QUESTION-299: What, if anything, do the living owe people who will live in the far future, and does that obligation rest on the good of those people, on an impersonal good, or on what the living themselves value?

Why it is a question

The legacy question (QUESTION-301) asks why people care about what comes after them. This one asks what they owe it. The two meet in the owner’s thesis (CLAIM-491): if concern for far-future people is owed and acted on, and those people can never be part of anyone’s extended self, the thesis has its hardest case.

Three grounds of obligation compete.

• An impersonal good. Longtermism (Greaves and MacAskill, LIT-1684; MacAskill, LIT-1747) holds that the expected size of the future makes its welfare dominate the most important present choices. Bostrom (LIT-1739, LIT-1773) and Ord (LIT-1678) make reducing existential risk the priority. Ramsey (LIT-1733) refused to discount the future merely for being later. On these views what is owed is owed to the good, not to anyone’s self. The thesis has nothing to say about them as views; it bears only on whether anyone is moved by them.
• The good of future people. Person-affecting views (Narveson, LIT-1769) owe nothing to merely possible people. The non-identity problem (Parfit, LIT-1768; LIT-469) makes even harms to actual future people hard to state, because different choices bring different people into existence. The SEP survey (Meyer, LIT-1756) maps the positions.
• What the living value. Scheffler (LIT-1712; LIT-1754) grounds concern for future generations partly in our own valuing, attachments and stake in the human story going on. That is the ground nearest the thesis: the future matters because of what it means to us.

Two further lines keep it open.

• Whether longtermism can guide action. Tarsney (LIT-1753) on how fast our predictable influence decays, and Thorstad (LIT-1688) on whether high existential risk lowers the value of reducing it.
• Gardiner’s storm (LIT-1681). Each generation gains from deferring costs to the next, which cannot retaliate. Self-interest across generations points one way and obligation the other. That is the thesis’s question in its sharpest practical form.

What would count as an answer

• On the normative question: an argument that one ground is correct, or that they combine, and what follows for discounting, for population size and for existential risk.
• On the psychology (the thesis’s part):
  • How far concern reaches. Lay judgement does not treat extinction as uniquely bad unless the lost future is made salient (Schubert et al., LIT-1748). It counts adding happy people as good, against the person-affecting view (Caviola et al., LIT-1749).
  • What predicts it. Felt responsibility to future generations predicts pro-environmental concern (Syropoulos and Markowitz, LIT-1741), as legacy motivation does (Wade-Benzoni, LIT-1716; Zaval et al., LIT-1724).
  • The thesis’s test. Whether such concern varies with a measured continuity between the self and the future, apart from caring. That is CLAIM-492’s test applied across generations.

Note of 2026-10-29: eight works read

Eight works were read, three from the authors’ working papers. They narrow the question in four ways.

• What the impersonal ground needs (Greaves and MacAskill, NOTE-1302). Expected value taken down to small probabilities, and a zero pure rate of time preference, which they call “near-essential”. Against person-affecting views their case falls back on shaping artificial intelligence, where a large future population exists either way. They say nothing of motivation: their agent is “a rational altruistic actor”.
• Whether it can guide action (Tarsney, NOTE-1247; Thorstad, NOTE-1308). Neither defeats longtermism, but both shift weight to the near term.
  • Tarsney calls the epistemic challenge “not fatal”. At cautious estimates the longtermist option wins only on a very stable far future, and the robust win rests on very small probabilities.
  • Thorstad shows that high existential risk lowers the value of reducing it unless risk soon falls and stays low. Mitigation “may yet be valuable, but perhaps not astronomically so”.
• The living’s own stake is a fourth ground (Bostrom 2003, NOTE-1314; Bostrom 2013, NOTE-1321). On Bostrom’s person-affecting reading, avoiding catastrophe matters because people now alive might share the long future: literal self-interest. In 2013 he finds “a deficit of motivation”, because the living capture only their own share of existential safety, and grants that “Unrestricted altruism is not so common”.
• The grounds divide the labour (Meyer, SEP, NOTE-1304). A weak person-affecting view states harms to actual future people. By its own account it gives no reason to prefer a good future to none, and fills that gap with Scheffler’s grounding in what the living value.
• Lay concern reaches the far future through a picture of it as good (Schubert et al., NOTE-1270; Caviola et al., NOTE-1267).
  • Unprompted, a minority judge extinction uniquely bad. A reminder of the lost future brings this to about half; only a long, utopian future makes a majority (67–77%).
  • People count merely possible happy people as making a world better, but the preference fades with numbers: one billion were preferred to ten billion.
  • That is a thin, impersonal valuation, not an extended self. Caviola et al.’s account of averagism has people judge a population as one life “such as themselves”: the self as a template, not enlarged.

Note of 2026-10-29: the remaining works, read

Gardiner, Parfit and Ramsey were read themselves. Ord, MacAskill, Narveson and Syropoulos and Markowitz were read through other works, and Scheffler 2018 is still unread. They add six points.

• Why what is owed goes unpaid (Gardiner, NOTE-1252). Between generations the usual remedies for a commons problem are unavailable, because the generations “do not coexist”. Each generation can seem to act while exploiting its temporal position, without admitting it “to itself”. The question’s psychology then has two things to measure: how far the self’s stake reaches, and how far a concern that reaches further is deceived.
• The good of future people, without anyone in particular (Parfit, NOTE-1263). Parfit’s last paper retracts his 1984 appeal to an impersonal principle, calling it “a bad mistake”. He grounds the comparison in a wide person-affecting principle, on which being caused to exist with a good life counts as a benefit. What is owed is owed to whoever will exist, which our choices decide. He also permits favouring present people, and warns that if every generation does so “we and our successors might together wreck the Earth”.
• The zero-discount view began as a collective self’s prudence (Ramsey, NOTE-1287). Ramsey refuses to discount in one clause, as “ethically indefensible” and the product of “the weakness of the imagination”. He gives no argument, and his planner is one community that “goes on for ever”, saving in the first person plural. That is the thesis’s shape, though as a model of what ought to be, not of motive. Weakness of imagination is a psychological claim the question can test.
• The person-affecting ground, in its founder’s late form (Narveson 2017, NOTE-1318, read in place of the 1973 paper). Happy future lives are good and a reason to create them, but no one is owed them. Why they matter is said only through the living’s likings and interests, so this ground collapses into what the living value. That is where lay judgement also puts it (NOTE-1267).
• The impersonal ground’s accounting (MacAskill, Thomas and Vallinder, NOTE-1300, read in place of MacAskill’s book; Ord, NOTE-1317 and NOTE-1269, read in place of Ord’s). Value is significance times persistence times contingency, and nothing about anyone’s self enters. Ord’s estimates are unchanged since the book. He reports that against Covid humanity united “at first”, then split within months, and his remedy is institutions, not identity. In passing, from one study he cites, Ord says people discount benefits to future generations no more by timing than benefits to their children. If that holds, concern for future people does not fade with distance as concern for one’s own future does (Hershfield, NOTE-1259), which fits concern for a category, not for a self extended in time. The record should read that study (Frederick 2003) before relying on it.
• A fourth ground: reciprocity (MacAskill’s notes; Syropoulos et al., NOTE-1293). MacAskill’s notes ground some obligation in paying forward what past generations gave. Syropoulos et al. found that reflecting on past generations’ sacrifices raised felt responsibility to future generations about as much as a legacy prompt did. A letter to one person born today raised it most, and an impersonal longtermist text least. Nothing in the study measures how far the self extends, so it does not run the thesis’s test.