Hide table of contents

This is a summary of the GPI Working Paper “The weight of suffering” by Andreas Mogensen (forthcoming in The Journal of Philosophy). This summary was written by Rhys Southan.

Does the happiness in this world balance out its suffering, or does misery have the upper hand? In part this is a measurement question: has there been more happiness or suffering in this world to date, and what should we expect the balance to be in the future? But this is also a philosophical question. Even if we knew precisely how much happiness and suffering a possible future for the world would have, we still need to decide how to morally weigh suffering against happiness. In “The weight of suffering,” Mogensen considers a view that, if true, would give us a partial answer to this philosophical question.

The view is lexical threshold negative utilitarianism (LTNU), and it states that there is some level of lifetime suffering that cannot be counterbalanced by any amount of well-being experienced by others. In other words, if we create someone who suffers enough, no number of happy lives we create alongside them can make up for this. Mogensen shows how a few plausible premises lead to LTNU before exploring some disturbing implications of LTNU and the challenges of trying to escape it. He then looks at ways to reject LTNU and, in particular, discusses one of the premises — called always outweighable (AO) — that he thinks is important in exploring LTNU as well as in other aspects of population ethics. 

Five plausible premises

The view that it is worse to create an extremely tortured individual[1] and any number of happy individuals than to create no one at all might sound radical, but Mogensen shows we commit ourselves to LTNU if we accept five plausible premises. The first three are what he calls “structural principles”. These are completeness, transitivity and separability

In the context of LTNU, the first of these is the most important to understand. Completeness means that if we have all the relevant information about two populations, we can always compare them and determine which is better, which is worse, or whether both populations are equally valuable.  

The final two premises are CON-RRC and AO.

CON-RRC is short for Contrary of the Reverse Repugnant Conclusion. As formulated by Mulgan (2002), the Reverse Repugnant Conclusion (RRC) is the claim that compared to a large population of people who live “long lives of unalloyed excruciating agony,” a population of people with lives which are just barely worth not living can be morally worse if it is large enough. In other words, it can be better to create lives of pure torture than lives which are barely worth not living when the latter are of a significantly large number. As its name implies, Mulgan finds this view repugnant. CON-RRC is the contrary of RRC: there is some number of tortured lives which it is worse to create than any number of lives which are barely worth not living. 

AO stands for “always outweighable.” This means that the value of any number of good lives can always be outweighed by sufficiently many lives that are worth not living. Creating any number of people who all have excellent lives and some number of people with negative lifetime welfare levels makes things worse if the latter are sufficiently numerous—even if these negative lives are just barely worth not living.

Deriving LTNU from the premises

Mogensen then shows that his structural premises combined with CON-RRC lead to a stronger version of CON-RRC, which he calls CON-RRC*. It states that a population of just a single life lived at a bad enough “tortured” level is worse than a population of any number of people whose lives are only barely worth not living. Rejecting CON-RRC* while accepting the structural premises as well as CON-RRC results in a contradiction, so Mogensen continues his argument with CON-RRC* as a premise.

He then shows how we get LTNU from CON-RRC*, AO, transitivity and separability. In short, AO tells us that any number of additional happy lives cannot compensate for the simultaneous addition of a high enough number of mildly negative lives. CON-RRC* tells us that a single tortured life is even worse than this many merely mildly negative lives. Together, this implies no number of additional happy lives can compensate for even a single additional life at the tortured welfare level.

Objections to LTNU

Accepting a few plausible premises has led us to LTNU. This is fortunate if we happen to like this view. But Mogensen finds LTNU counterintuitive.

His main objection is that LTNU appears to support the desirability of human extinction. It could seem obvious that so long as we keep existing, some new “tortured” lives will come to be—in which case, none of the happy lives that also result from continued human existence could outweigh them if LTNU is right. However, Mogensen points out that LTNU could set the threshold for a disqualifyingly bad life at a level of horrific wretched misery that is so unrelenting and grotesque that we might be reasonably uncertain whether such a life will ever in fact occur. In addition, accepting LTNU doesn’t necessarily commit us to bringing about human extinction even if the threshold for disqualifying bad lives isn’t unthinkably high. LTNU is a theory about the value of populations but on its own, it doesn’t tell us what to do. Even if we expect the continuation of humanity to make the population worse because some future lives will almost certainly meet the suffering threshold, this might not outweigh other considerations in favour of assuring humanity’s future, like that humanity’s continuation brings meaning to human lives (Scheffler 2013, 2018) or that humanity itself may have intrinsic value that we ought to preserve (Frick, 2017). 

Ways to reject LTNU

One way to reject LTNU could be to reject at least one of the structural principles that helped commit us to it. Mogensen can’t see giving up transitivity or separability, as these are too fundamental to how most of us believe value works. Completeness, in contrast, is up for debate. We could accept that some population outcomes might be neither better than, worse than, nor exactly equally as good as one another. But this leads to paradoxes and does not offer as clean a break from LTNU as we might want. We need completeness to get from CON-RRC to CON-RRC*, but combining the original CON-RRC with AO, transitivity and separability would still establish a weaker variant of LTNU, WLTNU, according to which there is some number of tortured lives that cannot be outweighed by any number of happy lives. WLTNU provides less strident support for extinction than LTNU since we might say the number of tortured lives that cannot be outweighed by any number of happy lives is extremely high. But even if we set the threshold at 10 billion, WLTNU still seems to support extinction. Newberry (2021) has argued that even under conservative estimates, the expectation of the number of future people is at least 1025 (10 billion billion billion). It is reasonable to worry that 10 billion future lives could be tortured if the future is this large. 

The only options left for avoiding the disturbing implications of LTNU are to reject either of the non-structural premises that led us to LTNU: CON-RRC or AO. 

The most obvious way to reject CON-RRC is to accept the Reverse Repugnant Conclusion (RRC). But RRC is controversial, so Mogensen considers whether we can reject both CON-RRC and RRC. Indeed we can, but only by giving up the completeness principle. The most acceptable way to reject both CON-RRC and RRC is to accept there is some possible population of tortured lives which is neither better than, worse than, nor precisely equal to a sufficiently large population of lives that are only barely worth not living. But this just leaves us with a more ambiguous version of WLNTU. We now have to accept the sudden extinction of all sentient life would not be worse than humanity’s continuation, no matter how many blissfully happy lives extinction prevents, if extinction also prevents some absolute number of tortured lives as well.

Not everyone would balk at such a conclusion, but most people seem to think that as long as the future of sentient existence would contain a suitably high ratio of happy lives to tortured lives, it would be a shame if all life were to be painlessly wiped out right now.

Mogensen therefore considers rejecting AO. One way to reject AO would be to claim there are some lives which are so good that creating even just one of these can justify also creating any number of lives which are just barely worth not living.

 So, using CON-RRC*, we don’t get LTNU anymore but the weaker statement that a single tortured existence would be worse than any number of lives which are barely worth not living—but these lives could be outweighed by a single really good life. So, outweighing tortured lives with happy lives may not be entirely out of the question.

The importance of AO

Whether or not we accept AO has implications beyond our commitment to LTNU. AO has independent relevance as well. This is because the vast majority of existences may be lives which are just barely worth not living.

Most wild animals are small creatures with short lives who have reproductive strategies that involve producing a gigantic number of offspring which are even more short-lived than the few lucky adults. These lives are too short to count as truly tortured. Nevertheless, the value of such short lives is probably dominated by the pain of their deaths, so it seems likely that most of these lives are barely worth not living. There is also reason to think that the billions of broiler chickens brought into existence every year typically have lives which are no worse than barely worth not living (Norwood and Lusk, 2011).  

AO has mostly been ignored in population philosophy. Yet with so much of sentient existence consisting of lives which are just barely worth not living, AO could turn out to be incredibly important. If AO is true, it could well be better for all sentient existence to cease, since there might well be too many lives which are barely worth not living to be outweighed by any number of happy lives. If AO is false, the addition of very good lives, such as happy humans, could justify the continuation of sentient existence, for now.


Frick, Johann (2017) On the survival of humanity. Canadian Journal of Philosophy 47, 344-67. 

Mulgan, Tim (2002) The Reverse Repugnant Conclusion. Utilitas 14, 360-4. 

Newberry, Toby (2021) How many lives does the future hold? Global Priorities Institute Technical Report No. T2-2021.  Accessed 28.02.2022. 

Norwood, F. Bailey and Lusk, Jayson (2011) Compassion by the pound: the economics of farm animal welfare. Oxford: Oxford University Press.

Scheffler, Samuel (2013) Death and the afterlife. Oxford: Oxford University Press. Scheffler, Samuel (2018) Why worry about future generations? Oxford: Oxford University Press.

  1. ^

    Mogensen’s definition of LTNU does not specify just how bad a life needs to be so that it is always worse to create it no matter how many happy lives are created alongside it, but that isn’t necessary to explore the view’s implications.





More posts like this

Sorted by Click to highlight new comments since:

Executive summary: Lexical threshold negative utilitarianism (LTNU), the view that some level of lifetime suffering cannot be outweighed by any amount of others' well-being, can be derived from five plausible premises but has disturbing implications that are difficult to avoid.

Key points:

  1. LTNU can be derived from five premises: completeness, transitivity, separability, CON-RRC (Contrary of the Reverse Repugnant Conclusion), and AO (always outweighable).
  2. LTNU appears to support the desirability of human extinction, but this may not necessarily follow depending on the suffering threshold and other moral considerations.
  3. Rejecting LTNU requires giving up at least one of the five premises, but this leads to other problems or weaker variants of LTNU.
  4. The AO premise has important implications beyond LTNU, as the vast majority of animal lives may be barely worth living.
  5. If AO is true, the cessation of all sentient life may be preferable; if false, creating very good lives could justify the continuation of sentient existence.



This comment was auto-generated by the EA Forum Team. Feel free to point out issues with this summary by replying to the comment, and contact us if you have feedback.

Curated and popular this week
Relevant opportunities