don’t think even they would believe are nearly as consequential if they spent some time writing out how those things compare to malaria. But people often just never think to do these comparisons at all. If they took the time to seriously consider how they use their resources, I think they would often come to different conclusions. My claim here isn’t that malaria is the one big cause every smart person needs to care about, only that it’s evidence that many people won’t even take the first step in seeking out ethical market failures unless really pushed. Malaria looks like a cause that would get a lot, lot, lot more attention if people actually sat down and thought step by step about what they’re doing with their charitable donations. It wouldn’t fade into the background as one interesting problem among many. For a lot of people, donating to fight malaria would seem much more urgent than most other things they could donate to, for the same reason it would seem drastically important for a single county to prevent a major war if it could.

There are of course problems with this thought experiment. Malaria kills around 600,000 people every year, so that’s the very maximum malaria charities could possibly prevent. And malaria charities hit diminishing returns. It gets harder to prevent each additional death once the easier cases are covered, so the cost per life would rise well before we got anywhere near that many people saved in a year. But there are a lot of other ways to spend $5,000 that save or come close to saving a life (malaria is just one of the causes GiveWell directs money to). Once malaria crosses some cost threshold, you can divert money to the next cheapest way of saving a life, and a single county in Virginia can still prevent a major war’s worth of deaths in a year.

Individuals, and large groups of individuals, could make moves like this whenever they’d like, each preventing at least one unnecessary death per year by donating to the right places, and collectively preventing major wars’ worth of them. Yet it often doesn’t get talked about at all. When people first hear the basic argument for “effective” giving, that we should direct our donations wherever they save the most lives, their reaction is often “That’s so obvious, why would you even need a special name for that?” It may seem obvious, but almost no one does it, anywhere. Walk around where you live and, as you pass people or homes, ask yourself “What are the odds this person has actually sat down to work out where their donations would go the farthest in saving lives or otherwise doing good? Have they compared the options available to them?” For the vast majority of people I’ve spoken to, the answer seems to be no, even though a few hours spent comparing the options would probably let the same amount of money change vastly more real people’s lives. Why not? Why doesn’t this happen?

There are a lot of reasons, but it’s useful to keep in mind that most people aren’t invested in thinking step by step about their donations, rigorously following premises to conclusions without first sussing out how those conclusions fit into their social sense of what will make them seem important. Thinking step by step is often arduous, we didn’t evolve to do much of it, and we often don’t even do it when planning our own careers!

If you talk to people about donating to malaria bed nets, they’ll often quickly exit the act of thinking through the problem with you, and let their parallel-processing sense that “this isn’t actually one of the cool things to care about right now” take over. They’ll come up with strange justifications you know they wouldn’t later endorse. They’ll repeat simple slogans they haven’t thought much about. Some of my more left-wing friends have told me bed nets are bad because “we need systematic change” instead, but then don’t go on to criticize similar one-off donations to people in need when they’re local.

This doesn’t always happen, to be clear. People also give thoughtful answers about why it makes sense to donate elsewhere, or not to donate at all. I’m not saying that people’s beliefs about malaria nets are a sign they’re ideologically asleep. What does strike me is that so many people, who spend so much of their lives guided by specific modern moral systems, treat the fact that we could collectively prevent major wars’ worth of deaths about the same as any other random fact about the world, and won’t follow along with thinking through what it implies about how we should behave, or will come up with on-the-spot justifications for not donating to effective causes that they probably wouldn’t endorse with a little more thought. Again, this suggests there could be massive, simple ethical market failures created by our status systems that most big ideologies today miss, and that could be found by just applying serial thinking, of writing out a Google Doc of what we actually know about the options available to us. EA is about trying to find those market failures with serial thinking, and to work out which look most promising to work on and which a lot of different moral views converge on once the facts are in front of them (or in slogan form, “using reason and evidence to do the most good we can with our time and resources”).

Another key part of seeking out ethical market failures is that they should look reliably good from a lot of different ethical perspectives. The idea that you ought to prevent someone from dying of malaria when the cost to you is a few minutes of thinking carefully about your donations seems like a very strong reason to do it whether you’re a utilitarian or a deontologist, a Christian or a Muslim. This isn’t to say there’s some objective morality written into the universe that we believe we can find, only that we want to take actions that look good under a lot of different possible moral systems, and it’s surprisingly easy to use serial thinking to find actions that look “robustly” better than others, meaning clearly good across a wide range of ethical assumptions.

So to recap: the core of EA is the idea that simply applying step-by-step thinking to the ways you do good (which, again, is rare, because we mostly didn’t evolve to do it and are often socially incentivized against it) can yield new options that look so robustly better for our values that letting them fade into the background of everything else we do looks ridiculous, and that people should just take this step and do things like simple cost-benefit analysis of the charities they donate to. When we do, we often find that the best things we could do are hidden from us, can have thousands of times the impact we care about (like saving lives) compared to what we were doing before, and look surprisingly compelling from a lot of different basic moral perspectives. So, as just some guy in DC, I can reliably prevent someone from dying of malaria every year because I’m careful about comparing where to donate. If I weren’t, it’d be extremely easy for my charitable giving to never prevent a single death.

If you want a sense of how this serial thinking looks in practice, a good place to start is the 80,000 Hours problem profiles, which guide you step by step through what they see as the most important problems in the world to work on. Reading a lot of writing like this in the 2010s, which tried to walk rational people through the full comprehensive deal with complex global problems without relying on social passwords or going along with the coolest things to work on at the time, and which was one of the few places online that actually tried to put numbers on all these different interventions, had a big influence on how I write. I credit a lot of my own blog’s success to the style I picked up from the blogs I was lurking on back then.

The main cause areas of EA (global health, animal welfare, AI safety, biosecurity) grow out of this intuition. Each is based on a plausible idea that doesn’t get much social status, so it doesn’t get considered much, and that also has the potential to be massively, robustly important:

• People in rich countries like the United States can save the lives of people in poor countries with modest donations to the right charities, which often drastically outperform local charities in lives saved per dollar.
• Animals aren’t just automatons, and may have complex forms of happiness and suffering that matter morally.
• Humanity could last a very long time, so if we go extinct or civilization collapses, this seems like a uniquely large loss.
• AI might gain most of the economically relevant abilities of the human mind sometime this century, and the consequences could be insane. (This one has gone mainstream.)

Each of these is complicated, and brilliant people disagree with them. But there are also (especially for animals) very strong social biases against even considering them in the first place, which leaves open the possibility of a massive market failure. At best, a lot of people react with “Oh wow, neat idea” and keep these in their pockets as talking points to bring up every now and then. A big goal of effective altruist thinking is to not do that, to not collect neat ideas and fold them into your current status game, but to behave as if they’re really real, as if they reflect something actually happening in the world that ought to shake you into behaving at least somewhat differently, and to act the way your future or past self would want you to if you suddenly discovered you really lived in a world like this. To then reason through what else we can know and do about the problem, and try our best to act on it under uncertainty instead of letting it become another neat conversational move.

Behaving as if the world exists beyond our status games

I worked at a summer camp when I was a teenager. Once, a camper crossing a bridge suddenly gripped the railing and started breathing heavily. A counselor asked what was wrong, and the kid turned to him and said “This is all REAL.”

“What? Yeah.”

“No, really, like us being here right now, and the bridge, and the world. It’s all REAL.”

I think this is a relatively common experience. A lot of the time, the world can feel somewhat fake. We go through the motions of our lives and a lot of what we see starts to feel like a dream, or like some kind of rehearsal while we wait for the life where things are real to start. It can take effort to remember that the faces we pass every day have lives just as complex as ours, never mind that the rest of the world is actually happening too, with billions of people in it. Sometimes the sense that our lives are real, that this is all really happening, suddenly floods us, or comes to us gradually as we live more in line with what we actually think is valuable.

For a lot of people moved by the basic EA story of the world, I think a key shared intuition is that thinking step by step about how to do good is a way of letting the world beyond our status games actually reach us, of taking it seriously as really existing out there beyond our little games, and of trying our best to work out how to live most coherently with our values in a world far bigger than the part of it we can see. To not laugh it off because it doesn’t perfectly align with the social status detectors in our brains, and instead to follow the real facts about it to whatever conclusions they support about how we ought to behave, to do the best we can with the actual state of the world beyond our little corner of it. To put numbers on things, to compare the costs of saving abstract people far away, is to treat those people as really existing, whether or not we ever hear the specific story that would make us feel for them. Someone just as real as a close friend could get to live out their life because of your decision to donate to a malaria charity.

Another motivation is to behave like adults, like how our past selves would have wanted adults to behave. The idea “you should carefully compare the charities you donate to to figure out which will have the best effect on the world” seems so obvious that you’d expect most adults to do it, and yet these “adults” mostly don’t exist. The world is reliably more nuanced, but also more goofy, than you immediately expect, and its full goofiness can often be revealed by just following a chain of thought like “are people actually taking the incredibly easy opportunities to impact the world in line with their values?”

I also find that a lot of people who are into effective altruism, myself included, have at some point felt burned by a moral flare-up we were part of. Things that gave us so much meaning, that seemed so important in the moment, that turned out to be a game we were playing with ourselves and the people around us, that wouldn’t have survived much serial thinking at all. Maybe not even a single Google Doc page of clear premise-to-conclusion thinking about what we were trying to do. As you get older, you start to look out for what you won’t regret, for what will let you feel at home with yourself later on. There’s a deep appeal in using serial thinking not just to access the world, but to live coherently with what your past and future self would’ve hoped you’d do. To gain some distance from any one specific cause you care about, and to invest more of your emotions into the act of adjusting, over time, toward what you actually think is worthwhile. You want to live sincerely.

Over the past 10 years, the world has felt more and more prone to mass movements and moral panics. While EA has generated a lot of weird behavior itself, it has also in some ways been a useful ideological haven from the constant demands to get maximally fired up about the thing of the moment, which usually dissipates into nothing shortly after. Some people have developed strange defense mechanisms against being captured by these panics, leaning hard into nihilism or performative cruelty to show they won’t be mind-hacked. I think this gives way too much power to the sources of the mind-hacking, and I’d rather calmly and technocratically assess the claims from an emotional distance.

This worldview only goes so far

Obviously, this story is also incredibly dangerous. Any story that tells you that you can easily ascend beyond your local status games should be treated with extreme suspicion, because humans are much more social and status-oriented than this simple story implies, and what’s waiting for you beyond your immediate status games can be much worse. Those games exist in part to protect us from new and worse social dynamics that can warp our thinking, be exploited by charming sociopaths, or lead us to believe the normal ethical rules constraining everyday people don’t apply to us anymore. While I’ve loved, and been excited to represent, large parts of the network of people who identify with EA, it doesn’t surprise me at all that loose associations of people organized around the idea that we should investigate the ways we’re taught to do good and try to transcend them through reason have at times led to everything from stultifying groupthink to outright cataclysm. Nothing written into this worldview prevents people who believe it from committing horrible evil, or from being led astray by the strange, invisible new status systems that exist in these peripheral social spheres. Holding views that normal society doesn’t reward can itself turn into a status symbol, and into a status market of its own. EA at its worst can also fall into what Dan Williams calls “naive realism,” where we believe we’ve simply ascended to the brute ethical truth of the world and can easily perceive it.

So whenever I write about what I think is so special about effective altruism, I’m not going to hype up the community. I think the community has collectively saved at minimum tens of thousands of lives and done incredible good for animals, and I increasingly feel proud of a lot of our work in AI safety, but I also see horrible decisions continuing to be made that harm people into the idea, and the world more broadly.

I’ve always approached EA with a pretty heavy epistemic conservatism, and a wariness of abstract human reason. The piece of writing that had the single biggest effect on the way I think was Charles Sanders Peirce’s Some Consequences of Four Incapacities, which notes:

Philosophy ought to imitate the successful sciences in its methods, so far as to proceed only from tangible premisses which can be subjected to careful scrutiny, and to trust rather to the multitude and variety of its arguments than to the conclusiveness of any one. Its reasoning should not form a chain which is no stronger than its weakest link, but a cable whose fibers may be ever so slender, provided they are sufficiently numerous and intimately connected.

I can’t follow a lot of people in EA when their thinking gets too abstract. But I don’t think it’s possible to engage deeply with EA arguments and come away honestly dismissive of the idea that EA has stumbled on some profound ethical problems that deserve way more help than they’re getting, or of the more general point that there’s disturbingly little serial thinking in the world about how to do good (or even how to pick a career). One of my favorite ethics papers is Evan Williams’s “The Possibility of an Ongoing Moral Catastrophe,” which lays out the case that, just given human history, we are very likely in the middle of unseen, gigantic moral catastrophes right now, exactly the kind we’re often stunned that past people lived alongside without trying to solve or even thinking about much. EA is one specific tool for finding these and trying to act on them, and it needs to be surrounded by a lot of outside moral and epistemic rules to keep you from drifting into the more dangerous dynamics of believing you’ve ascended above mere human social games. Here, my personal rules for not going ideologically crazy have come in handy.

It seems obvious to me that a central tragedy of the human condition is that two things are true at once: we are almost certainly living with and tolerating horrific ethical catastrophes that we would act on if we could see them clearly, and attempting to rise above our social situation to identify and solve those invisible catastrophes is itself unspeakably dangerous, carrying the threat of ideological extremism to the point of cults, groupthink, or a flattening of our intellectual and emotional engagement with the world. I think the only way to be morally serious is to try to occupy the emotionally difficult middle ground between these extremes, to allow ourselves to be deeply troubled by counterintuitive moral arguments, and to act on them when the evidence and reasoning are overwhelming, even if that violates our social sense of what the cool good people do. But also to stay intensely wary of, and somewhat apart from, the social and intellectual dynamics we land in when we follow these arguments.

I’ve seen effective altruism as an amazing and dangerous button to press. It scrambled my brain a bit and complicated most of the ethical narratives I used to take as given, and of all the clusters of ideas that could have done that to me, I feel especially proud of the decisions I’ve made as a result of this one, and of the balance I’ve struck with the story it tells. Among other things, I’m sure I wouldn’t have had the success I’ve had with this blog without the tools for thinking I picked up from EA writing. Even my data center stuff came out of the basic observation that applying some simple serial thinking and math to a widely discussed problem (like whether it’s an environmental disaster to use ChatGPT) can, if done well, take you far past the public debate, because way more people are just vibing it out and repeating slogans than you might think.

More broadly, the donations I’ve made, the brilliant people I’ve helped get careers in DC policy areas that I still think are some of the most important in the world, and even some of how I try to talk to friends directly and serially instead of running on social autopilot, all trace clearly back to the ways the EA story of the world disrupted my background sense of status and importance. But I’m also happy I didn’t chase status within EA scenes in ways that would’ve warped my thinking and made me much more confident about AI, animals, or global health than is warranted. And I’ve seen disastrous decisions made in the name of EA, and emotionally stunting social dynamics among true believers, which have reinforced my own epistemic conservatism and my wariness of straying far from my otherwise pretty ordinary ethics. Right now this middle ground, where EA is one important idea but deathly serious idea among many, feels like the most authentic way I can use it. I can’t use it to structure my life or give me a deep sense of meaning, but I can tap into it for a few very broad stories about the world, and after wrestling with the arguments for a long time, occasionally see a massive problem coming a decade early. And I definitely can’t turn away from the basic arguments, go back to my cozy human-scale world, and pretend I never saw them.

My goal in writing more about EA is not to make anyone like us more as people, trust us with power, or put on a lightbulb shirt and show up to an EA conference. Instead, I’d like the aspects of EA that I think are so valuable, and that have shaped my thinking for the last 15 years, to be more available in the general intellectual buffet that smart people choose from to interpret the world. I’d like people to encounter, be challenged by, and use the very best mental tools I’ve personally pulled out of the sprawling intellectual ecosystem EA has spawned, without feeling that they need to give any status to the people (including me) who happened to get into this cluster of ideas in the first two decades of its existence. I think the EA cluster of ideas could have a much bigger positive effect if it were used by a much larger number of people. The specific scene that happened to get into these ideas early is mostly a coincidence, though it also carries deathly important lessons about both the promise and the danger of believing you can use reason and evidence to override the status games around you and do the most good.