2. Stage setting

The view that there are important respects in which we should treat people, places, and times the same for the purposes of practical reasoning is supported by an ingenious argument of Parfit’s (Reference Parfit1984, Part Two). Parfit contrasted three different positions, each with a different structure: a present-aim theory, which someone like Williams (Reference Williams1981a, Reference Williams1981b, Reference Williams1985) might adopt; a self-interest theory, which has been a dominant conception of individual rationality since the early Greeks (Plato, The Republic); and a neutralist theory, favored by classical utilitarians like Mill (Utilitarianism) and Sidgwick (Reference Sidgwick1907). Parfit offered an intriguing strategic metaphor, suggesting that the self-interest theory occupied an indefensible ‘no man’s land’ between the other two positions. Parfit noted that the present-aim theory is a ‘pure’ theory, relativizing the reasons one has to a particular person, place, and time.Footnote ¹ On such a view, for the purposes of practical deliberation, only an agent’s own interests matter, and only those that the agent has here and now, at the time and place where the practical deliberation is occurring. Likewise, neutralist theories are ‘pure,’ in that, for the purposes of practical deliberation, they treat all people, places, and times neutrally, giving equal weight to the interests of all people at all places and all times.Footnote ² By contrast, the self-interest theory is a ‘hybrid’ theory, it relativizes with respect to people, telling each person that she only needs to give weight to her interests, but it is neutral with respect to space and time, holding that a person should give equal weight to each moment of her life, no matter where, or when, it occurs.

Parfit suggested that the self-interest theory’s hybrid position may be unstable and indefensible. In particular, he claimed that the strongest arguments that the self-interest theory might muster in opposition to the present-aim theory, against relativizing with respect to place and time, might analogously tell against relativizing with respect to individuals, and so may carry one from the present-aim theory, beyond the self-interest theory, and all the way to the neutralist position. Likewise, Parfit suggested, the strongest arguments that the self-interest theory might muster in opposition to the neutralist position, against being neutral with respect to people, might analogously tell against being neutral with respect to place and time, and so may carry one from the neutralist position, beyond the self-interest theory, and all the way to the present-aim theory.

Thus, Parfit suggested, there are powerful reasons for any theory of practical reasoning to be ‘pure,’ and to treat people, places, and times alike: either we should be relative with respect to all three, or we should be neutral with respect to all three, but what we should not do is to treat people, places, and times differently insofar as we believe that reasons are either relative or neutral.

Many theorists accept the view that we should treat people, places, and times the same, and that in fact we should be neutral between them. This is the view of many consequentialists, including all of the classical utilitarians.

The view that rationality requires us to be neutral with respect to time was expressed nicely by Henry Sidgwick, when he wrote that:

Similar claims might be made with respect to people and places. Thus, echoing Sidgwick, one might hold, regarding persons, that:

Likewise, one might hold, regarding space, that:

Now, as an egalitarian, who also believes in certain agent-relative duties and permissions, I have never been attracted to the sort of ‘pure’ neutralist position of classical utilitarianism, whose sole focus is on how much utility obtains, without regard to how it is produced or distributed by, and across, different people, places, and times. Still, there is a powerful attraction to the kind of reasoning expressed by Parfit and Sidgwick, in support of the views that in some important respects, we should treat people, places, and times the same, and be neutral with respect to all three. Unfortunately, here, as elsewhere, the devil is in the details and, sadly, an adequate account of the details has not yet been given. More to the point, whatever kernel of truth there may be to the sort of views expressed by Parfit and Sidgwick, I believe that there is good reason to reject any blanket suggestion that we should treat people, places, and times the same, or be neutral between them, beyond the standard agent-relative objections that have been mustered against such positions.Footnote ³ While I cannot fully defend my view here, in the remainder of this article, I shall present a few of the considerations that underlie my thinking about this matter. As we will see, the question of whether we should treat people, places, and times the same goes well beyond the issue of whether reasons should be neutral with respect to each, relative with respect to each, or neutral with respect to some but relative with respect to others.

3. Some musings about space and time, and worries about treating them the same

It may seem obvious that we should treat space and time the same. However, I’m not so sure about this, especially if it is supposed to be an a priori truth that holds regardless of the metaphysics of space and time. Suppose, for example, that we lived in a universe that extended infinitely in all directions, spatially, and infinitely towards the past and future, temporally. Suppose, further, that time’s passage is not an illusion, and that time is ‘directional,’ such that the past is receding at a constant rate from the steadily changing present, even as the future is steadily moving at the same constant rate towards the present.Footnote ⁴ Perhaps such a view is incoherent, or metaphysically impossible, but if not, must we treat space and time the same for the purposes of practical deliberations?

Consider the following thought experiment. Suppose I learn that our civilization will live in our galaxy another 1000years, and then die out. I also learn that in a distant galaxy, another civilization will exist for the same 1000years, and then die out. I then learn that this is also the case in some third and fourth distant galaxies. I find this all quite interesting. It is somewhat pleasing to me to learn that there are, in fact, advanced civilizations living in galaxies far away.

Next, suppose I also learn that beyond the fourth galaxy, there is nothing but cold, empty, space. This, too, I find interesting. However, I must confess that learning that fact bothers me some, but not very much. Indeed, if events beyond the fourth galaxy were about to unfold which would make those distant reaches inhospitable to all life forms in perpetuity, I wouldn’t think it especially important for us to make significant sacrifices, if we could, to prevent that from happening.

Suppose, on the other hand, I vary the story. As before, I learn that civilization in our galaxy will die out in 1000years, but I learn that after ours dies out, another civilization will arise and persist for 1000years in a second galaxy. I also learn that this will happen again, a third and fourth time. But after that, I learn, there will be nothing but cold, empty, space, forever. For some reason, that knowledge would bother me a lot. Indeed, if events were about to unfold which would make the universe uninhabitable for any life forms 4000years from now, unless we made significant sacrifices to prevent that from happening, I would feel quite strongly that we should do so, and I would feel that way even if I knew that our civilization was going to die out in 1000years, and that the distant future civilizations would do nothing to further our particular hopes, projects, or ideals.Footnote ⁵

Here is a variation of the example. Suppose that I am living in a world where the only sentient beings are human, where our civilization will have persisted for a total of 10,000years before dying out, and where there are 10 billion people alive during each period where our civilization exists, each of whom is at a high level, h. I believe that no other civilizations exist elsewhere in space, and that no other civilizations existed before ours, or will exist after ours. I then learn that I am mistaken in one of two ways. Either I am mistaken about there being no other civilizations in space, and in fact there are 10,000 other planets that will be populated by 10 billion beings also at level h during the same time period that our planet is populated, but I am right that no other civilizations will exist in the universe prior to, or after, our civilization; or, alternatively, I am mistaken that no other civilizations exist during another time period, and in fact that are 10,000 other civilizations of 10 billion beings, each of whom is living at level h in a distinct non-overlapping 10,000year time period of its own, but I am right that there are no other civilizations living elsewhere in space during the time period where our civilization persists. Here, I believe it would be much better if I were mistaken in the second way than the first. And that is because I believe it would be much better for there to be 100,010,000years where different groups of 10 billion sentient beings are living at a high level, than for there to be only 10,000years where the universe is occupied by sentient beings with high-level lives, even if, during those 10,000years, there would not merely be 10 billion people alive, but 100,010 billion people alive.

Note, in both cases, there would be the same sum total of utility. Indeed, in both cases, the very same people might exist at the very same levels of well-being. Still, I believe that the alternative where the many people with high levels of well-being are dispersed throughout time, so that there are lots of cases of high-quality lives stretched over many eons, is better than the alternative where the many people with high levels of well-being are dispersed throughout space, so that there are lots of cases of high-quality lives stretched over many miles, or acres.

Let us add another variation to the example. Suppose that I was mistaken in both respects. The reality is that our civilization is at the end of a long line of 10,001 contemporaneously populated planets, like ours, extending deep into space along a single ray from the center of our planet, and, in addition, it is the last of 10,001 non-overlapping civilizations, also like ours, extending deep into the past for roughly 100,000,000years. I then learn that long ago, an infallible predictor set matters in motion that would determine how the universe would unfold, depending on the actions of the members of our planet.

If we took certain steps, A, which would lower the quality of our lives by 20%, then, in fact, there would be an additional 10,000 contemporaneous civilizations extending deep into space along another ray from the center of our planet, each of whose members would be at level h. If we didn’t do A, then there would be no other planets elsewhere in space populated by sentient beings during our civilization’s existence, other than the 10,000 contemporaneously populated planets about which I already know. In addition, I learn that if we took certain other steps, B, that would also lower the quality of our lives by 20%, but, in that case, there would be an additional 10,000 non-overlapping civilizations, like ours, each of whose members would be at level h, extending forward in time for a total of another 100,000,000years after our civilization dies out. If we don’t do B, the universe will remain utterly devoid of all sentient life once our civilization, and the other 10,000 contemporaneous civilizations, come to an end.

Faced with the knowledge of these alternatives, I think it would be important that our civilization took steps B, to ensure that high-level sentient life persisted in the universe for another 100,000,000years, and that it would almost certainly be wrong of us not to do so. I think it would be much less important for our civilization to take steps A, to ensure that there be even more high-level civilizations living in space at the same time as ours, in addition to the 10,000 other such civilizations that will already be existing elsewhere in space during that time. Moreover, I think it would probably not be wrong of our civilization to fail to do A.

My own view about this case is that the universe might well go best if we did both A and B, go second best if we did B but not A, go third best if we did A but not B, and would go worst if we failed to do either A or B. Hence, I am not denying that, other things equal, it might be important to populate different regions of space with high-quality lives. But the key point, for my present purposes, is that I think we should treat time and space differently, in this context. Specifically, I believe that, in certain cases at least, we should give greater priority to filling differing periods of time with quality life, than to filling different locations in space with quality life.Footnote ⁶

Is it crucial to this example that the different periods of time to be filled come after our civilization will die out? Not to my mind. I would feel similarly about the greater importance of doing B, rather than A, if as a result of our doing B, which would lower the quality of our lives by 20%, there would be an additional 10,000 non-overlapping civilizations, like ours, each of whose members would be at level h, extending backward in time for a total of another 100,000,000years before the 100,000,000years of civilization that preceded our civilization’s existence. Thus, my thought is not merely that it is more important for high-quality life to be dispersed into the future, rather than to be dispersed across space, but rather that, more generally, it is more important for high-quality life to be dispersed across time rather than across space.Footnote ⁷

Here is another example. As before, suppose, that our civilization consisted of 10 billion people, all at the high level of h, and that altogether our civilization would last 10,000years, before dying out. I then learn that we were the first sentient beings to exist in the universe, and that no other sentient beings will exist for another 7,000,000,000,000,000,000,000,000,000 (7octillion)years, after which there will be one more sentient civilization, like ours, with 10 billion people that will last for 10,000years, before the universe grows cold, forever. I would, of course, feel incredibly lucky to know that we were one of only two sentient civilizations that would ever exist in the whole duration of the universe. But I would also feel that the overall goodness of the universe was a pittance compared to what it might have been. I would think it a great cosmic tragedy that the vast majority of time periods in the universe were utterly devoid of high-quality sentient life. Correspondingly, if our civilization could somehow find a way to ensure that high-quality sentient life would continue for the seven octillion years after our civilization dies out, I believe there would be powerful reason for us to do so, and I believe this even if it would require substantial sacrifice on the part of our civilization to bring about the better outcome.

Consider, next, the following. If one travels by plane from one end of the US to the other, and gazes out one’s window, one may be struck by the fact that outside of a few major metropolitan areas, much of the US consists of vast unpopulated tracts of land. The same is true for Canada, Australia, Russia, and much of Africa, Asia, South America, and Northern Europe. Iceland is almost devoid of people; Greenland, the Artic, and Antarctica even more so. Moreover, 71% of the Earth’s surface is water. When I think about these facts, I don’t think that it is a great cosmic tragedy that the vast majority of spatial locations, right here on Earth, are utterly devoid of high-quality sentient life. Should I?

Nor do I believe that my reaction here is solely due to the assumption that if all those spaces were filled with sentient life, none of it would be of high quality. Even assuming that everyone who would live on the Earth would have a high-quality life, I don’t see a compelling reason to increase the size of the Earth’s human population from 7billion to, say, 100billion or more, even if we could.

Or let us turn our gaze inward, for a moment. It is estimated that 1% of every atom is composed of protons, neutrons, and electrons, and that the other 99% is empty space.Footnote ⁸ And it is estimated that the average human adult has 7,000,000,000,000,000,000,000,000,000 (7octillion) atoms in his or her body. There are over 7billion humans on the Earth. Given all this, should we regard it as a great cosmic waste that the vast majority of spatial locations within each human body is devoid of high-quality sentient life? Would our world, or the universe, be much better if within each atom of each human being (not to mention all the other atoms on Earth!), there were subatomic sentient beings possessing high-quality lives? If, contrary to fact, our civilization could somehow find a way to create subatomic sentient beings with a high quality of life, would there be powerful reason for us to fill each of the 7octillion atoms of each ‘average’ adult human, as well as all of the other innumerable atoms of the rest of the 7billion members of the human population, as long as when doing so our own quality of life remained the same? Would there be powerful reason to bring about such an outcome, even if doing so required a substantial sacrifice on the part of our civilization?

Suffice it to say, when I think about all the locations of space on Earth devoid of sentient beings with high-quality lives, I’m not moved in anything like the way I am when I think of vast regions of time that are devoid of such beings. In sum, in general, I think it important that many times be filled with flourishing beings, but not nearly as important that many spaces be filled with flourishing beings.Footnote ⁹ Perhaps I’m mistaken about all this, of course. But, for now, at least, I see no compelling reason to abandon my views about this matter.

I have been focusing on cases involving the high-quality lives of sentient beings. Unsurprisingly, my judgment would flip regarding the relevant importance of filling time vs. space if the lives in question were miserable – well below the level at which life ceases to be worth living.

Suppose, for example, that there are two ways in which the universe might unfold. In one, there are 10 billion planets, each populated with 10 billion people, each of whom lives for 100years, and all of whom are suffering unrelenting agony. In the other, there are 10 billion planets each populated with 10 billion people, each of whom lives for 100years, and all of whom are suffering unrelenting agony. Assume that the very same people exist in each universe, and that no one else exists, other than the 100,000,000,000,000,000,000 people in sheer agony. From the subjective standpoint of each individual, it won’t matter which universe exists. And from the standpoint of total disutility, each universe will be identical. Suppose, however, that in the first alternative, each person lives simultaneously, while in the second alternative, each planet is populated during a different time period. In that case, I think the second alternative is worse than the first. I think a universe where there is vast suffering, but where the suffering only lasts for 100years, is much better than a universe where there is the same total amount of vast suffering, yet there are great numbers of people in great agony not merely for 100years, but for 1000trillion years. So, in my judgment, it is bad if many spaces are filled with agonizing lives that are worth not living, but it is worse if many times are filled with agonizing lives that are worth not living. Thus, the relative importance of filling times, or spaces, with sentient beings, depends on whether the value of those being’s lives are worth living, or worth not living.Footnote ¹⁰

Let me turn, next, to a different point. Suppose that God were deciding to populate an infinite number of planets, and time periods of 100 years each, with sentient beings whose lives were miserable. Her plan is to have 10 billion miserable beings living on planet one billion and one, during time period one billion and one, 10 billion different miserable beings living on planet one billion and two, during time period one billion and two, 10 billion different miserable beings living on planet one billion and three, during time period one billion and three, and so on, for all of eternity. Just before doing so, God decides that She will create the very same miserable beings as She was originally intending to, and that each of them will live during the very same time periods as She was originally intending for them, but that She will shift which particular planets they occupy, so that each person’s place in space would be different.

Specifically, suppose that God decides to put the people who would have occupied planet one billion and one on planet one, instead, the people who would have occupied planet one billion and two on planet two, instead, the people who would have occupied planet one billion and three on planet three, instead, and so on. One might, if one likes, imagine that each of an infinite number of planets are spaced an equal distance apart, say k miles, along an infinite, straight, Euclidean line, so that God’s choice involves placing each person on the planet She originally intended for them, or, instead, shifting each person k billion miles in the same direction along the line of planets to a different planet. To my mind, the difference between these two prospects has no moral significance. As between these options, where everything is the same except for where in space the infinite people lived, I would be utterly indifferent.

Suppose next, however, that God decides that She will create the very same miserable beings and place them on the very same planets as She was originally intending to, but that She will shift when they live. Specifically, suppose that God decides to put the people who would have occupied time period one billion and one in time period one, instead – where, we are assuming, time period one begins one billion years earlier than time period one billion and one – that She decides to put the people who would have occupied time period one billion and two in time period two, instead, the people who would have occupied time period one billion and three in time period three, instead, and so on. I find the difference between these two prospects to be morally significant. Notwithstanding the so-called ‘fact’ of infinity that tell us that, over the course of time, there will be just as much miserable existence in each of the two alternatives, I believe there is reason to favor God’s original plan, over Her revised plan, in which miserable existence will begin one billion years earlier, and then continue, unabated, afterwards.

If you aren’t convinced by the previous example, consider a variation that would directly affect you. Suppose that God tells you that She is planning to send you to Hell, where you will suffer unbearable pain. She then tells you that once you are sent to Hell you will remain there forever, but that She is willing to give you some choice as to when your agony begins. You can either start immediately, or you can start sometime later; however, if you choose to start later, God will put you in a state of suspended animation between now and when you start, so that you will experience nothing between now and when you begin your torments in Hell. Dismayed, you begin asking how long you can delay it. A week? Sure. A Month? No problem. What about a year? A decade? A century? God is happy to go along with any of those choices. You decide to be bolder. You ask if you can delay by a million years. She agrees that that, too, is possible. You decide to be bolder still. Having recently learned how big an octillion is, you ask God if you could delay your entry to Hell by an octillion number of years. At this point, you have tried even God’s (infinite!) patience, and She replies that yes, indeed, you can delay your entry by that much, but no longer!

At this point, what would you decide? Would you reason that since you won’t be gaining anything positive by delaying, and will be spending an eternity in Hell once you’re there, it doesn’t matter when you start, since at the end of time, as it were, you will have spent just as much total time in Hell? Or would you choose to enter a state of suspended animation, and delay your entry as long as possible, taking the option of entering Hell in an octillion years? I know what I would choose, for myself, or anyone else that I dearly loved. I would choose the latest possible entry date that God permitted, and I believe that it would be perfectly rational for me to do so, and irrational for me not to do so!Footnote ¹¹

Suppose, next, that God tells you that she is planning to send you to Hell immediately. She further informs you that Hell consists of an infinite number of planets spaced an equal distance apart, say k miles, along an infinite, straight, Euclidean line, each of which is labeled by an integer. As it happens, she is planning to place you on planet 1, where you will remain for one year, after which you will be on planet 2 for a year, then planet 3 for a year, and so on, for all of eternity. However, if you want, you have the option of entering Hell on any of the other planets located along the infinite straight line, where you will remain for one year, after which you will be moved to the next highest numbered planet where you will again remain for a year, and this pattern will continue for all of eternity. You immediately ask if any of the planets are less torturous than the others. She assures you that they are not. They are all equally torturous. At that point, would you bother to shift your entry point into Hell to a different planet at a different location in space? Would you try to bargain with God to please let you move your entry point from planet 1 to planet 2, planet 100, planet 1 million, or planet 1 octillion? I see no reason why one would. If, in fact, Hell’s planets are all equally bad, it seems clear that there would be no rational basis for preferring to be in one particular location in space rather than any other.

I realize that these past two cases involve infinity, and that our intuitions about such cases are notoriously problematic. Nevertheless, I don’t think my views about these cases are implausible, and they suggest a further asymmetry about space and time for the purposes of practical reasoning. In some infinite cases, at least, merely shifting the spatial locations of sentient beings will be morally irrelevant, whereas shifting the temporal locations of sentient beings may be significant.Footnote ¹²

4. Dominance principles with respect to people, places, and time

Parfit’s argument against the Self-Interest Theory suggests that rationality requires that we treat persons, places, and times the same in certain key respects, and, in particular, that if we should be neutral with respect to one, we should also be neutral with respect to the others. Reasoning along similar lines, it might seem natural to assume that if we accept a dominance principle with respect to one of these categories, then we should also accept a similar dominance principle with respect to the other categories. Consider, for example, the following three dominance principles regarding utility.

Intuitively, many would find each of the preceding dominance principles plausible. Moreover, as indicated, influenced by reasoning of the sort appealed to by Parfit, many might assume that if one of the dominance principles is true, then the others must also be true. But this assumption is clearly false. To see this, consider Diagram 1.Footnote ¹³

Diagram 1.

O₁ is one possible outcome. In that outcome, there is one person, P₁, living in time period one, and spatial region one, who has a good life, well above the level at which life ceases to be worth living, but there are twice as many other people, P₂ and P₃, who have bad lives, well below the level at which life ceases to be worth living. In time period two, P₁ through P₃ have moved to spatial region two, where they all enjoy good lives, but unfortunately, in that time period, and at that location, twice as many other people, P₄ through P₉, have come into existence, and their lives are as bad as P₂ and P₃’s lives were during T₁. In time period three, P₁ through P₉ have all moved to spatial region three, where they all enjoy good lives, but unfortunately in that time and location, twice as many other people, P₁₀ through P₂₇, have come into existence, and their lives are as bad as P₂ and P₃’s lives were during T₁. Outcome One continues to unfold, in this ever-expanding manner, forever, with each time period lasting for one day, and each person living for 100years total, before dying. Here, and below, we assume that the positive value of each good moment is the same, the negative value of each bad moment is the same, and that the two values sum to zero, so that a life containing an equal number of moments of good and bad life will have a net value of zero, a life containing more moments of good life than bad life will have a positive net value, and a life containing more moments of bad life than good life will have a negative net value.

Outcome Two contains the very same people as Outcome One, P₁, P₂, P₃, etc., and is analogous to, though the reverse of, Outcome One. Specifically, in Outcome Two, there is one person, P₁, living in time period one, and spatial region one, who has a bad life, well below the level at which life ceases to be worth living, but there are twice as many other people, P₂ and P₃, who have good lives, well above the level at which life ceases to be worth living. In time period two, P₁ through P₃ have moved to spatial region two, where they all suffer bad lives, but fortunately, in that time period, and at that location, twice as many other people, P₄ through P₉, have come into existence, and their lives are as good as P₂ and P₃’s lives were during T₁. In time period three, P₁ through P₉ have all moved to spatial region three, where they all suffer bad lives, but, once again, fortunately in that time and location, twice as many other people, P₁₀ through P₂₇, have come into existence, and their lives are as good as P₂ and P₃’s lives were during T₁. As before, Outcome Two continues to unfold, in this ever-expanding manner, forever, with each time period lasting for one day, and each person living for 100years total, before dying.

Given our assumption that each moment of good life would balance equally against a moment of bad life, and the further moral assumption that in this example, there is no reason to favor one person over that of any other, how do Outcomes One and Two compare regarding utility?

On an Impersonal Neutralist View, of the sort favored by classical utilitarians, O₁ and O₂ might be judged as equally good. After all, if one is indifferent to where in space, time, or lives goods or bads are located, and only pays attention to how much total good and bad exists in the world, then one may judge that O₁ and O₂ are equally good, since, in each outcome, there would be an infinite number of good days, and an infinite number of bad days lived, and in each case the orders of infinity of the number of good and bad days would be the same.

Alternatively, if one compares the two outcomes place by place, or moment by moment, Outcome Two would be clearly better than Outcome One, in accordance with the Spatial and Temporal Dominance Principles. This is because for every spatial region, S _n , and every temporal region, T _n , there will be twice as many people with good lives as with bad lives in Outcome Two, while there will be twice as many people with bad lives as with good lives in Outcome One.

So, should we conclude that, regarding utility, Outcomes One and Two are equally good, in accordance with the Impersonal Neutralist View, or that Outcome Two is better than Outcome One, in accordance with the dominance principles with respect to space and time? I find such judgments very hard to believe. After all, in O₁, each person has exactly one bad day, and the rest of his or her 100year life is good. In O₂, on the other hand, each person has exactly one good day, and the rest of his or her 100year life is bad. I know which of these outcomes I would want to obtain for myself, a loved one, or anyone else who was not pure evil!

O₁ is a world where everyone has lives that are very good every single day but one. O₂ is a world where everyone has lives that are very bad every single day but one. Clearly, every member of O₁ has a life which is, on balance, well worth living, whereas every member of O₂ has a life which is, on balance, well worth not living. Given all this, I firmly believe that Outcome One is better than Outcome Two, in accordance with the Personal Dominance Principle regarding utility.

In this example, we can accept the dominance principle regarding people, or we can accept the dominance principles regarding space and time, but we cannot do both! Here we have a proof that, unless we reject all three dominance principles, in some cases, at least, we should not, and cannot, treat space and time the same way as we treat people. So, should we reject all three dominance principles, in favor, perhaps, of the Impersonal Neutralist View? I don’t see why. At least in the sort of case that we have been considering, the Personal Dominance Principle seems clearly true!Footnote ^14, Footnote ¹⁵

The preceding argument suggests that, for certain cases, at least, we should give priority to distributions of well-being across people over distributions of well-being across time. And earlier, I suggested being more concerned about distributions of well-being throughout time, than throughout space. The priority rankings of people over both time and space, and time over space, for some cases, at least, might be further buttressed by considering Diagram 2.

Diagram 2.

In Diagram 2, O₁ and O₂ are outcomes with an infinite number of people, P _i or Q _j , with each person, located at a particular location in space, S _k , and a particular location in time, T_l, at a level corresponding to one of the integers. So, for example, in Outcome One, person P₀ is at level 0, at temporal location 0 and spatial location 0, while in Outcome Two, person Q₋₄ is at level −4, at temporal location −3 and spatial location −5. For the purposes of this example, I am assuming that the metaphysics of space and time allow for the identification of the same spatial and temporal locations across different possible outcomes, so that for each k and l, S _k corresponds to the very same location in space in each outcome, and T_l corresponds to the very same location in time in each outcome. If such an assumption is coherent, then Diagram 2 illustrates that, for some cases, at least, the Spatial and Temporal Dominance Principles are incompatible with each other. Thus, for such cases, we can reject both, but we can’t accept both. This is because, in Diagram 2, Outcome One is better than Outcome Two at every point in time, but it is worse than Outcome Two at every point in space.

Now assume, temporarily, that the populations of the two outcomes are wholly distinct. In that case, I can see why someone might claim that each outcome is equally good, since each involves an infinite number of people, such that for each integer, there is exactly one person whose level of well-being is accurately represented by that integer. In that case, one would be rejecting both the Spatial and Temporal Dominance Principles. But my own judgment, in this case, is that we should accept the judgment yielded by the Temporal Dominance Principle, and reject the judgment yielded by the Spatial Dominance Principle. That is, in this case, I would judge Outcome One as better than Outcome Two, since it is better at each moment in time, and, to my mind, there is neither a compelling reason to ignore this consideration, nor a countervailing reason outweighing it.

But, of course, as the previous discussion makes plain, I believe that there could be such a reason. In particular, if the same people would exist in each outcome, and they would each be better off in one of the outcomes than the other, then, in accordance with the Personal Dominance Principle, I would regard the outcome in which they were all better off as better than the other outcome, regarding utility, regardless of how the two outcomes compared in accordance with either the Spatial or Temporal Dominance Principles.

5. Avoiding cases involving infinity

Some people will worry about some of my examples because they involve appeals to infinity about which our intuitions are notoriously unreliable. For those who have such worries, let me make several comments.

First, I believe that one must distinguish between different kinds of examples involving infinity. Some rely on moves that are clearly dubious, for example, when one ‘reorders’ the different members of an infinite sequence, say, by ‘moving forward or backwards’ certain members of the sequence, but not others, in order to shift our intuitions about the overall value of the infinite sequence, either when considered by itself, or in comparison with some other sequence. I fully agree that any intuitions that we might have about such ‘reordered’ infinite sequences are not to be trusted, but I note that none of my appeals to infinity involve such dubious moves. Indeed, I believe there is nothing ‘tricky,’ artificial, or dubious about the infinite sequences that I have considered in this article, that provides good reason to doubt our intuitions or judgments about them. To the contrary, although I cannot pursue this further, here, I believe that there are good reasons to accept our judgments about the various examples I have invoked involving infinity.

Second, I remind the reader that not all of my arguments involved appeals to our intuitions about infinite cases. Many of my arguments focused on finite cases.

Third, while I find some of my cases involving infinity particularly compelling – which is why I employ them – I believe that my main conclusions could have been argued for without appealing to such examples. In particular, I believe that there are a host of strong reasons, that don’t appeal to infinity, to worry about any ‘pure’ neutralist position requiring us to be strictly neutral between any ‘locations’ of people, places, or times at which utility might obtain. I also believe that there are some finite cases where we should reject the rankings generated by the Spatial and Temporal Dominance Principles. In support of these claims, I offer the following observations.

First, consider the widely held view that the shape of a life matters, so that a life that begins poorly, but steadily improves, and ends well, would be better than a life that begins well, but steadily declines, and ends poorly, even if the two lives contained the same total amount of well-being. Clearly, this view is not strictly neutral as to when, in a life, well-being occurs.

But notice, most people holding this view regard space and time differently in this respect. They don’t believe, for example, that if someone was at level 100 for the first 20years of his life, at level 400 for the next 20 years, at level 700 for the next 20years, and at level 1000 for his final 20 years, that the overall quality of his life would vary depending on where, in space, he lived during those different periods. Such a life would be equally good if the person lived in one place all his life, or in different places. Indeed, as long as the levels for each period weren’t affected, he could move to any combination of different places, in any order, without affecting the overall quality of his life. So, this is a non-infinite example where many believe that space and time should be treated differently, as the ordering of well-being in time seems relevant for the overall quality of a life, in a way that the ordering of well-being in space does not.

Next, consider three principles that I presented in my book, Rethinking the Good:

I don’t have time to repeat my arguments for these principles here, but in my book, I noted that most people, including many consequentialists, accept such principles. Such principles reflect an anti-additive-aggregationist approach, which opposes the simple additive-aggregationist approach of classical utilitarianism. For most people, we don’t simply care about how much utility obtains in a given outcome, we also care about how that utility is distributed throughout the outcome, and that means that most people are not strictly neutral, as classical utilitarianism requires, as to where utility is located, with respect to people, places, and times.

Moreover, importantly, our concern about how utility is distributed does not merely reflect our concern for other distributive principles, such as equality or justice. It reflects our judgment about what is relevant to making one outcome better than another, even regarding utility. Thus, there is room for judging that though one outcome has more utility than another, it is still worse regarding utility. And this is because, for many, one fundamental concern is about the way in which different amounts and distributions of utility affect people, for better or worse, and this, I believe, reflects a person-affecting view, and not simply an impersonal neutralist position.

The anti-additive-aggregationist principles have wide appeal, both within, and between, lives. For example, they help explain why many find Parfit’s Repugnant Conclusion repugnant (Parfit Reference Parfit1984, 381–390; Temkin Reference Temkin2012, 34–35, 37, 41–42, 324–328). Even if there is more total utility in Z than in A, Z’s utility is dispersed across many lives, so that each person’s life is barely worth living, whereas A’s utility is consolidated among far fewer lives, so that each person’s life is well worth living. Here, our ranking of A as better than Z reflects the wide person-affecting view that focuses on how the people in the two outcomes fare, and rests on the anti-additive-aggregationist position that lots of tiny benefits spread across innumerable masses don’t add up, normatively, in the way that they would need to to make Z better than A.Footnote ¹⁶

Similar results apply within lives, where most people judge a long life that includes two years of excruciating torture and fifteen mosquito bites per month, as worse than a long life that contains no torture but sixteen mosquito bites per month, even as they acknowledge that if the life were long enough, the total amount of disutility would be greater in the latter life than the former. Here, too, our anti-additive-aggregationist principles tell us that the discomfort of one extra mosquito bite per month doesn’t add up, normatively, in the way that it would need to to outweigh two years of excruciating torture. Here, too, we are not merely concerned with the total amount of disutility in a life, and utterly neutral as to where, when, and to whom (in this context, which person stage) it obtains; rather, we are concerned about how the disutility is distributed throughout the life, and how the person is affected, for better or worse, by that distribution. The plain fact is that some distributions of vast amounts of total disutility can be benign, while some distributions of much smaller amounts of total disutility can be disastrous.

So, bearing all this in mind, return to the example depicted in Diagram 1, but this time, imagine that the outcomes don’t extend infinitely, but ‘only’ for a billion years. How do the two outcomes compare in that case? I believe, in accordance with the spirit of the Disperse Additional Burdens and Consolidate Additional Benefits Views, that O₁ would still be better than O₂, even though there is twice as much badness in O₁ as in O₂, and twice as much goodness in O₂ as in O₁.

How could that be? Well, by now, the answer is familiar. The question, for me, isn’t merely about how much total goodness and badness exists, but about how the people are affected for better or worse, by the distribution of whatever goodness and badness there is. In O₁, there will be many people with good lives for every day but one, for 100years, many with good lives for every day but one, for 99years and 364days, many with good lives for every day but one, for 99years and 363days, many with good lives for every day but one, for 99years and 362days, and so on. To be sure, on the last day of O₁’s existence, there will be a vast number of people who live only one day, whose lives, for that day, will be very bad. But still, the badness for them only lasts a single day, and I don’t believe that that badness adds up, normatively, across the lives of the many who will live only one day, in the way that it would need to to outweigh the really good lives that would have been lived for many years by many others.

In O₂, on the other hand, there will be many people with bad lives for every day but one, for 100years, many with bad lives for every day but one, for 99years and 364days, many with bad lives for every day but one, for 99years and 363days, many with bad lives for every day but one, for 99years and 362days, and so on. To be sure, on the last day of O₂’s existence, there will be a vast number who live only one day, whose lives, for that day, will be very good. But still, the goodness for them only lasts a single day, and here, as before, I don’t believe that that goodness adds up, normatively, across the lives of the many who live only one day, in the way that it would need to outweigh the really bad lives that would have been lived for many years by many others.

In sum, looking at how all the different people are affected for better or worse in each outcome, and taking account of the anti-additive-aggregationist principles that I find compelling in cases like this, as well in many other cases, including the Repugnant Conclusion, I would judge that, even regarding utility, Outcome One is better than Outcome Two. Of course, in making this judgment, I am not denying the obvious truth that Outcome Two has more utility than Outcome One. Of course it does! Rather, I am reflecting the appeal of person-affecting views, that pay attention not merely to how much utility there is in any outcome, but to how the people are affected, for better or worse, by the distribution of however much utility there is.

If my judgment about this case is correct, and I realize that not everyone will agree with me about this, then we have reason to reject the Spatial and Temporal Dominance Principles even in finite cases. After all, as before, Outcome Two is better than Outcome One at every location in space and time. At first blush, this is a somewhat surprising result. However, on reflection, I believe it is the right one.

6. The Limbo Man, the Capped Model, and other unresolved issues

In discussing the badness of death, Kamm (Reference Kamm1993, 19, 49–55) introduces several variations of a case she calls the Limbo Man. Kamm suggests that the finality of death, its permanence, the extinction of our lives may play a special role in our explaining some of the attitudes toward death that are almost universally shared. Kamm suggests that the recognition that once our lives are over, they are over forever, opens the possibility that there might be reason to be concerned about when our lives begin, as well as when our lives end.

Kamm suggests, for example, that if the universe spanned a given period of time, and we were going to live our one and only life for 70years sometime within that span, then even if one assumed that the conscious experiences of our life would be exactly the same on either scenario, and that all the (non-location-in-time-related) goods of life that we would possess would be exactly the same on either scenario, there might still be reason to want our 70-year life to obtain towards the end of the universe rather than towards the beginning.

Kamm suggests that there might be at least two related reasons for this. First, as long as we have not yet existed, the potential for our existing will still be there, and one might believe that there is value in such potential. Second, one might believe that it is better for us, or our lives – though not due to any impact on our states of consciousness or the goods that we possess during our lives – if the time during which we will never again exist is as short as possible, so that our extinction comes as late as possible. On this view, the badness of never existing again is distinct from, and has special significance, relative to the mere badness of not existing, which, of course, will also be true of us at each moment before we come to exist.

For Kamm, then, if someone had a choice of living a normal lifespan, filled with a given set of experiences and goods of life, or a chance of starting that life for a period of time, going into limbo for an extended period of time, and then finishing off the remainder of one’s life many centuries or more later, there could be reason to do the latter, even if it were no better in terms of one’s set of experiences and (non-time-related) goods of life. As indicated, for Kamm, this is because there might be something good about both preserving, as long as possible, the possibility of a period of one’s future existence, and minimizing, as much as possible, the period during which it is true that you will never exist again.

In many respects, Kamm’s discussion is orthogonal to my own. She is concerned about the badness of death, and the asymmetry between our attitudes towards death, and the period during which we won’t exist that comes after it, and our attitudes towards birth, and the period during which we won’t exist that comes before it. I am focused on civilizations, containing large populations, and my concern is not with the prospect of any given civilization coming to an end, but with the possibility of there being large periods of time devoid of any civilizations at all whose members possess high-quality lives. But though Kamm’s concerns are different than mine, they have a bearing on mine in several respects.

First, I have pointed out an asymmetry in my thinking about space and time. I note that the ubiquitous fear of death that has been almost universally held throughout the history of humanity, and which has inspired so much art and literature, is a time-related attitude. It is not a space-related attitude. Kamm’s discussion takes as its starting point the commonplace that the vast majority of humans fear their own death. For many, the prospect that there will be an eternity of time after they die during which they will never exist again is terrifying. Many can’t bear to even contemplate that prospect, and many others simply refuse to accept it. Indeed, throughout human history, religions have arisen to help people confront their earthly deaths, holding out the promise of eternal life.

Nothing akin to this holds regarding the many different spaces beyond ours that we will never, ever, occupy! Looking out into space, we might feel tiny and insignificant. We might even find ourselves deeply disappointed that we never get to explore that great unknown. But we don’t look into space with utter terror at the realization that we will never be there. We have no trouble contemplating, or accepting the fact, that there may be an infinity of spaces, beyond all those that we will ever occupy. And no religions have arisen to help us cope with that reality!

Kamm’s Limbo Man lives for a period, puts himself in limbo, then lives out the duration of his life at a much later period of time. In doing this, he gains nothing in terms of his experiences or the (non-time-related) goods of life, but he succeeds in significantly delaying the day when it will be true that he will no longer ever exist again. Kamm thinks that there could be reason to be this kind of Limbo Man. But Kamm doesn’t consider a Limbo Man who lives in different locations, puts himself in limbo, then lives out the duration of his life at a place much further away in space than the place he was when he went into limbo. This is, I believe, no accident.

If Kamm is right, there can be rational significance to when we live in time, connected to the desirability of both the potential that we shall one day exist in the future, and the delay of the time after which we shall never exist again. But there seems to be no analogous rational significance to where in space we live. Assuming that our experiences and (non-space-related) goods would be the same, the potential of our coming to exist at one point in space seems no more valuable than the potential of our coming to exist at any other point in space, far away from the first. Nor does there seem to be any rational reason to want to start our lives at one point in space, and finish them at another, merely so as to reduce the amount of space outside the first point that will never be occupied by us.

In sum, I suggest that Kamm’s discussion of our attitudes towards death – attitudes which are decidedly time related but not space related – implicitly support my contention that there is reason to treat space and time differently for the purposes of practical reasoning. However, Kamm’s discussion is of interest to me not merely for that reason, but because it raises a host of interesting questions about individual lives that might be similarly raised about large groups of lives.

In this article, I have contended that it is more important that large gaps in time be filled with many high-quality lives, than that large gaps in space be filled with high-quality lives. But even if that is so, there are a host of other, Kamm-like questions that might be raised. For example, suppose that there will only be one super-large civilization in the whole of the universe’s existence, and that the very same people will exist in that civilization and all have lives of the very same high quality of existence, no matter when in time that they existed. If there would be, say, a quintillion (1,000,000,000,000,000,000) people, each of whom lived for 100years with a high quality of life, would it matter if they lived later, rather than earlier, in the life of the universe? For example, if the life of the universe, and time itself, had a beginning and an end, and all of the people were going to be alive during the same 10,000-year period, would it be better if they all lived in the middle rather than the beginning, but better still if they all lived at the end?

Similarly, even if I am right that, in general, it would be good if the different ‘empty’ periods of time were ‘filled’ with high-quality lives, that still leaves many possibilities open. For example, suppose that we thought that 100billion people were enough to ‘fill’ any given 10,000-year period of time. Then, on my view, it would be better if the quintillion people were spread out in time, so that there were 1000 distinct 10,000-year periods, each filled with 100billion people. Still, that leaves it open whether we think it matters how those different periods were dispersed throughout the life of the universe.

Retaining our previous assumption that the universe, and time, had a beginning and an end, one might think that it doesn’t matter. Or one might think, perhaps influenced by Kamm, that it would be best if the different periods were all bunched together, so that there was a steady run of high-quality life for 10million years coming at the very end of time. Perhaps one might think that it was important that high-quality life span the entire life of the universe, but that all this required is that the very first 10,000-year period be populated with lots of high-quality life, and that the very last 10,000-year period be populated with lots of high-quality life, but that beyond that it wouldn’t matter how the different 10,000-year periods of high-quality life were distributed through time, as long as they remained non-overlapping. Alternatively, perhaps one might think it best to distribute the different periods of high-quality life throughout time, so as to minimize the length of any period of time during which there would be no high-quality life!Footnote ¹⁷

These questions are not intended to be exhaustive, but merely indicative of the wide range of issues that need to be considered once one acknowledges that, ceteris paribus, it is good if ‘empty’ periods of time be ‘filled’ with high-quality lives. And, of course, similarly issues will arise if one believes that there is some reason to ‘fill’ empty locations in space with high-quality lives, even if one grants that such reasons are less weighty than the analogous ones pertinent to time. Moreover, we may or may not have the same view of these issues, depending on whether or not we think that time and space are finite or infinite.

I mention these issues, only to leave them aside. They are a reminder – as if any were needed! – of the very preliminary nature of this work.

In Rethinking the Good, I introduced a position I called the Capped Model of Moral Ideals (Temkin Reference Temkin2012, 328–350). On a Capped Model, in certain contexts of comparison, there may be an upper limit on how good a given outcome may be regarding any particular ideal and, similarly, an upper limit on how good an outcome can be, all things considered. I acknowledged that the Capped Model faced a host of serious objections, and that there were a large number of unresolved questions regarding it. Even so, I argued that we needed something like the Capped Model in order to capture the anti-additive-aggregationist views that underlie certain of the judgments that most people hold about how different outcomes compare. Thus, for example, in the Repugnant Conclusion, most people believe, in accordance with a Capped Model, that no matter how many people there may be in an outcome where everyone who exists have lives that are barely worth living, the value of the well-being in that outcome will never rise to the point that it outweighs the value of the well-being in an alternative outcome where 10 billion people exist, all of whom have extremely high-quality lives. And likewise, most people believe that the all things considered value of the former outcome will never exceed that of the latter.

I cannot repeat, here, my lengthy discussion of the Capped Model in Rethinking the Good. But I note that much of this article is implicitly reflecting different views as to how best to develop the Capped Model. For example, I am suggesting that we should not have a single cap for how good an outcome may be regarding well-being over the course of time, but rather that we must have separate caps for each ‘substantial’ period of time. Regarding space, however, I am tempted to the view that a single cap that covers all of space at any given time period might be appropriate. On the other hand, if we do have different caps for different regions of space, perhaps the levels of those caps will be lower than the levels of the caps for time, or perhaps the levels of the caps for space may vary depending on how full other spatial regions already are.

Many of the unresolved issues in this paper are connected with unresolved issues regarding the Capped Model. Correspondingly, thinking further about what to say about the kinds of cases presented in this article may help illuminate how we should think about the Capped Model. By the same token, in the Rawlsian spirit of reflective equilibrium (Rawls Reference Rawls1971),Footnote ¹⁸ thinking further about how best to understand and interpret the Capped Model may help inform our judgments – sometimes confirming and sometimes leading us to revise our intuitions – about the sorts of cases this article considers.

Unfortunately, I cannot explore the Capped Model here. But let me point out just one way in which thinking more about the Capped Model might help illuminate our thinking about the kinds of issues this article raises. Regarding well-being, I believe that for each time period, we will need a different cap – a different upper limit – on how good an outcome can be regarding well-being, that will largely be a function of the (well-being) level of the best-off people in that outcome, and the number of people at that level. So, for example, for any given time period, the upper level of how good an outcome can be, regarding well-being, when everyone has a life that is barely worth living will be much lower than the upper level of how good an outcome can be, regarding well-being, when everyone has a life that is well worth living. More generally, I believe that the higher quality lives a population possesses, the higher cap there may be on how good an outcome can be regarding well-being.

Let us suppose that if people have a very high quality of life, k, the upper level on how good an outcome can be when everyone in a given time period has lives of quality k, is n. Let us further suppose that if people have a very high quality of life, k+x where x is a positive number, the upper level on how good an outcome can be when everyone in a given time period has lives of quality k+x, is n+y. If x is a relatively small, we can assume that the difference between n and n+y won’t be too significant. Next, suppose that a population of 10 billion people is sufficiently great, that if 10 billion people are all at level k, the value of that outcome will be very close to the upper limit for that outcome, namely n. And likewise, suppose that a population of 10 billion people is sufficiently great, that if 10 billion people are all at level k+x, the value of that outcome will be very close to the upper limit for that outcome, namely n+y.

By hypothesis, it will then be the case that the value of the outcome in which 10 billion people are at level k+x, will be greater than the value of the outcome in which 10 billion people are at level k, but not much greater. By the same token, given the nature of the Capped Model, since, by hypothesis, a population of 10 billion is enough to get a population very near the upper limit for how good that outcome can be regarding well-being (depending on the levels of those 10 billion people), it follows that if 20 billion people were at level k+x, during the very same time period, rather than only 10 billion people, that would improve the overall value of the outcome regarding well-being slightly, but only slightly. And the same is true, of course, if 20 billion people were at level k, during the very same time period, rather than only 10 billion people. That, too, would improve the overall value of the outcome regarding well-being slightly, but only slightly. Since, by hypothesis, we have assumed that the overall value of n+y is not much higher than the overall value of n, it follows that having 20 billion people at level k+x, during any given time period, will be not much higher than having 20 billion people at level k, during any given time period.

Now suppose that the only people living in the universe would be the 20 billion people at level k. But that they could either all live during the same time period, in which case the total value of well-being in that outcome would be very close to n, or there could be 10 billion people living in each of two distinct time periods, each living at level k. In that case, there would be two distinct time periods during which the value of the well-being would be nearly n. Even if one is not a simple additive aggregationist for determining the value of an outcome – as one won’t be if one adopts a Capped Model – it seems, given the considerations of this article, that there might be strong reason to attach significantly greater value to the outcome in which the 20 billion people were spread out over two distinct time periods, than to the outcome in which all 20 billion people lived during the same time period.

If this reasoning is all correct, then we might believe that there is strong reason to attach greater value to the outcome in which 20 billion people are all at level k, but spread out over two distinct time periods, than to the outcome in which all 20 billion people are at level k+x, but they all live during the same time period. But notice, we might suppose that it would be the very same people in the two different outcomes. In that case, we would be valuing one outcome more highly than another, because of how the well-being in that outcome is distributed temporally, even though the presumably better outcome would be worse for everyone who ever lived! This is because, in the supposedly better outcome, everyone would be at level k, while in the supposedly worse outcome, everyone would be better off, at the slightly higher level k+x.

I have argued at length, elsewhere, that there are a host of impersonal ideals – ideals that have value beyond the extent to which they are good or bad for people (Temkin Reference Temkin1993a, Reference Temkin, Frey and Morris1993b, Reference Temkin, Clayton and Williams2000, Reference Temkin2003a, Reference Temkin2003b, Reference Temkin2003c). Such ideals open up the distinct possibility that one outcome could be worse than another, even though there is no one for whom it is worse and, similarly, that one outcome could be better than another, even though there is no one for whom it is better. Clearly, if the way in which an outcome’s well-being is distributed in time or space can be relevant to our assessment of the overall value of that outcome, beyond the extent to which it affects the quality of lives (the experiences and non-time-related and non-space-related goods) of the sentient beings in that outcome, then the relevant temporal and spatial factors are impersonal in nature. This is an important but, on reflection, not surprising result.

Earlier in this article, I considered some examples that led me to conclude that, for certain cases and contexts, I favor people over times, and times over space. At the time, the reader may have been puzzled as to why I qualified my remarks in the way that I did. The preceding explains why. While there are some cases where I clearly favor how well-being is distributed across people, over how well-being is distributed across time, in some cases, such as the one I have just discussed, my judgment goes the other way. Here, as elsewhere, morality is enormously complex, and there are few, if any, simple principles that hold uniformly, and without exception or qualification, across all cases.