Utilitarianism is a common ethical viewpoint, especially here, but it is not free of problems. Two of these problems (here collectively called the Undesirable Conclusions) will be discussed here and one of the two given a better name. Origin adjustment will then be used to solve these problems, but at the cost of creating new ones. However, these problems are significantly lesser than the existing ones.
Throughout, I will discuss utility as if it only applies to people. Many argue that it does not, but that adds substantial extra complexity to the wording, and does not significantly change the arguments. It also includes a degree of weighting by importance that adds even more complexity.
I will also assume that all utilitarianism is hedonic in nature. While there are some problems there, it is the easiest to work with, and many utilitarians are hedonic utilitarians, so this isn't a weak man of utilitarianism.
To start with, I will discuss Undesirable Conclusions. These are not simply conclusions that are bad, they are specifically bad-seeming (paradoxical, immoral, etc.) conclusions caused by additivity. The focus here is on two of them: the Repugnant Conclusion and the Apocalyptic Conclusion, and it is these that are actually called Undesirable Conclusions here.
The first Undesirable Conclusion is Parfit's well-known Repugnant Conclusion. It states that, for any world-state, there is a better world-state in which every individual has very low positive welfare.
The proof is as follows:
This is, as the name suggests, repugnant, but relies on two main features:
Following Arrhenius (2000), the first principle will be called Mere Addition and the second called the Egalitarian Principle. While his main impossibility theorem is more complex than this, it should be clear from the above argument that Mere Addition and the Egalitarian Principle imply the Repugnant Conclusion.
There has been a lot of discussion on the Repugnant Conclusion, with many possible ways to deal with it. Here, I will focus on the critical-level solution, which adjusts the origin such that the resulting conclusion is not repugnant.
Now for the second Undesirable Conclusion, the Apocalyptic Conclusion. This states that, given negative utilitarianism, one should destroy the universe. More precisely, it states that any world-state with anyone experiencing non-zero suffering is worse than the world-state with nobody.
The proof of this conclusion is as follows.
Here, the controversial principle, and the easiest to reject, is the first, but this is key to negative utilitarianism.
The actual principle that will be rejected is implicit here. It is that someone experiencing zero suffering has a life with neutral welfare.
The two Undesirable Conclusions are in an imporant sense dual. Each of them looks at a life with little pleasure and little suffering and adds a large number of them up, to create a world that is either very good or very bad. There are three clear solutions to this:
Point 3 is obviously wrong. Addition does, in fact, work like that. The closest to a serious argument in this vein is lexical-threshold negative utilitarianism, where some levels of suffering are lexically worse. This can be considered as them being infinitely bad with respect to lesser sufferings.
Point 2 leads to things like average utilitarianism and maximin. But these clearly have problems.
Average utilitarianism implies an extreme form of the negation of the Repugnant Conclusion, where a world with one person with extremely high welfare is better than a world with a large number of people with negligibly lower welfare.
Maximin, on the other hand, has a problem with utility monsters. It is better for a world to have a large number of people with extreme negative welfare rather than one person with a negligibly more negative welfare and everyone else with extreme positive welfare.
So, the focus here will be on point 1. Why is a life with zero pleasure and zero suffering one that can be added or subtracted without adjusting the utility? And what do either of these have to do with barely being worth living?
In some philosophies (Epicureanism, for example), this life is the ideal, while in others, it is distinctly bad. Just because the number zero is used in these concepts doesn't mean they're the same.
As such, I use a concept I refer to as origin adjustment to deal with this. It is fundamentally taken from critical-level utilitarianism, which refers to a critical level of welfare such that lives below that are neutral or negative, but still worth living.
Fundamentally, the concept of origin adjustment decouples "a life such that adding it keeps the total utility the same" from "a life with zero pleasure and zero pain", as well as "a life barely worth living". An upwards adjustment of the origin means that a life with neither pleasure nor pain would be bad, while a downwards one means that it would be good.
For the Repugnant Conclusion, the appropriate adjustment is upwards. A common description of a life barely worth living is one spent working on an assembly line, eating only potatoes and listening to only muzak.
But this doesn't sound like a neutral life, it sounds pretty crap!
As such, perhaps the origin needs to be shifted upwards, so that a neutral life is somewhat better than that, even if that life is not so bad that it would be better not to exist.
This is the view of some forms of critical-level utilitarianism, and is the view I hold to. While it is still possible to form a Repugnant-style Conclusion, this conclusion only holds that a world with a huge population barely above the critical level is better than a world with a small number of incredibly happy people, and this is entirely acceptable, depending on what the critical level is.
However, there are still problems here. This sort of argument leads to the Sadistic Conclusion:
This is because the lives in G, while worth living, are below the critical level.
However, in Embracing the "sadistic" conclusion, the author argues (convincingly, in my opinion) that the Sadistic Conclusion isn't that bad. The creation of G is a negative thing, as is the creation of B, while the description and name suggests the much stronger claim that there exists a set B of people with lives not worth living such that it is good to create B. This does not follow.
However, in negative utilitarianism, it might.
Here, we have the opposite problem. Even lives that are going well contain some suffering, and so a purely negative utilitarian perspective would have them be negative in value.
As such, the necessary adjustment here is downwards, so that a neutral life can have some positive amount of suffering in it.
There is a history of something similar in the Epicurean concept of aporia and ataraxia (basically, lack of physical and mental suffering, respectively) as actively good states, rather than merely neutral ones. However, contra Epicurus, I treat these as lack of suffering rather than as pleasure, and so consider Epicureanism to be a form of negative hedonism.
With this adjustment, there is a region of non-zero suffering that is considered to be a good life, and so, assuming that there are enough people above that line, it is not good to destroy the universe. If there are not, then it might in fact be good to destroy the universe, but that isn't a problem with negative utilitarianism as such.
The counterpart to the Sadistic Conclusion here appears to be something like the following:
This is because a large number of slightly positive lives (ones with a small amount of suffering) can outweigh the arbitrary amount of suffering.
While this is a problem, it is also one found in regular utilitarianism and negative-leaning utilitarianism. Lexical-threshold utilitarianism could be used instead of negative-leaning, creating a threshold of infinitely bad extreme suffering as well as one of non-bad mild suffering, but this has the same problem as any other lexical-threshold utilitarianism, so I would just bite the bullet and accept the Omelas Conclusion.