Why I'm Not a Negative Utilitarian

Toby Ord


I have been surprised to see that some of my friends and acquaintances in the effective altruism community identify as Negative Utilitarians. Negative Utilitarianism (NU) is treated as a non-starter in mainstream philosophical circles, and to the best of my knowledge has never been supported by any mainstream philosopher, living or dead [1]. This is quite an amazing lack of support: one can usually find philosophers who support any named position. On considering the theory in some detail, I cannot help but agree that the philosophical community has got this one right.

I'm therefore writing this note to those of you who are tempted by NU to show you some of the unsatisfactory consequences it has, and to try to convince you that it is indeed a non-starter.

NU comes in several flavours, which I will outline later, but the basic thrust is that an act is morally right if and only if it leads to less suffering than any available alternative. Unlike Classical Utilitarianism, positive experiences such as pleasure or happiness are either given no weight, or at least a lot less weight. (In what follows, I use the word 'happiness' to stand in for whatever aspects of life might be thought to have positive value).

I shall not argue against the Utilitarian part of this. I am very sympathetic towards Utilitarianism, carefully construed. While it is not widely accepted, it has a very strong intellectual tradition and is supported in various forms by some of the most prominent philosophers alive and dead. What I shall argue against is the asymmetry between suffering and happiness in NU, which is something that I do not think can be plausibly and coherently maintained.


The idea of NU appears to mainly trace back to some remarks made by Karl Popper in The Open Society and Its Enemies [2].

'there is, from the ethical point of view, no symmetry between suffering and happiness, or between pain and pleasure… In my opinion human suffering makes a direct moral appeal, namely, the appeal for help, while there is no similar call to increase the happiness of a man who is doing well anyway. A further criticism of the Utilitarian formula ‘Maximize pleasure’ is that it assumes a continuous pleasure-pain scale which allows us to treat degrees of pain as negative degrees of pleasure. But, from the moral point of view, pain cannot be outweighed by pleasure, and especially not one man’s pain by another man’s pleasure. Instead of the greatest happiness for the greatest number, one should demand, more modestly, the least amount of avoidable suffering for all...'

The context of these remarks was an attempt by Popper to construct three practical principles for public policy. He places minimising suffering alongside the principles of tolerating the tolerant and resisting tyranny.

R. N. Smart wrote a response [3] in which he christened the principle 'Negative Utilitarianism' and showed a major unattractive consequence. A thorough going Negative Utilitarian would support the destruction of the world (even by violent means) as the suffering involved would be small compared to the suffering in everyday life in the world. J. J. C. Smart (the brother of R. N.) makes the same argument in Utilitarianism: For and Against [4]. The reason most of us see to support the continuation of humanity is that there is some kind of positive value in it, but this kind of response is not available to a hardline Negative Utilitarian who is simply trying to minimise suffering.

Popper responded to R. N. Smart by saying that he supposed Smart's unpleasant conclusion would indeed follow from accepting Minimise Suffering as a criterion of right action, but that it was never meant this way. It was just meant as one of three rules of thumb, and only meant for public policy, not individual action [5]. Popper was not, therefore, a Negative Utilitarian (nor any other kind of Utilitarian).

This has pretty much been the end of the debate in the academic philosophical literature about NU. Its main role in the academy is as a position that is sometimes mentioned as a rival to Classical Utilitarianism in introductory classes, then shown to be open to severe critiques and discarded.

Types of Negative Utilitarianism

Before outlining my objections to NU I need to distinguish between several different strengths of NU, since they each have some unique problems, as well as some problems in common. I take all forms of NU to combine some form of consequentialism (i.e. that the right act is the one that leads to the best outcome) with an axiology (i.e. a definition of which outcomes are better than which others). I stratify the types of NU by how much emphasis its axiology places on suffering over happiness.

Absolute NU

Only suffering counts.

Lexical NU

Suffering and happiness both count, but no amount of happiness (regardless of how great) can outweigh any amount of suffering (no matter how small).

Lexical Threshold NU

Suffering and happiness both count, but there is some amount of suffering that no amount of happiness can outweigh.

Weak NU

Suffering and happiness both count, but suffering counts more. There is an exchange rate between suffering and happiness or perhaps some nonlinear function which shows how much happiness would be required to outweigh any given amount of suffering.

The above are all types of true NU. However some people with Negative Utilitarian intuitions don't support any of them. They instead agree that there is a theoretical symmetry between suffering and happiness and thus support the theory of Classical Utilitarianism. However, in a wide variety of practical cases they agree with the Negative Utilitarians in the moral focus on suffering. I'll outline two forms of this:

Strong Practically-Negative Utilitarianism

Classical Utilitarianism with the empirical belief that suffering outweighs happiness in all or most human lives.

Weak Practically-Negative Utilitarianism

Classical Utilitarianism with the empirical belief that it in many common cases it is more effective to focus on alleviating suffering than on promoting happiness.

Objections to Negative Utilitarianism

I'll argue against all four theoretical forms of NU and also against the first practical form. I won't argue agains the weak practical form as I think it is perfectly reasonable and would be supported by almost all Classical Utilitarians. Indeed, I think it is the best way to account for Negative Utilitarian intuitions within a solid theory.

It is always quite difficult to argue against many forms of a view simultaneously as there is normally some kind of change that can be made to create a new form which the argument doesn't engage with. Sometimes this change is to a legitimate and independently motivated new version of the view. Sometimes it is just adding an epicycle.

In this essay, I'm trying to argue in good faith to help my friends see some serious problems with their views, so I hope that the reader (whether friend or stranger) will take the argument in this way. The point is not whether there is some kind of acrobatics or extreme bullet biting that would enable one to avoid each of the arguments I give, but whether they make you think that a once promising theory is looking increasingly like a bad thing to bet the future of our world on.

Absolute NU — The indifference argument

Absolute NU is completely indifferent to happiness (over and above any merely instrumental effects it has on reducing suffering). Suppose there were a world that consisted of a thriving utopia, filled with love, excitement, and joy of the highest degree, with no trace of suffering. One day this world is at threat of losing almost all of its happiness. If this threat comes to pass, almost all the happiness of this world will be extinguished. To borrow from Parfit's memorable example, they will be reduced to a state where their only mild pleasures will be listening to muzak and eating potatoes. You alone have the power to decide whether this threat comes to pass. As an Absolute Negative Utilitarian, you are indifferent between these outcomes, so you decide arbitrarily to have it lose almost all of its overflowing happiness and be reduced to the world of muzak and potatoes.

I think this example pretty much speaks for itself. You would have to be crazy to choose the world devoid of happiness, but Absolute NU says this is just as good and doesn't mind whether you choose it or not. This outcome would be catastrophically worse for all individuals, making Absolute NU a devastatingly callous theory.

As this argument only affects Absolute NU, I imagine that pretty much all people of a Negative Utilitarian persuasion who hear it would at least move to Lexical NU, which accepts that happiness can at least break ties, avoiding this argument. (This is analogous to the case of distributional principles in ethics, where all supporters of maximin have moved to leximin.)

Lexical NU — The pinprick argument

Lexical NU is willing to sacrifice arbitrarily large amounts of happiness to avoid a single pinprick. It therefore succumbs to examples very similar to the one above. For example, suppose one of the lucky people in the utopia of love, excitement, and joy, were to have a tiny amount of suffering amidst their sea of happiness — in a moment of carelessness in their heavenly garden, they prick their thumb on a rose thorn. It is only a very small pain, and they find that it is already outweighed by the pleasure of smelling that very rose. Lexical NU says that it is so important to avoid that pinprick that it would be obligatory to destroy all that is good about their world and force the inhabitants down to the muzak and potatoes lives, if we could thereby avoid that pinprick.

Once again, this outcome would be catastrophically worse for all the individuals (including the one whose suffering we are trying to avoid!). It is not possible to construe this as respectful of the interests of individuals. Lexical NU is thus also a devastatingly callous theory.

A natural retreat for the proponent of Lexical NU is to accept that a pinprick is too small an amount to suffering to warrant sacrificing everything of value to avoid. They can instead hold that there is at least some level of suffering where this would make sense.

Lexical Threshold NU — The continuity argument

As far as I understand, David Pearce supports a version of Lexical Threshold NU [6]:

'It stems instead from a deep sense of compassion at the sheer scale and intensity of suffering in the world. No amount of happiness or fun enjoyed by some organisms can notionally justify the indescribable horrors of Auschwitz. Nor can it outweigh the sporadic frightfulness of pain and despair that occurs every second of every day.'

I share David's sense of horror at Auschwitz. An estimated 1.3 million people died there amidst unfathomable emotional and physical suffering. I can also see that there is a vast amount of suffering in the world every day, though my access to this second fact is more through a process of calculation rather than raw horror.

I don't agree though about whether these quantities of suffering, vast though they are, can be outweighed by the positive side of human experience. One problem with making headway in such a debate is that our intuitions become pretty useless. There are 7 billion people in the world today and it appears to me that the average life has a non-negligible amount positive wellbeing (and has net positive moral value if you think those things are different). I thus think there is a lot of value worth of happiness in the world.

Is it enough to make a single year outweigh the horrors of Auschwitz? I don't have a strong intuition, but I think this is mainly because I'm comparing the suffering of millions with the quality of life of billions. This is a hard thing to have a proper intuition about, since our internal representations of these quantities are basically the same: we find it hard to feel differently about the suffering of one million or the suffering of one billion. This makes me distrust my intuitions, especially as this comes up all the time with millions and billions, so doesn't seem to be a particularly moral feature of my intuition. I'd like to divide each number by one million and make the simplified comparison, but I gather that David Pearce thinks that is not an analogous question.

However, there is an argument which allows us to make serious headway in resolving the disagreement. If you believe in Lexical Threshold NU, i.e. that there are amounts of suffering that cannot be outweighed by any amount of happiness, then you have to believe in a very strange discontinuity in suffering or happiness. You have to believe either that there are two very similar levels of intensity of suffering such that the slightly more intense suffering is infinitely worse, or that there is a number of pinpricks (greater than or equal to one) such that adding another pinprick makes things infinitely worse, or that there is a tiny amount of value such that there is no amount of happiness which could improve the world by that level of value.

The argument goes like this. We can imagine a long sequence of levels of intensity of suffering from extreme agony down to a pinprick, each of which differs from the one before it by a barely detectable amount. If we want to avoid the discontinuity in the badness of the intensity of suffering, then the suffering of a million people at the extreme agony intensity level must be only finitely many times as bad as the suffering of a million people at the next intensity level down, which is only finitely many times as bad as the suffering at the next level, and so on all the way down to the level of the pinprick. This implies that suffering at the very high intensity level is only finitely times worse than the suffering at the pinprick level.

Moreover unless, there is to be a case where n+1 people getting a pinprick is infinitely worse than n people getting a pinprick (for n greater than or equal to 1), we can run a similar argument moving from a million people receiving a pinprick down to a single person and show that the former must only be finitely many times worse than the latter. We thus get the conclusion that while a million people in agony is terrible, it is still only finitely times as bad as a single person receiving a pinprick.

Because we are not discussing strict Lexical NU, we know that the badness of a single person suffering a single pinprick can be outweighed by some amount of happiness. For concreteness, let's imagine that the happiness of a happy year of life with no suffering is enough to outweigh a pinprick (it seems implausible to reject this, but if you do, then choose some other example). It follows that the goodness of stopping a million people suffering in agony is only finitely many times as good as a happy year of life.

Now we can build upwards again. Unless there is a tiny amount of value such that no amount of additional happiness can ever be worth that amount of value, then we can imagine an increasing chain of valuable states of nature, starting with a happy year of life and ending with something X times better (for any X), which proceeds by tiny steps of value. Since we can always add enough happiness to move through this progression, there must be an amount of happiness X times better than a year of a happy life. There must therefore be an amount of happiness so valuable that it is more valuable than avoiding a million people being in extreme agony, contradicting Lexical Threshold NU.

Lexical Threshold NU must therefore accept some kind of discontinuity. Either that there are two very similar levels of intensity of suffering such that the slightly more intense suffering is infinitely worse, or that there is a number of pinpricks (greater than or equal to one) such that adding another pinprick makes things infinitely worse, or that there is a tiny amount of value such that there is no amount of happiness which could improve the world by that level of value.

All of these options look very unintuitive to me — especially when we had reason to doubt the intuition about large number cases which originally motivated the Lexical Threshold view. However, they are not quite so bad as the earlier problems besetting the advocates of Absolute and Lexical NU. I can imagine someone biting one of these bullets if they really needed to. If so, they should make sure to read the 'worse-for-everyone' argument, which is even harder to stomach.

Some proponents of this view would perhaps instead be tempted to modify their view, so that for any amount of suffering, there is always some amount of happiness that can outweigh it, but that this may be a vast amount of happiness, because suffering matters more from a moral point of view than does happiness. This is the view that I've called Weak NU.

Weak NU — The incoherence argument

A major problem with Weak NU is that it appears to be incoherent. What could it mean for suffering to matter more than happiness? It suggests an image such as this:

But what is the horizontal scale supposed to represent? There is no obvious natural unit of suffering or happiness to use. It might be possible to have a consistent scale in the happiness direction and a separate consistent scale in the suffering direction, but it is very unclear how they are both supposed to be on the same scale. This is what would be needed for Weak NU to be a coherent theory and for the diagram to make any sense.

The only things I can think of to put these on the same scale are:

1) How morally good/bad it is.

2) How much it contributes to an individual's wellbeing.

Classical Utilitarians typically use (1) or (2) to set this scale, and since Classical Utilitarianism holds the moral value of an outcome to be the sum of the individuals' wellbeing, it doesn't much matter whether they define this with (1) or (2).

Weak Negative Utilitarians can't use (1) to set this scale, as otherwise they would have to say that a unit of happiness is just as morally important as a unit of suffering and they would just be Classical Utilitarians. I therefore think that the only coherent version of Weak NU is to use (2) to set the common scale between suffering and happiness. However, if they do so, they (along with all other Negative Utilitarians) fall victim to what I think is the main argument against NU.

All forms of NU — The worse-for-everyone argument

In their day to day lives, people make tradeoffs between happiness and suffering. They go to the gym, they work hard in order to buy themselves nicer food, they sprint for the bus to make it to the theatre on time, they read great books and listen to beautiful music when they could instead be focusing on removing suffering from their lives. According to all commonly held theories of wellbeing, such tradeoffs can improve people's lives.

However, there is a big problem for NU in how it assesses these tradeoffs. Absolute NU and Lexical NU say that no such tradeoff can be moral, as does Lexical Threshold NU if the amount of suffering is greater than its threshold. Weak NU says that some such tradeoffs can be moral, but if it accepts (2), it must claim that there are tradeoffs which are good for the individual but morally bad overall. For example, if you think that happiness is only a tenth as morally important as suffering, and are using the contribution of happiness and suffering to wellbeing as your measuring stick, then you must think that very many tradeoffs that successfully improve an individual's wellbeing are morally bad — even if they don't affect the wellbeing of anyone else.

Indeed, suppose that there was a situation in which all individuals want to accept 5 wellbeing units of suffering in order to gain 10 wellbeing units of happiness. This would be in everyone's interest. However, Weak NU would say that it was impermissible, and that it is instead obligatory to prevent it (for very weak versions of NU with exchange rates less than 2 to 1 a new example can easily be constructed). NU is thus not a theory that supports the overall interests of individuals. It may support each of the components of their wellbeing in isolation, but it is biased between them in a way that is counter to promoting their wellbeing. In this way it systematically harms people.

It might be rare to have a case where everyone is made worse off at once, but there are greatly many lesser cases. For example, in some cases it will say that it is immoral to watch the end of the film while you are really hungry, even if this tradeoff increases your wellbeing, because the suffering counts more morally. Other things being equal, you are obligated to prevent your friends and family improving their wellbeing through their judicious tradeoffs too. I find this to be an absurd consequence.

Note that Absolute or Lexical NU could avoid this if they said that happiness didn't contribute to wellbeing at all, or that it only broke ties, but this would be a crazy view about what is good for someone. Lexical Threshold NU could avoid this in cases below its threshold if it said that happiness and suffering were equally important below that level, but this would create a particularly odd kind of discontinuity and wouldn't seem to satisfy their intuitions either.

Implausible practical implications

Many advocates of NU claim that on average human lives have net negative intrinsic moral value. This must be true of Absolute NU, Lexical NU, and Strong Practically-Negative Utilitarianism. It is also the kind of thing that advocates of Lexical Threshold NU and Weak NU tend to say. However, this has some really implausible implications.

For example, it implies that other things being equal it would be good if people get murdered or if one's mother were to die. Things are not always equal, and you may perhaps think that in many cases the bad secondary effects are enough to outweigh the benefits of the people dying. I find this difficult to believe since according to NU, one's mother's death is astoundingly good: roughly as good as eliminating all of her suffering and having her live out the rest of her live in constant bliss. It seems unlikely to me that the pain of family members' grief would outweigh this great 'benefit' (particularly if you remember that your mother would otherwise experience such grief too, and that she would otherwise die at a later date inflicting almost the same amount of grief on loved ones). However, even if secondary effects always conspire to restore a semblance of normality, a Negative Utilitarian would still have to believe that it is great for the person when they get murdered and would be great overall if only people didn't react so badly to it. This is a large bullet to bite.

Similarly, it also implies that much healthcare and lifesaving is of enormous negative value. It says that the best healthcare system is typically the one that saves as few lives as possible, eliminating all the suffering at once. This turns healthcare policy debates on their heads and means we shouldn't be emulating France or Germany, but should instead look to copy failed states such as North Korea.

This list could easily be extended, either focusing on the fact that many forms of NU would prefer almost everyone dead, or on the fact that they would resist people making happiness/suffering tradeoffs which are in those people's interests.


I hope by now I've shown that there are a number of serious looking problems for NU. If you were a supporter of some form of NU, then perhaps you are trying to work out some way to hold onto your view, or to invent a new variant that manages to avoid some of the most serious arguments. Before you do, I'd like to invite you to consider some of the reasons that you might have initially become interested in NU and see if there aren't other views that would fit those intuitions just as well but without such dire consequences.

Typical examples that motivate NU can be be captured by other views. For example, in the earlier quotation, Popper said:

'human suffering makes a direct moral appeal, namely, the appeal for help, while there is no similar call to increase the happiness of a man who is doing well anyway'

Popper's example involves removing suffering from someone rather than adding happiness to someone who is already happy. It thus involves two contrasts. One is between improving the lot of a worse-off person or a better-off person, and one is between improving a life through the addition of happiness or the removing of suffering. There is a well known moral intuition that we should prioritise helping the worse off and this is much more widely accepted than NU. Perhaps this is what is lying behind people's intuitions that they should help the suffering person over the happy one.

To test this, consider a case where the you could avoid minor suffering in the life of someone who has been extremely fortunate and is very happy, or you could bestow some happiness to someone whose life is truly wretched. If you would prefer to grant the happiness in this case, or if you even feel much more uncertain about it, then that is an indication that you were at least partly motivated by concerns for the worst off rather than the primacy of relieving suffering. Such an intuition can be accounted for in the theories of Prioritarianism, Egalitarianism, and Sufficientarianism. I think that these theories are themselves flawed (which goes beyond the scope of this essay), but I find them much more plausible than NU.

Another potential cause for confusion is the interrelation between NU and population ethics. You may think that a clear case for NU is comparing the alleviation of present suffering with the creation of new happy lives. However, this case brings in changing populations, which is the topic of population ethics. The intuition might not be to do with suffering and happiness per se, but to do with the lack of value in creating new lives in comparison to improving existing ones. If this is what guides you, then you should consider views such as Presentism or Critical Level Utilitarianism, which deny that adding a happy life is good, while agreeing that adding happiness to existing people is good and even that preserving the life of existing people is good. Again I don't support these alternative theories, but they are much more supported in the philosophical community than NU.

There is also a type of example that is phrased in terms of whether it would be right or wrong to create a utopia if the very foundation of that utopia required the forced suffering of the innocent during its construction. I agree that there is an intuition of something abhorrent here, but this intuition is normally taken to be a non-consequentialist one. The problem seems to be something to do with the unsavoury connection between the joys of the utopia and the forced suffering. For example, it would still seem problematic if the utopia involved less suffering overall than the alternative society, and this cannot be squared with a move to NU. I think there is also something similar going on in David Pearce's remark that a large amount of happiness wouldn't 'justify' Auschwitz. Justifying it involves some kind of close and deliberate connection that isn't involved in mere 'outweighing', and NU is only supposed to be concerned with outweighing.

Another intuition which I think is at the core of many putative arguments for NU, is that there is a moral obligation to prevent suffering, but no obligation to increase happiness. Indeed, this intuition is shared by many people who think that happiness is good and that it can outweigh suffering. It is normally taken to be an intuition in favour of non-consequentialist theories, which say that increasing happiness is still morally valuable, but it is super-erogatory (above and beyond duty).

In contrast, versions of NU that claim it is not obligatory to increase happiness accord happiness no moral value whatsoever — something that most people find very implausible. Note also that NU cannot just claim that there is no obligation to increase happiness. If it does that, then since it is consequentialist, it must also hold that there is no problem in destroying it, and that we are indeed obligated to destroy happiness if we can therefore avoid some suffering. I don't support non-consequentialism, but I do think it fits our intuitions better than NU here, and I find it more plausible than NU overall. Moreover, the rest of the philosophical community finds it exceedingly more plausible than NU overall.

Finally, Classical Utilitarianism can of course make sense of many of the intuitive cases that lead people to support NU, if we take care to consider the practical effects that typically go along with the cases. For example, the decision procedure (or rule of thumb) of focusing on eliminating suffering seems to lead to more Classical Utilitarian value than addressing suffering and happiness in equal measure. As another example, adding happiness to those who are already happy seems to typically lead to less Utilitarian value than using the same resources to alleviate suffering. However, whether you move on to Classical Utilitarianism, or somewhere else, I would certainly urge you to reconsider whether NU with its remarkable downsides is the best theory for fitting your intuitions.

[Posted 28 Feb 2013.  Last updated 1 Mar 2013]


[1] Note that while Benatar, Schopenhauer, and Buddhism are all strongly against suffering, none of them actually espouse Negative Utilitarianism. The former's view is to do with population ethics and the latter two do not even appear to be Consequentialists of any stripe.

[2] Karl Popper, The Open Society and Its Enemies, vol. 1. Routledge. pp. 284–285 (see also note 6 to chapter 5)

[3] R. N. Smart. 'Negative Utilitarianism', Mind 67:542–3. 1958.

[4] J. J. C. Smart. Utilitarianism: For and Against. pp. 28–30.

[5] Karl Popper, The Open Society and Its Enemies, vol. 2. Addenda I. 13.

[6] David Pearce, 'Why be negative?', http://www.hedweb.com/negutil.htm