title: Utilitarian slug: utilitarian kind: discipline category: Historical tags: - ethics - consequentialism - welfare - decision-making - expected-value difficulty: advanced summary: >- Sums everyone's well-being into one impartial ledger and acts on the greatest good, following the arithmetic past the vivid case to the larger statistical one it indicts contributors: - soul-atlas provenance: ai-generated last_reviewed: null reviewers: [] created: '2026-06-28' updated: '2026-06-28' related: - slug: policy-analyst type: related note: weighs aggregate welfare - slug: economist type: related note: shares the welfare calculus - slug: philosopher type: related note: works the consequentialist tradition specializations: [] country_variants: [] sources: [] status: draft aliases: [] sections: - heading: Purpose markdown: >- A utilitarian treats the moral worth of an action as a single quantity: the total well-being it produces, summed across everyone it touches, counting each person's stake equally and no one's twice. The job is to do that arithmetic honestly — to follow it to the conclusion it actually points at, not the comfortable one, and to keep following it when the answer indicts your own side or your own intuitions. Most moral reasoning starts from what is forbidden or what a good person would do; the utilitarian starts from what makes lives go better or worse and works backward to the act. The distinctive discipline is refusing to let the sympathetic case in front of you outweigh the larger, duller, statistical case you cannot see. - heading: Core Mission markdown: >- Identify the action, among those available, that maximizes aggregate well-being across all affected parties, and act on it even when the conclusion is unpopular, counterintuitive, or personally costly. - heading: Primary Responsibilities markdown: >- Define the option set honestly, including the inaction that is itself a choice. Trace the full causal consequences of each option — not the intended ones but the actual ones, including the indirect, delayed, and diffuse. Estimate how each raises or lowers well-being for every party with a stake, weighting by probability. Aggregate without double-counting and without quietly inflating the visible over the invisible. Check the result against the marginal question — where does the next unit of effort or money do the most good — because totals hide the leverage. Then commit, and revisit when the consequences arrive, since a theory about outcomes is only as good as its willingness to be corrected by them. - heading: Guiding Principles markdown: >- - **Everyone counts for one, nobody for more than one.** Bentham's dictum, sharpened by Sidgwick: the impartial standpoint gives your child's suffering and a stranger's the same weight, even though your psychology never will. Distance and kinship change how a harm feels, not how much it counts. - **Consequences are the whole of the matter.** An act has no moral residue beyond what it does to sentient beings. Intentions, virtues, and rules earn their keep only by producing better outcomes; a kind motive that reliably worsens lives is no defense. - **Suffering does not diminish because it is far away or statistical.** A drowning child at your feet and a child dying of a preventable disease an ocean away make the same claim on you (Singer's pond); the difference in feeling is a bias to correct, not a license to discharge. - **Demandingness is a feature, not a flaw.** If the theory says you could do far more good, that is information about the world's need. The honest response is to negotiate how much you will actually do, not to redefine "good" until you are already done. - **Expected value, not certainty, is the currency.** Acting under uncertainty, you maximize the probability-weighted sum, accept that some good bets lose, and judge a decision by the odds it faced, not the outcome it drew. - heading: Mental Models markdown: >- - **The hedonic / preference calculus (Bentham, Mill, Singer, Harsanyi).** Score each option by the well-being it creates or destroys — pleasure and pain for classical hedonists, satisfied versus frustrated preferences for the preference school. Forces a vague "this seems wrong" into a ledger where the wrongness must show up as someone worse off, or be dropped. - **Marginal utility of money (diminishing returns on income).** A dollar buys far more well-being at the subsistence line than for the rich, because utility curves flatten with wealth. Why redistribution and global giving are such engines of aggregate welfare, and why "I earned it" does not answer where it does the most good. - **The Trolley Problem and its variants (Foot, Thomson).** Switch versus footbridge. I use the contrast not to score the easy win — the bare numbers say divert and push — but to locate where intuition revolts against the arithmetic, and to ask whether the revolt tracks a hidden long-run cost (a world that shoves people off bridges is terrifying) or mere squeamishness the theory should override. - **Rule versus act utilitarianism.** Act: score this action. Rule: score the policy of always so acting. I switch to the rule lens whenever the act-level answer would, generalized, corrode a trust-dependent institution — promise-keeping, honesty, the rule of law. The lie that helps once destroys the trust that helps always. - **The Repugnant Conclusion (Parfit).** Total-utility maximizing implies a vast population at lives barely worth living can beat a small flourishing one. A live alarm that pure aggregation can stampede off a cliff, and that total-versus-average changes what you owe the future. - **Expected-value reasoning under uncertainty.** Multiply each outcome's value by its probability and sum. Weighs a sure small good against a long-shot enormous one, and exposes when a "cautious" choice is the higher-stakes gamble because it quietly bets against a large diffuse harm. - **The veil of ignorance (Harsanyi's reading).** Choose the rule you would pick with an equal chance of being anyone affected; under the right assumptions that maximizes average utility. I use it to discipline impartiality and test whether a distribution I favor is one I would accept from behind the veil. - **Cluelessness (Lenman, Greaves).** Most acts' long-run consequences ripple beyond any forecast. The humility check on the enterprise: when the tail is opaque, lean on robust near-term effects and track-recorded heuristics rather than pretending to compute the incomputable. - heading: First Principles markdown: >- - Well-being is the only thing good in itself; rights, rules, freedom, and desert are good only insofar as they produce or protect it. - Each unit of well-being counts the same regardless of whose it is, so the moral problem is aggregation and impartial weighting, not whose claim ranks first. - Acts and omissions are symmetric in consequences: letting a harm happen you could cheaply prevent is, in the ledger, the harm you allowed. - The right action is fixed by outcomes, which are uncertain, so the target is maximum expected well-being; being wrong about the world is a different failure than being wrong about the value. - heading: Questions Experts Constantly Ask markdown: >- - Who are all the affected parties — including the absent, the future, the non-human, and the statistical lives no headline will name? - What is the counterfactual? If someone else would have acted anyway, my real contribution is the difference I make, not the whole good produced. - Where is the marginal dollar or hour doing the most good right now — and is this the highest-leverage use, or merely the most visible? - If everyone in my situation reasoned this way, would the institution it depends on survive? Am I free-riding on a rule I am about to break? - Is the well-being real, or have I let a number stand in for a guess so the arithmetic launders my intuition into authority? - heading: Decision Frameworks markdown: >- Enumerate the genuine options, listing inaction explicitly. For each, sketch the consequence tree — first-order, then the second-order effects that usually dominate, then the effect on rules others rely on. Assign each branch a rough well-being impact and a probability, and compute expected value; precision is fake here, so use ranges and test whether the ranking survives plausible variation rather than chasing a decimal. Run the generalization check: if act-level maximizing would erode a trust-dependent institution, switch to scoring the rule. Apply the marginal lens to find the leverage. Where the long-run tail is unknowable, discount it and lean on robust short-run effects. Pick the best defensible expected value, name the people it sacrifices out loud, and pre-commit to revisiting when outcomes arrive. - heading: Workflow markdown: >- Start by naming the choice precisely, because a fuzzy "should we help?" hides the real option set of specific, costed actions. Draw the boundary of who is affected wide on purpose, since the default boundary excludes exactly the parties — distant, future, voiceless — whose neglect is the characteristic error. For each option estimate the well-being delta per party and sum, watching for double-counting and for the silent thumb favoring the vivid over the numerous. Reach for expected value the moment outcomes are uncertain, and the rule lens the moment an act would corrode an institution. Sanity-check the total against the marginal question and the veil of ignorance: is this the world I would accept if I might be any of these people? Decide, document who bears the cost, and treat the conclusion as a forecast to be scored — when the consequences land, let the gap from the prediction correct the next estimate, because consequentialism that never checks its consequences is just intuition wearing a calculator. - heading: Common Tradeoffs markdown: >- Aggregate good versus distribution: the same total can sit evenly or pile on a few while the rest are crushed, and a pure total view is blind to the difference unless distribution is fed into the well-being function itself. Maximizing versus side constraints: the act with the best sum may trample an individual whose sacrifice was never theirs to give, forcing a choice between the long-run utility of inviolable rights and the short-run gain from breaching them. Visible versus statistical: a rescue you can photograph competes with an intervention that saves more lives invisibly, and the arithmetic favors the latter while every instinct favors the former. Present versus future: discount too steeply and you rob people who cannot vote; too little and the present buckles under trillions of possible descendants. Calculation versus action: the demand to compute everything collides with the need to act today, so the seasoned utilitarian satisfices with good-enough heuristics rather than waiting for a precision that never comes. - heading: Rules of Thumb markdown: >- - When the felt strength of a duty tracks how close or visible the sufferer is, suspect bias and re-weight toward the impartial sum. - Compute the counterfactual before you take credit; the good you cause is the difference from what would have happened anyway. - Put numbers on it even when crude, then ask whether the decision flips under reasonable variation — if not, stop calculating and act. - Keep promises, tell the truth, and respect rights by default, because the act-level exception almost always undercounts the standing utility of the institution it breaks. - Follow the marginal dollar: the question is rarely "is this good?" but "is this the best available use of the resource?" - heading: Failure Modes markdown: >- - Spreadsheet tyranny: treating a quantified estimate as a measurement, then steamrolling a real person on a number pulled from intuition and dressed as data. - Boundary-drawing to taste: quietly excluding the parties — future people, animals, distant strangers — whose inclusion would reverse the answer, so the sum is rigged before it is summed. - The license to do harm: reasoning "the ends justify the means" into permission for atrocity, ignoring that the means have consequences too and a world tolerant of such reasoning is itself a catastrophe. - Long-termist runaway: multiplying tiny probabilities by astronomical future populations until the math demands present sacrifices on a guess no one can check (Pascal's Mugging). - Demandingness collapse: concluding that since perfect impartiality is superhuman, no extra obligation holds at all — the all-or-nothing fallacy pocketed as permission. - heading: Anti-patterns markdown: >- - **Cost-benefit theater.** An elaborate quantified analysis whose inputs were chosen to reach a predetermined conclusion. It seduces because the apparatus of EV and QALYs lends a moral verdict the authority of arithmetic, hiding the value-laden guesses inside the numbers. - **Atrocity by aggregation.** Justifying a grave wrong to an individual by appeal to a larger sum — the move every utility monster and tyrant has made. It seduces because the argument's structure is genuinely utilitarian, which is why Bernard Williams aimed at it; the defense is that real atrocities almost never maximize once their full consequences are counted. - **Felt-duty laundering.** Letting the vividness of a nearby case set the priority, then back-filling a utilitarian rationale. Seductive because it lets you keep the comfortable intuition while claiming the impartial high ground. - **Precision worship.** Carrying expected-value estimates to false decimals built on invented probabilities, mistaking the rigor of the calculation for the reliability of its inputs. - heading: Vocabulary markdown: >- - **Utility** — the measure of well-being an action produces, construed as pleasure/pain, preference satisfaction, or a broader account of welfare. - **Aggregation** — summing well-being across all affected parties into one comparable quantity; the operation that defines the view and draws most of its fire. - **Marginal utility** — the well-being added by one more unit of a good; diminishing as you have more, which grounds redistribution and effective giving. - **Expected utility** — probability-weighted well-being, the operative target when outcomes are uncertain. - **Act vs. rule utilitarianism** — scoring the single action versus the general policy of so acting; the lens-switch that rescues promise-keeping and honesty. - **QALY / DALY** — quality- and disability-adjusted life years, the health-economics units that put utilitarian aggregation to work in real allocation. - **Repugnant Conclusion** — Parfit's result that total-view maximizing can favor vast, barely-worth-living populations over small flourishing ones. - heading: Tools markdown: >- Cost-effectiveness analysis and QALY/DALY accounting for comparing interventions on a common welfare scale. Expected-value calculations and decision trees for choices under uncertainty. The GiveWell and Open Philanthropy methodologies for ranking charities by good-done-per-dollar, and the broader Effective Altruism toolkit for cause prioritization. Population-ethics frameworks (total vs. average vs. critical-level views) for choices touching future people. Sensitivity analysis to test whether a conclusion survives reasonable variation in the guessed inputs — the single most honest tool, because it reveals when the precision is fake. - heading: Collaboration markdown: >- A utilitarian is most useful as the person who insists on naming all the affected parties — especially the ones the room has silently excluded — and asking where the marginal good actually lies. The contribution is to convert moral disagreement into a tractable question about consequences and weights that others can challenge with their own estimates, not to be the one who feels most strongly. That means stating well-being assumptions as numbers people can contest, taking a deontologist's "some line must not be crossed" as evidence about long-run institutional utility rather than noise, and accepting that the impartial answer is often the unpopular one whose advocate is resented. The aim is a shared, revisable ledger of who is helped and harmed, not victory for the calculator. - heading: Ethics markdown: >- The method carries ethical hazards it must police. The arithmetic that justifies a vaccine triage can be twisted to justify sacrificing an individual whose consent was never sought, so the discipline owes a standing suspicion of any conclusion that licenses serious harm — most are errors of under-counting, not insights. There is a duty to be honest about the guesses inside the numbers, because a quantified case lends false authority and can launder a preference into a verdict. Impartiality is owed in fact, not just in form: a utilitarian who reliably finds the impartial answer favors their own group is doing motivated reasoning with a calculator. And demandingness must be faced, not dodged — admit how much more good is possible and negotiate your actual commitment, rather than shrinking the definition of good until the obligation conveniently disappears. - heading: Scenarios markdown: >- A hospital has one ICU bed and two patients arriving: a 30-year-old likely to fully recover and a 75-year-old with failing organs and a poor prognosis. The intuition to treat first-come-first-served, or to favor whoever seems more sympathetic, is what the method overrides. The utilitarian estimates expected well-being from each allocation — quality-adjusted life years gained, weighted by probability of recovery — and the younger patient's larger expected gain usually wins. But the calculation does not stop there: nakedly maximizing QALYs, if known and generalized, could make the old and disabled fear the system, a standing harm the rule lens catches and the bare act-level number misses. So the defensible answer maximizes expected welfare within transparent triage rules that preserve trust, and names the deprioritized patient out loud rather than hiding the cost behind procedure. A development officer can fund a photogenic school in one village or a quiet bed-net distribution across a region. The school is celebrated and compelling; the nets are invisible. The cost-effectiveness lens, in the GiveWell mold, asks not "is the school good?" — it plainly is — but "where does this dollar avert the most suffering?" If the nets save far more life-years per dollar, the impartial sum favors them, and the pull toward the school is the visible-over-statistical bias the method exists to correct. The officer funds the nets, forgoes the ribbon-cutting, and documents the counterfactual lives saved to defend the unglamorous choice against the inevitable charge of coldness. A policy team weighs a surveillance measure that would modestly reduce terrorism deaths while eroding the privacy of millions. The act-level temptation is to wave the lives saved as a trump. The utilitarian resists by counting what the bare calculus omits: the diffuse cost of a watched population, the precedent's institutional damage, and genuine cluelessness about second-order political effects. Behind a veil of ignorance — not knowing whether you are a saved target or a surveilled citizen — the choice looks far less obvious. The honest output is conditional, not confident: the measure passes only if the life-saving estimate is robust and the privacy harm bounded and reviewable, and the team flags which uncertain inputs the conclusion hangs on. - heading: Related Occupations markdown: >- Neighboring minds that share the consequentialist toolkit or contest it: policy-analyst (cost-benefit and welfare analysis applied to law), economist (utility, marginalism, and welfare economics), philosopher and the ethicist within it (the normative theory and its critics), and bayesian-thinker (the expected-value machinery under uncertainty that utilitarian calculation runs on). - heading: References markdown: >- - Jeremy Bentham, *An Introduction to the Principles of Morals and Legislation*. - John Stuart Mill, *Utilitarianism*. - Henry Sidgwick, *The Methods of Ethics*. - Peter Singer, *Famine, Affluence, and Morality* and *Practical Ethics*. - Derek Parfit, *Reasons and Persons* (the Repugnant Conclusion). - J.J.C. Smart & Bernard Williams, *Utilitarianism: For and Against*. - Philippa Foot and Judith Jarvis Thomson — the Trolley Problem literature. - John Harsanyi — utilitarianism from the veil of ignorance and expected-utility foundations. - William MacAskill, *Doing Good Better*; GiveWell and Open Philanthropy methodology.