Does The Omelas Kid Have An Obligation To Suffer?
I was 16 when I realized I had to kill myself. I was in a Denny’s with my friends, looking at an empty glass of Diet Coke. The glass was produced by taking the raw resources of a foreign country, exploiting its workers in horrible factories, and sent to me to drink out of. And I didn’t really have any other option for getting liquid from a nearby water source into my body, it had to travel through moral atrocity along the way.
It wasn’t just one glass, of course. It was everything. It was the shoes I was wearing (shoes were a big deal in those days) and the flooring I walked on and the food I ate. The only moral act was to kill oneself, and failure to do that meant you accept your role as a vicious monster.1 (The depression helped, but maybe the depression/guilt causality was reversed.)
This wasn’t unique to me. This is the driving impetus behind the desire to destroy modernity and capitalism among Westerners. This is the core conceit behind The Good Place, which posits that all humanity is damned to literal Hell, and has been for many centuries, for exactly this reason. It was very popular show and this reasoning resonated with huge swaths of the population.
(I now think that actually modernity and liberal capitalism are among the greatest forces for good we’ve ever created. I had believed an incredibly false thing for the same reason I believed quite a few other very obviously very false things. But this post isn’t about that.)
The Singer Swindle
This post is about moral duty. People have certain moral duties.2 To provide for their children. To not recklessly endanger others. Moral duties are a big deal because if you fail to meet them you risk losing your standing as a decent human being, and all that entails. Naturally introducing new moral duties can touch off major cultural battles (like whether one has a duty to memorize bespoke pronouns for anyone that demands them).
Mostly moral duties are removed rather than added over time—our society has far fewer than it used to, for both good and ill. But one really important moral duty we’re all agreed upon—if you see a child drowning, and you can save it with no risk to yourself but it would ruin your expensive suit, you have a duty to save that child even at the cost of your suit.
Peter Singer pointed out that the principle of “you have a duty to sacrifice your $1000 possession to save a child’s life” is fully generalizable in his now-famous swindle. If you have a duty to save a dying child even at the cost of $1000, then you have a duty to transfer $1000 to anywhere that this would result in saving a child’s life. Importantly—he’s correct under basic utilitarian assumptions. It’s unfair of me to call this the Singer Swindle because he didn’t invent it, he simply pointed it out. It is an inescapable conclusion of utilitarianism that comes simply from taking utilitarianism seriously. It's a perfect self-consistent system using axioms we already accept.
Singer’s Drowning Child Swindle is the core sacred narrative of Effective Altruism. The vital drive of the movement is summed up in that story, it speaks to the heart of that movement, and it’s been rightly adopted as the foundational myth. It’s a powerful story and has motivated thousands of people. It has also immiserated hundreds of people, and if unchecked will result in the destruction of entire communities.
Coming Back Home To Omelas
The natural conclusion of the Swindle is that everyone has a moral duty to work the maximum amount possible for the benefit of the least fortunate humans until we're all roughly equal in DALYs or QALYs or pick-your-util-measure. The logic cannot lead anywhere else. That's the point of his drowning child pond example. That's why he chose that example and not one that doesn't have an obvious moral duty along with it.
The unspoken result of this is that one has a duty to suffer if it would relieve greater suffering elsewhere. For almost all of us, money is life. We trade away time with our families and our children for money. We sacrifice joy, companionship, even health, for money. Money is bought with some amount of personal suffering. However the small amount of suffering we experience for $5000 is significantly less than the large amount of suffering of a child’s death. So we have a duty to exchange one for the other.
Yes, a duty. Or would you stand by and watch a crying, gasping child drown because you don’t want to buy another suit?
This is strikingly similar to the logic of the tormented child in Omelas. In the original (very short) story an entire city of eternally joyous and maximally-happy people—without worry or strife or violence or disease—are supported by the relentless, grinding suffering of one imbecilic child locked in a basement cell. The suffering (negative utility) of that one child is outweighed by the vast amounts of joy and flourishing (positive utility) it brings to the rest of the city. If one has the moral duty to trade small losses of personal utility for great gains in global utility, one has a duty to become the Omelas child.
Among those who take both utilitarianism and moral duty seriously, we do see this happening in the real world. This particular form of guilt and depression at being unable to force themselves into permanent Omelas-ness is not uncommon in EAs, and it’s a perennial topic of concern. The Singer Swindle weaponizes moral duty to place everyone in a position of permanent evil until all suffering is banished (ie forever). It cripples them for life.
“It’s OK to be a little bit evil”
The edifice of Effective Altruism has not collapsed, because it’s full of good, smart people. And good, smart people will move to fix a failing system so it doesn’t destroy itself and everyone within it. The standard solution is to cap one’s moral duty at 10% of income.
The broader EA/Rationalist/Utilitarian society has adapted pretty well to this problem. Sure, not working the maximum amount and giving it all away is technically failure, but the 10% solution is well defended. The defenders rightly point out that “but humans don’t work that way” or “humans can’t function like that” or “it’s a little bit selfish, but some selfishness is necessary to keep going!” This year’s Bay Area Solstice was dedicated to normalizing the idea that you don't have to kill yourself with altruism! (It was incredibly moving, and I have thought about that Solstice Service many times this year. Ozy did an incredible job.)
All of EA is very accepting of the level of selfishness needed to keep EA alive and doing good (and it’s a lot of good! It checks out on consequentialist grounds!). Scott continues to write apologia for it, as one must with a doctrinal flaw at the very core of a doctrine, most recently talking directly about the Drowning Child Swindle. It mostly works, and the movement continues to be extremely effective at increasing the QALYs of distant strangers. Everyone can both enjoy life and save drowning children.
And yet there is the underlying guilt. The damning knowledge gnawing away at the edges of your soul that you drew the line not because your moral system tells you to, but as a defense against what it tells you to do. You have a duty you are shirking, and that shirking is measured in lives.
Replace The Whole Damn System
As rationalists we want coherent, consistent systems that work on sound principles. I was recently talking with my love about why the hell rationalists care about Decision Theory so much. She said “That whole lecture would've been much better if they led with 'This is important for things that run on formal systems, and thus could be very valuable for aligning machine intelligences, and that's why we care. For humans this boils down to "be trustworthy and people will trust you." ' ”3
That's true, but it's not sufficient for us. Rationalists (and our ilk) actually want to to have coherent formal systems for ourselves too. That is what separates things that actual work from superstition. EA society is saying very loudly “this 10% solution is ok!” because the natural conclusion from the Singer premises is that this is NOT OK at all, and everyone with enough intelligence to follow the argument that leads you to EA in the first place can see this. Yes, their society implementing a patch that says “just ignore the moral system you believe in and stop at 10%, really!” is what is saving these people, and their movement. Saving them is good. That there is something they need saving from is what I have issue with. The system itself is the problem. “You're doing a bad thing4 if you're happy and your community is thriving, but it's OK, everyone's allowed to be bad” is not a philosophy that can exist in a healthy society.
A coherent system that works from its foundational principles and can be logically extrapolated into a full working theory of morality… except that you have to introduce an arbitrary 10% cutoff in the most important practical application or the whole thing implodes… is extremely unsatisfying. Either it works or it doesn’t.
We need an actual systemic decision-theory-style explanation for why spending money on the happiness and well-being of our friends, our parents, our children, and our community is far more important than spending it to probably save a stranger on the other side of the world. Even if they are a child. Saying "it's OK to be selfish" doesn't cut it. We do not have any such system right now. We’re stuck out here without a map, and the one we’re being given with a 10% cut-off scrawled on it with crayon is obviously not going to cut it. We require something that gives clear, systemic reasoning as to which things/people are owed which moral duties and why. I suspect we won’t have anything like that for quite a while. But I’m looking, and if anyone sees something promising I’d love to hear about it.
Even back then I’d prefer death to a return to subsistence farming. This should have been a clue.
At least in the christian-descendant West, I dunno about the rest of the world.
This is a simplification. Per her - “once I understood that it was about translating human ideas about trust and cooperation into language that computers can understand it clarified things. And the process of doing that makes us have to break things down and inspect them rather than using intuition which is a wonderful human capacity but can also lead us astray. And since then B has been convincing me that it isn't just for alignment”
A bad thing? Yes, again, Singer deliberately draws a straight line from not donating $1k to watching a child drown in front of you. The moral duty to act is explicitly the same in both, so they are equally bad. That’s the point the hypothetical is designed to demonstrate!





The point of civilization/morality/society is to make there be good places to live. If you take a good place to live (your community) and induce everyone there to ruin their lives in order to cause more people to live in another, different, shitty place (the kinds of places people die of malaria) then you have not improved anything, and indeed, you've made the world worse: you've taken a world with one good place to live and turned it into a place with zero good places to live (ignoring the parts of the world unaffected by these actions).
This is something that comes up a lot in my meditations, so I've got a lot of thoughts. But I don't have the time to make them coherent, and I don't want to bury you with a massive comment. So here's what I feel is if not the most philosophically satisfying, is at least pretty compelling pragmatically:
You may have a moral obligation to help. But 'you' are just a tiny little conscious entity riding around on the back of an elephant, and that elephant has no moral obligation to anyone but itself and the other animals around it. 'You' may feel a genuine desire to help and give everything you have, but unless you can get your elephant onboard, you're not going to be able to give much of anything.
Talking about 'you' as a coherent singular entity that is subject to moral obligations and persuadable by reason just doesn't make sense. Any moral framework that is going to work in the real world is going to have to take into account that the rational agent to whom guilt can be applied to effect change is just one of the concerned parties. If you can get your elephant to agree to giving 10% of your income to charity, then you're doing better than most, and you shouldn't beat yourself up over your inability to alter reality through sheer force of will.