Sunday, 21 April 2013

Manipulating the distribution of lifespans

Suppose that you have the power to alter people's lifespans, but only on a statistical basis, not person by person. Perhaps you can do this through genetic engineering, or through putting something in the water supply. (As is normal in philosophical thought experiments, we shall not worry about practicalities.) Furthermore, you cannot ask people's permission before acting, and you are the only person in the world who can take this action.

To be precise, you can take, or refrain from taking, action that will have the following effects. The effects are only on offer as a single package: you cannot pick and choose selected effects.

A. The mean of the distribution of life expectancies for the affected group will rise.

B. The dispersion of that distribution will fall, but the overall shape of the distribution will remain unchanged. If, for example, the distribution was normal, it will remain normal, but with a lower standard deviation. The fall in the dispersion will be great enough to ensure that despite the increase in the mean, there is some age, above which the high-lifespan tail of the new distribution will fall below the high-lifespan tail of the old distribution. (Both distributions are to be given in terms of proportions of the populations to which they apply, so as to avoid issues about the effects of a higher mean on the size of the population.)

C. The mean of the distribution of years of ill health or disability at the end of life, and the overall shape and the dispersion of that distribution, will remain unchanged. Note that this effect is given in terms of years, not in terms of proportion of life. So the number of years of ill health or disability remains the same, even if lifespan increases.

D. Any correlation between position in the distribution of lifespans, and position in the distribution of years of ill health or disability, remains unchanged.

Thus, average prospects will improve: longer life, with no more years of ill health or disability at the end of life. But there is a tail of the distribution of lifespans, in which prospects are made worse by the change.

Now we come to the questions.

1. Should you take this action, if it is to apply to all people already born or currently in the womb, but not to anyone else?

2. Should you take this action, if it is to apply to all people who have not yet been conceived, but not to anyone else?

3. Should you take this action, if it is to apply to all people already born, currently in the womb, or not yet conceived?

Two points should be noted, before we consider answers.

The first point is that you are confronted with a single choice: take the action, in accordance with the terms of whichever one of 1., 2., and 3. is in force, or do not take the action at all. You are told which of the three is in force. You do not get to choose which one is in force.

The second point is that in drawing a line at conception, I do not mean to take a stance on whether embryos are people. I simply want to recognize the difference between entities that already have a determined genetic endowment, and potential entities that do not. The significance of genetic endowment, is that it is likely to have quite a lot to do with prospects for lifespan and for infirmity in old age.

The case against taking action in 1. looks quite strong. Given that genetic endowment and lifestyle to date have a substantial influence on lifespan, there would be some group of people, ones with good genetic endowments and healthy lifestyles, who constituted a modest proportion of the population, such that we were aware of the criteria for membership of that group, and such that we could say that taking the action would shift the odds against members of that group, to an extent that one would not regard as trivial. Kantian objections to the use of people as means to other people's ends come to mind. The other people in question would be the people outside the group.

The fact that the members of the group would be people who were already well off, in that they would be people at the high end of the distribution of lifespans, hardly seems to be a satisfactory response. Imposing reductions in lifespans seems to be rather more fundamental than imposing high tax rates on high incomes.

My initial feeling is that the obvious utilitarian case for taking action in 1. is outweighed by considerations such as these.

One objection to this view, would be that a decision not to take the action would itself be a decision to act: that is, that refraining from action would not be abstention, but a positive choice for the other side, and that one would be just as responsible for that as one would be for a choice to take the action, and responsible in the same way. Then a choice not to act would amount to deliberately disadvantaging those who were not in the group with long life expectancies. I shall not explore this view here, but it is a view that could be argued.

The case against taking action in 2. looks much weaker. People as yet unconceived have neither any particular genetic endowment, nor any particular lifestyle. It is tempting to say that there would be, in the future population, some members of a specifiable group of people (the group of people with good genes and healthy lifestyles), who would be worse off than they would have been, had the action not been taken. But that objection would not be rightly phrased. The word "they" would have no referent. The complaint that some members of a specifiable group would be worse off, once re-phrased to remove this difficulty, can amount to no more than a complaint that the distribution of lifespans, for the whole population, would be different. Given that, the utilitarian case seems to be a good one. In 2., a choice to take the action would be the right choice.

It might seem that the word "they" would have a referent, at least in the next few generations, if there was a strong hereditary element to longevity. It could refer to the descendants of people who were currently alive and who had good genes. But I doubt that this would be enough to create a referent. Since we are concerned with people not yet conceived, there would only be potential descendants. Any particular person currently alive and with good genes might not have any descendants, even though it would be very probable that some people or other, drawn from the set of those currently alive and with good genes, would have descendants.

There is a parallel with John Rawls's veil of ignorance. It is a veil of total ignorance. The people behind it have no particular characteristics, until they are dropped into the society that they have designed. So they can only sensibly think about overall distributions. There is a sense in which they cannot take up arms on behalf of a particular group on the basis that some arrangement would do an injustice to its members, even though criteria for membership of the group may be perfectly clear, because no-one has, at the time of design, characteristics that would determine whether he or she was a member. "Its members" lacks a referent. To make the parallel closer, we can imagine Rawls's deliberators thinking about possible changes to distributions that already apply in an actual society, a society which all the designers will join as a complete replacement population, in roles and with characteristics that will be allocated at random, once all existing members of the society have died.

I shall not reach any conclusion on the third possibility, applying the change to current and future people. The key question is this. If the future population will be, in total over the centuries, very much larger than the current population, can the utilitarian case for action outweigh the case against action that was set out in relation to 1.?

There is one more complication. Is it in the interests of a person to have descendants who enjoy long lives? If it is, then that would influence one's thoughts in relation to 3. Currently living people with good genes might lose some of their own lifespans. But suppose that the beneficial influence of good genes on lifespan was not passed down the generations to any significant extent, so that, for example, the great-grandchild of someone with genes that substantially improved prospects had no better chance than the average in the population of having genes that conferred such good prospects. Then the good genes of the current generation would not tend to keep a significant proportion of their descendants in the group that would potentially lose from the action. Then the members of the current generation with good genes might have self-regarding reason to favour the action.


  1. Thank you for this thought-provoking thought experiment. These are some of the questions it raises in my mind.

    Would the Kantian objection apply within the society imagined by Rawls? Is it valid to see people as being made use of for another's ends if they had an equal chance of benefitting if the positions had been reversed? Or is this co-operation?

    Should we view society through the 'veil of ignorance' or take into account personal circumstances?

    From the 'veil of ignorance' point of view, the changes to lifespan distribution apply equally to all and Rawls' deliberators would presumably have voted for them. Whether the changes are beneficial or adverse to any individual will depend on his or her personal circumstances, but this would then be a matter of being lucky or unlucky within a basically fair system. Someone has to draw the short straw.

    On the other hand, if we consider personal circumstances, we can argue that the potentially long-lived group are adversely affected. But if we start doing this, where do we stop?
    a) The group in question only has a higher probability of longevity, based on current scientific knowledge; they do not have a guarantee of extra years. Is a probability enough to entitle them to special consideration?
    b) The group is statistical and the definition will be somewhat arbitrary. Any number of statistical groups could be defined. There will be issues with cut-off points.
    c) The definition of the group, and therefore its members, would change over time as science progresses. What if a longevity gene was to be discovered after the distribution of lifespans had been changed? Would its possessors have a valid claim for compensation?

  2. Many thanks for your questions, Isabel. I need to stop and think about them, but I shall post a proper response within the next day or two.

  3. Hello Isabel,

    In response to your first two paragraphs ("Would the Kantian objection apply ..." and "Should we view society ..."), my feeling is that Kant's way of looking at our interactions only works when we are determinate human beings, with particular lives and characteristics. So Kant would have nothing to say to people behind the veil, except that they should not construct a society in which people would get used merely as means. (The "merely" is important here, if we are to make rooms for Kant's rather unattractive comments on rights to persons akin to rights to things, in Metaphysics of Morals, Ak.6:276-284.)

    I read your two sentences, "Is it valid ... co-operation", as addressing two points, about how people behind the veil should design their future society.

    The first point is that the future society had better allow for contractual dealings, and Kant would have expected and welcomed that too. When you have a contract, each party may derive some benefit, a point that is neatly reflected in English law by the requirement for consideration (except when contracts are executed as deeds). But one may derive much more benefit than the other, and may in that sense make a big profit at the other's expense. If X buys something from Y very cheaply, and then sells it to Z for a much higher price, Y could have made the sale directly to Z, and taken the profit himself. Is that X using Y merely as a means? I do not think it is, given that Y voluntarily enters into the contract. Rather, I would see it as co-operation. But in any case, the remedy would be to urge people to negotiate contracts in a reasonable way, not to forbid the free negotiation of commercial contracts.

    The second point is that of whether, behind the veil, people should take a chance on being on the used end of relationships of use merely as means, given that they might well turn out to be on the user end. Rawls, with his focus on the position of the worst-off, would say that if they were rational, they would not take that chance - although his concern was with objectionable outcomes, such as having to work too hard for too little reward, rather than with the objectionable logic of some relationships. Kant would, I think, say not merely that they should not take that chance, but that they were being something close to inconsistent if they both did take it, and thought they were creating a society in which all examples of the species homo sapiens would be accorded the status of persons.

    [Continued in the next comment, because of the limit on the number of characters in each comment]

  4. [Continued from the previous comment, because of the limit on the number of characters in each comment]

    On your paragraph, "From the 'veil of ignorance' point of view ...", I see the position of the deliberators somewhat differently from you. Yes, they would vote for the less dispersed distribution with the higher mean, but I do not think one could identify anyone who would draw the short straw. There are two reasons for my view, and I feel more secure about the first one than about the second one, although the second one is a generalization from the first one.

    The first reason is that no-one behind the veil has any particular life expectancy, so no-one has a starting point from which they can be made worse off.

    The second reason is that it is not at all clear to me that we can attach sense to the idea of selecting a function that takes each identified deliberator to an identified person in the post-veil world. It is tempting to imagine that there would be such a function, because we can say, to our actual selves, "Imagine that we were all put behind the veil; would we design a society like the one we actually inhabit, or would we arrange things differently?". So we imagine taking ourselves behind the veil, one actual person to one deliberator. We then think that we could simply take the inverse of that one-one function, and have a one-one function from deliberators to people in a post-veil world. I think we may be fooling ourselves about the sense in a function from deliberators to post-veil people. In order to make the Rawls story work, we have to attach sense to the idea of some unspecified function or other drawn from the class of all one-one functions from deliberators to post-veil people. But that is not the same as attaching sense to the idea that we could select some particular function, drawn from that class, as the one that would in fact operate.

    On your final points, the basic question seems to be, has a particular person someone lost anything if the odds shift against her, when there is no guarantee that she has in fact been made worse off? Arguably, there never will be any fact as to whether she has been made worse off, because there never is a determinate outcome under the old odds, since the outcome, age at death, arises only under the new odds. At least, if there is a determinate outcome under the old odds, it is certainly unknowable in practice.

    It feels as though the answer to this basic question should not depend on the precise odds. On the other hand, some odds make it more comfortable to conclude that there has been a loss, and other odds make it more comfortable to conclude that there has not been a loss. Suppose that someone was a member of a well-defined group, half of which would live to 97 or more under the old distribution. Under the new distribution, 99% of that group will die by age 92. That looks dramatic enough to ground a complaint. Alternatively, suppose that under the new distribution, 45% of the group will live to 97 or more. That looks like a mild enough change to think that complaints would not be well-grounded.

  5. Many thanks for this detailed response. As usual with your posts, it repays multiple readings.

    Re my paragraph, "From the 'veil of ignorance' point of view ...", I was not thinking of the deliberators as the individuals who would be affected. I was trying to express a sense that in the post-veil world we would worry less about the impact of a policy on specifiable groups as the starting point would have been as fair as possible. I see now that this is very similar to your future world where specifiable groups do not yet exist.

    Both give us a starting point where everyone is notionally equal. If the odds of being impacted adversely by a policy are also equal, then I feel we consider the affected people as having been unlucky rather than having been treated unfairly. The more the odds shift, the more we may worry about securing equality of outcomes in order to mitigate the unfairness.