One of the main theoretical questions within utilitarianism is whether one should favor total or average utilitarianism. The former aims to maximize total pleasure (assuming we’re talking about hedonistic utilitarianism) while the latter aims to maximize average pleasure. In other words, the average utilitarian would prefer the existence of relatively few very happy people rather than a lot of people who are just above the limit between happiness and unhappiness. In the latter case, there is a thought experiment, most famously discussed by Derek Parfit, which, according to him leads to a “repugnant conclusion”: if we have a population of billions of people with a good quality of life, we could always imagine a larger population that would be better, in the sense that the sum of happiness would be higher, as long as we just add more and more people whose lives are barely worth living. If we have a population of 10 billion with an average happiness of 3, we could simply compare that to a population of just over 30 billion with an average happiness of 1. The total utilitarian would, presumably, have to choose the latter scenario. But to some people this seems absurd, or even repugnant.
Some utilitarian philosophers (Torbjörn Tännsjö, for instance) simply bite the bullet and accept that they would choose the maximum happiness anyway. I, however, avoid the repugnant conclusion by being an average utilitarian – not simply because I want to avoid the repugnant conclusion, but because it seems more “intuitively” reasonable to me. I don’t see happiness as a kind of free floating mass of happiness particles. Sums of happiness are only thinkable when they are attached to existing individuals. Individuals that do not exist are not relevant to the calculation. Therefore, average utilitarianism seems most plausible. One consequence of such a view is that we should not create more people in the world if we think it would decrease average happiness. Of course, I do not think creating more people generally does decrease average happiness, but it is imaginable that there are points where overpopulation – and other kinds of considerations – might make it so (different points would apply globally, locally and individually).
But the average view can lead to some repugnant conclusions on its own. If we think that calculating happiness is only relevant when happiness is attached to an existing individual, then it seems to be the case that the death of someone is not really a bad thing if the death of that person does not affect anyone else (in the form of emotional pain, for instance), since the happiness level of that person drops out of the average calculation. On the total view, however, happiness decreases when someone dies (given that they had lives barely worth living or higher). The “repugnant” thing about the average view could, thus, be that killing people by itself might not be bad at all, at least for the person being killed. Death would only be bad for the people who mourn the loss of the dead person, and if there is no mourning (or other bad consequences) of killing one particular individual we could not really say it was bad. Or could we? Yes, we could always claim that even if the state of being dead in itself is not bad for anyone, fear of being killed is very strong among most people. This means that having a strong ban on killing is very beneficial for society, because constantly living in fear would decrease average happiness a lot (but there would be other bad consequences as well, for instance, consequences related to the probability of making miscalculations). All this being said, I must still concede that I do not mourn the deaths of the countless millions who have already died and become completely forgotten by the world, as I do not mourn the non-existence of all people who have not yet been born or who could have been born if we did not use contraceptives (regardless of whether they would have been happy people or not). Their state of non-being is irrelevant for calculating happiness in our present world.
However, there is another type of “repugnant” conclusion with average utilitarianism (but it can affect total utilitarianism as well). Imagine if we could raise average happiness by killing off unhappy people. Four people with happiness levels of 4, 3, 2 and 1 could raise their average by killing the person with happiness level 1 (they would get average 3 instead of 2.5). But this involves some illegitimate fudging of the numbers. It is true that the average would increase, but none of the individual numbers would increase; no individual would in reality be any happier as a consequence of the killing. This might mean that the sort of average utilitarianism I am contemplating must avoid this absurd conclusion by being reformulated in some way. In other words, we can compare two averages in different scenarios, provided that any individual numbers actually increase or decrease in the two scenarios. It seems we need an increase in both total and average happiness (total happiness among the survivors, that is).
Nevertheless, one might retort the following: what if some individual happiness levels actually increase in the previous scenario – if, in other words, we got levels of 4, 3, 3 after the killing (maybe because the killed person annoyed one of the individuals very much). Would that killing, then, be justifiable? Again, I would say that, in theory, the state of being dead would not be bad for the person who was killed, but it would be hard to imagine an organized society where things like that are condoned. On the other hand, we should remember that this sort om “removal” (if not killing) of people is something that all societies do in a milder form. For instance, if some people go around lowering other peoples happiness by stealing from them, raping them, abusing them, etc. we remove them (usually temporarily) from society so that happiness levels among the “survivors” (average and total) can increase again (and conversely, if happiness levels were never affected by such crimes it would by wrong of us to punish the perpetrators). Just like an organized society cannot tolerate individual decisions of removal, we must tolerate collective democratic forms of removal (but we should not accept non-democratic forms). In an even milder forms we practice this form of removal by stopping to socialize with people that make us unhappy – and it would, presumably, be hard to claim that this is immoral (unless, perhaps, we have not made any serious attempts to improve the relationship first).