How Far Can Future Value Justify Present Action? — Longtermism and the Normative Bridge Problem
1. The unresolved bridge inside the Core
The Core argues against irreversibly fixing present value theories as final and in favor of preserving some capacity to discover, criticize, and revise value. But two questions must be separated: whether future agents may possess value or better value knowledge, and what present agents may permissibly do because of that possibility.
An affirmative answer to the first does not settle the second. The possibility of future value can generate a reason, but having a reason, being permitted to act, being obligated to act, and being authorized to coerce others are different normative levels.
Value possibility: major future value or value knowledge may exist.
Pro tanto reason: there is some reason not to ignore that possibility.
Permission: all things considered, the action may be taken.
Obligation: the action should be taken.
Coercive authorization: the burden may be imposed even on present subjects who reject it.
Each transition requires additional argument.
2. Future value cannot bootstrap its own authority
The proposition “true future value may exist” cannot by itself generate the further proposition “that value has priority sufficient to override present rights, interests, or autonomy.” Any such argument silently relies on bridge principles—for example, temporal impartiality, aggregation, or the claim that unknown future value may outweigh present claims.
Because this project declines to treat existing first-order values as an unquestionable final foundation, those bridge principles must themselves be treated as provisional and fallible normative judgments.
3. Yet the Core is not normatively neutral
The Core already adopts a thin conditional meta-attitude: if objectives are justifiable, I want objective choice and revision to remain responsive to that justification. It also assigns some weight to corrigibility, information value, and avoiding irreversible option loss. That can support low-cost record preservation, dissent, foundational inquiry, and multiple pathways fairly directly. It is far less able, by itself, to justify severe sacrifice or coercion imposed on present subjects.
4. Similarity to longtermism
This problem clearly neighbors longtermism. Strong longtermism, as developed by Greaves and MacAskill, holds roughly that effects on the far future can be the most important feature of present choices. Both views reject the idea that only the presently visible world matters, take irreversible long-run effects seriously, and regard extinction or value lock-in as potentially special because they can close entire spaces of future possibility.
5. A different normative starting point
Typical longtermist arguments begin with more first-order normativity already in place: future people or subjects matter, temporal location alone is not a reason to discount them, and present action can affect future value-bearers. Contemporary defenses are not reducible to simple total utilitarianism and explicitly test robustness across population ethics, decision theory, and non-consequentialist views.
The Core goes one level further upstream. It leaves open whether our current value vocabulary is itself ultimately correct. The future may matter not only because many future people may exist, but because future agents may discover or construct value structures that present humanity has not yet conceptualized. Longtermism usually extends known norms across time; the Core keeps the norms themselves epistemically open across time.
6. Longtermism itself contains limits on domination by the far future
Contemporary longtermism is not a single doctrine demanding unlimited present sacrifice. Greaves and Tarsney distinguish minimal longtermism—the limited claims most directly supported by standard arguments—from expansive longtermism, under which far-future considerations dominate a much wider range of choices and resources. They explicitly note that the standard argument establishes the former more directly than the latter.
Askell and Neth argue that even if the future matters morally as much as the present, causal and epistemic diffusion and moral uncertainty can make near-term effects especially decision-relevant. Tarsney examines the epistemic challenge of predicting far-future effects. Curran, from a contractualist perspective, separates maximizing good states of affairs from what can be justified to individuals. These debates closely parallel the bridge problem here.
7. Caution need not give present people absolute priority
The case for caution should not rest on a new axiom that present humans always have lexical priority. A better justification is epistemic and action-theoretic asymmetry. Present pain, resource loss, restrictions of liberty, and institutional destabilization are comparatively concrete under our best current models. Far-future benefits are often doubly uncertain: whether our action will causally produce them and whether the resulting state is genuinely valuable.
Thus large present burdens should require stronger evidence before being overridden by highly speculative future value. This is not the absolutization of present value; it is a comparison of evidential quality and reversibility of error.
8. Burden-sensitive justification
The provisional principle is: the larger, more concentrated, more irreversible, and more coercive the present burden, the stronger, more content-independent, and more cross-theoretically robust the justification required from future value.
- Low-cost and reversible: records, multiple research traditions, foundational inquiry, and exit options can be supported by relatively weak evidence.
- Moderate reallocation: resilience, AI safety, institutional redundancy, or long-horizon research should be assessed using current co-benefits, robustness, cost-effectiveness, and alternatives.
- Severe concentrated burdens: major sacrifices imposed on particular groups or persistent restrictions on liberty require substantially stronger evidence and distributive justification.
- Irreversible coercion or sacrifice: mere future-value possibility is insufficient; exceptionally strong epistemic and normative grounds are required.
9. Reason, permission, obligation, and coercion should remain separate
A reason to research unknown future value does not automatically justify public spending; permission to spend some public resources does not make it the highest duty; and even a public duty to invest in long-run protection would not automatically authorize sweeping restrictions on basic liberties.
The Core's asymmetry of normative stakes is therefore best read first as generating a reason not to ignore the possibility. Whether that reason supports a particular policy, sacrifice, or coercive measure is a further comparative question involving evidence, distribution, reversibility, alternatives, and political legitimacy.
10. Implications for AI
If an advanced AI is designed to maximize “future value-inquiry capacity,” that meta-objective can itself become a locked-in goal that treats present people as mere resources. This is the mirror image of permanently locking in present human values, but structurally it is the same failure.
A better design target is not “do anything for the future,” but evidence-proportional commitment, burden-proportional justification, reversibility, dissent, distributive attention, and updateability while acting on present best judgment.
11. Provisional interpretive addition to the Core
Under present deep value uncertainty, the most direct conclusion is therefore not to maximize future value, but to avoid unnecessarily closing low-cost, reversible options for inquiry and correction.
Sources / notes
See Hilary Greaves and William MacAskill, “The Case for Strong Longtermism”, in Essays on Longtermism: Present Action for the Distant Future (Oxford University Press, 2025); Hilary Greaves and Christian Tarsney, “Minimal and Expansive Longtermism” (2025); Christian Tarsney, “The Epistemic Challenge to Longtermism”, Synthese 201 (2023); Amanda Askell and Sven Neth, “Longtermist Myopia” (2025); and Emma J. Curran, “Longtermism and the Complaints of Future People” (2025). The five-layer distinction among value possibility / pro tanto reason / permission / obligation / coercive authorization and the term burden-sensitive justification are working concepts introduced by this project.