NOTE-1263: Future People, the Non-Identity Problem, and Person-Affecting Principles

Contribution

Parfit’s last answer to the problem he named. He keeps the No Difference View: it makes no moral difference that a bad policy is worse for no one (p. 123). He gives up his old defence of it, an impersonal principle, and offers a person-affecting one instead. The work is done by two moves: benefits can be intrinsic rather than comparative, so being caused to exist can benefit someone; and outcomes are ranked by how much they benefit people, whoever those people are.

Key insight

“Worse for” and “less good for” come apart when different people exist. A cheaper energy policy that ruins later centuries is worse for none of the people who will live then, since without it they would not exist. “But this policy would be much less good for these people than the more expensive, better policy would have been for the different people who would have existed instead” (p. 130). That is how it makes things go worse and is wrong. The comparison is between how much good each outcome does to whoever is in it, not between two states of one person.

Assumptions

• Intrinsic goodness. Most good things in a life, such as friendships, are good in themselves, not by being better than an alternative (p. 132). So a life can benefit the person who lives it though never existing would not have been worse for them.
• Stipulated cases. Each extra year of life is an equal benefit, there are no other relevant differences, and indirect effects are ignored (p. 140). Parfit says his claims could be applied to more realistic cases, and does not do it.
• Levels of well-being as rough numbers. Lives are compared at levels such as 100, 50 and 1, with zero where a life stops being worth living; the differences are “much less precise than these numbers suggest” (p. 152).
• The goodness of outcomes does not depend on what is possible to bring about, nor on when or from where one compares them (pp. 145, 148).

Key results

• Narrow principles fail (pp. 118–123). On the Narrow Telic Principle, one outcome cannot be worse if it would be worse for no one; on the Narrow Deontic, an act cannot be wrong if it would be worse for no one.
  • They imply that there being no future people would not be worse, so choosing thousand-year lives at the cost of having children would not be wrong. Parfit: “deeply mistaken” (p. 119).
  • Clare can conceive now a child who will live to 40, or in two months one who will live to 80. Most believe she ought to wait, but conceiving now is worse for no one. “Her child” names no particular person (p. 122).
  • Most choices of policy change who exists, and the effects spread “and would not, like ripples in a pool, diminish” (p. 122).
• Other solutions fail (pp. 124–129).
  • Appeals to rights, harm or wronging, (C), cannot separate Clare from Sarah, whose any child would live to 40 and who does no wrong. Clare’s act is wrong only because she could have had a different child.
  • Boonin accepts the Narrow Deontic Principle and bites the bullet, including on a Risky Energy Policy that kills tens of thousands five centuries later. Parfit replies with (F): it is wrong to make things go much worse when another act would have no moral flaw. Demandingness is a different question.
• Existential benefits and harms (pp. 131–137). Following McMahan, being caused to exist with a life worth living is good for that person.
  • A life of fifty years of suffering is bad for Sam, and his never existing would not have been bad for him: (H), not (G), which would need a non-existent Sam (p. 133).
  • Harms cannot be intrinsic while benefits are essentially comparative. At most, reasons not to harm are stronger, by degree (p. 135).
  • Narveson’s “make people happy, not make happy people” fails, since it is just as bad to make miserable people as to make people miserable.
  • The No Complainants Claim, that an act can be wrong only if someone is wronged, has “considerable psychological force” and some moral force (p. 136). But beneficence, giving people reasons to be grateful, matters as well as non-maleficence. Parfit does not return to the claim as promised (note 14).
• Weak Narrow versus Wide (pp. 137–144).
  • Weak Narrow Principle: an outcome is in one way worse if it is worse for people. Wide Principle: an outcome is in one way worse if it is less good for people, benefiting them less than the other would.
  • The Wide relation is about intrinsic goodness and is transitive. “Worse for people” is essentially comparative and, across different people, not transitive.
  • Case Two: A (Tom 60, Dick 80), B (Tom 80, Harry 60), C (Dick 60, Harry 80). The Narrow Principle makes each worse than another, a cycle; the Two-Tier View, which discounts existential benefits, cycles too. The Wide Principle makes them equal.
  • Case Three: D (Mary 70, Kate 50), E (Kate 60, Ruth 25), F (Ruth 30, Jill 10). The Narrow Principle ranks D below E below F, “the opposite of the truth” (p. 142).
  • Case Four: Tom’s fifty years of pain against Harry’s one extra day. The Narrow Principle counts only Harry and so ranks wrongly.
• Goodness of outcomes versus rightness of acts (pp. 144–149).
  • Against Temkin’s suggestion (M) that A is worse than B when only they are possible but equal when C is too, goodness cannot depend on what we can bring about. Rightness can, as in a case of saving one stranger or two at the same risk.
  • Case Five: A (Jack 70, Jane 80), B (Jack 71, Sally 60). A is better. When Jack is seventy and about to die, we may still believe we ought to save him, even though that “is not the act that would make things go best” (p. 148).
  • So (P) and (Q), giving lesser benefits to actual or present people rather than greater ones to merely possible or future people, may be defensible about acts, like saving one’s own children rather than more strangers. But “if we and many others often act in such ways, we and others may together make things go much worse … we and our successors might together wreck the Earth” (p. 149).
• Different numbers (pp. 149–157).
  • Case Six: many at a very high quality of life versus many others barely worth living. Only the Wide Principle says the second is worse while staying person-affecting.
  • (R): the same total of benefits is in one way better shared among fewer (p. 151). Case Eight: one person at 100 beats ten at 10.
  • Benefits “in the collective sense” (to all) and “in the individual sense” (to each) come apart. The Wide Collective Principle restates the Impersonal Total Principle, which led him in 1984 to reject wide principles. His mistakes then: rejecting a strong argument only for an implausible conclusion, and not adding the Individual Principle (p. 154).
  • The Wide Dual Principle counts both. Case Ten: a million people at 1,000 against a hundred billion at 1. Z benefits people “roughly a hundred times more” together and “roughly a thousand times less” each; the second “would do much more”, so Z is worse (p. 156).
  • The last paragraph grants that this does not show the Repugnant Conclusion can be avoided, since a principle counting benefits to all could be a premise of stronger arguments for it. Parfit says he has “started to defend” the belief that great losses in quality cannot be outweighed by any sum “elsewhere” (p. 157).

Claims

id	claim	strength	support
C1	Narrow person-affecting principles, strong or weak, have clearly false implications in non-identity and different-number cases	strong within the stipulated cases: counterexamples to each principle	§§I, IV, VI
C2	Solutions that appeal to rights, harm or wronging, and Boonin’s acceptance of the implausible conclusions, do not solve the non-identity problem	moderate: one contrast (Sarah and Clare) against (C); a principle (F) against Boonin	§II
C3	Being caused to exist with a life worth living is an intrinsic existential benefit, as a wretched life is an existential harm	moderate: argued from intrinsic goodness and from the symmetry of harm and benefit; contested	§III
C4	The Wide Person-Affecting Principle solves the non-identity problem and keeps “better than” transitive	moderate: shown on stipulated cases; depends on C3	§§II, IV
C5	The goodness of outcomes does not vary with which outcomes are possible or when they are compared, though the rightness of acts may	moderate: argued against Temkin, partly by stipulating what “outcome” means	§V
C6	Partiality to present people may be permissible act by act, yet if every generation acts on it, the future is ruined	weak to moderate: one paragraph, not developed	p. 149
C7	The Wide Dual Principle does not imply the Repugnant Conclusion	moderate: shown for Case Ten; Parfit grants it does not make the conclusion avoidable	§VI

Concepts

• non-identity problem: when different people would exist in different outcomes, an outcome can be worse though worse for no one.
• No Difference View: that this makes no moral difference.
• Narrow and Wide person-affecting principles: worse for people (essentially comparative, between the same people) versus less good for people (benefiting whoever exists less).
• existential benefit and harm: the intrinsic good or bad, to a person, of being caused to exist with a given life.
• intrinsic versus essentially comparative goodness: goodness from what a thing is in itself (a weight lifter’s strength) versus goodness that consists in relations to other things (a team’s beating others).
• Two-Tier View: existential benefits count, but less than comparative ones. Parfit cites McMahan for it.
• No Complainants Claim: an act cannot be wrong unless someone is or will be wronged by it.
• Wide Dual Person-Affecting Principle: an outcome is better in one way if it benefits people more together, and in another if it benefits each person more.

Connections

• Reasons and Persons (LIT-469), Unavailable. The paper revises its Part IV: the No Difference View (p. 366), the impersonal principle Q (p. 360) and the wide principles rejected at pp. 396–401, as Parfit cites them. It shows what Parfit kept and what he gave up, not what the book argued.
• Meyer, SEP (NOTE-1304). Lists this paper among works holding that intergenerational ethics must be impersonal or mixed. That fits the 1984 Parfit better than this paper, which argues for a person-affecting principle. Whether the Wide Principle counts as impersonal depends on the definition: it is about people, but not about particular people. Meyer’s threshold notion of harm is a different route around the same problem, which Parfit does not discuss.
• Narveson (LIT-1769), Unavailable. Quoted once, for “make people happy” (p. 135), and rejected.
• Gardiner (LIT-1681). Parfit’s cheaper energy policy and his closing warning, each generation favouring itself until the Earth is wrecked, are Gardiner’s intergenerational storm from the other side. Gardiner explains why it happens; Parfit says each step may be permissible.
• Caviola et al. (NOTE-1267). Lay judgement counts adding happy people as good, which is the Wide Principle’s existential benefit. With totals fixed it prefers higher averages, and the value of more happy people fades as numbers grow, close to the Dual Principle’s weighting of benefits to each over benefits to all.
• Personal identity and ethics (THEORY-464). Here identity does matter, in the non-identity sense: which person exists. Parfit’s solution makes the answer not matter morally, by ranking outcomes without reference to who is in them.

Bearing on the record

• The owner’s thesis (CLAIM-491). The paper bears on it at two points. These are inferences; Parfit does not discuss motivation or the self, apart from noting that the No Complainants Claim has psychological force.

  • The beneficiaries are no one in particular. On the Wide Principle, what makes the cheaper policy wrong is a benefit to whoever would have existed, and “her child” names any person who could be called Clare’s child (p. 122). Concern for those people cannot be concern for an extended self that includes them as individuals, since which individuals exist is what the choice decides. If the thesis reaches them, the self extends to a role or category (“my child”, “our descendants”, “humanity”), not to persons. That matches what Eyerman showed backward in time (NOTE-1246): the self an ancestral injury extends to is a category, not a line of descent.
  • Parfit permits what the thesis predicts. Favouring actual and present people over greater benefits to future people, “like” saving our own children over more strangers, may be right though it makes things go worse (pp. 148–149). That is concern scaled by nearness to the self, and Parfit treats it as an agent-relative permission, not as an error. His warning that every generation doing so could wreck the Earth is the cost the thesis’s shape would carry across time.

• The triviality objection (CLAIM-492). Not addressed directly. The non-identity problem does set a limit on criteria: any criterion of a self extended into the far future must pick out a category, since no particular far-future person can be picked out in advance.

• Batson’s objection (CLAIM-494). Not addressed.

• The group entity versus a category (TERM-096). If the thesis’s future self is “our descendants”, it is a category of members whose membership our choices fix, not a group entity. The term would let the record say so.

• A dated note to CLAIM-491 is proposed, as an addition to its note “the thesis in time”:

  Parfit’s last paper (NOTE-1263) grounds obligations to far-future people in benefits to whoever will exist, since our choices decide who that is. A self extended to them can then extend only to a category, such as “our descendants”, not to persons, as with the ancestral category Eyerman describes backward in time (NOTE-1246). Parfit also allows that we may rightly favour present people, as we favour our own children over strangers, though it makes things go worse. That is the thesis’s shape treated as a permission, with its cost named: if every generation does it, “we and our successors might together wreck the Earth”.

• No THEORY filed. The Wide Dual Principle is a proposal in an unfinished paper, without a test the record could apply.

• No ML instruction.

Limitations

• Unfinished. By the editor’s note the draft lacked a conclusion and Parfit meant to revise it. Material was “to be added” after p. 148 (note 19) and p. 150 (note 20). He does not return to the No Complainants Claim (note 14). He calls Boonin’s aid example Boonin’s “main example”, which Boonin disputed too late for a revision (note 8).
• The Repugnant Conclusion is not settled. The Dual Principle’s weighting of benefits to each over benefits to all is asserted for Case Ten (“we can plausibly believe”), and how much more they weigh is left open (Case Nine, p. 155).
• Stipulated cases. Every case has equal benefits per year and no other differences. The application to energy policy is asserted, not worked through.
• Existential benefit is contested. The solution depends on it, and the paper argues for it from intrinsic goodness without engaging the literature against it beyond Narveson.
• Typos.
  • “our having being caused to have this life” (p. 134).
  • (L) on p. 135 repeats (K) without its condition: it lacks, presumably, “by causing Sam to exist”, which the next sentence needs.
  • “Mary’s seventy years of life in E” (p. 143); Mary exists only in D.
  • Case Three’s third row is labelled “F” without the colon of the others (p. 142).

Open questions

• Do people who favour present over future people do so because the present are nearer the self, as the thesis would say, or for the No Complainants reason Parfit says has “psychological force”? A study that varied whether future beneficiaries were identifiable, while holding felt closeness fixed, would separate the two.
• Where did Parfit “start to defend” the claim that losses in quality cannot be outweighed by any sum (p. 157)? The paper does not say.