Against Effective Altruism

Core claim

Responding to Scott Alexander’s “Tower of Assumptions” challenge, the author accepts EA’s foundational premises but rejects the second tier: in a highly taxed welfare state, charitable giving is supererogatory rather than obligatory; QALY maximisation is the wrong goal because culture and nature are valuable in themselves; and particular obligations of kinship, community and reciprocity are real duties that EA’s general utilitarianism displaces.

Subtitle: “Well, not so much ‘against’ as ‘nuanced disagreement’, but who wants that in a title?”

Framing

• Written as a response to Scott Alexander’s “Effective Altruism as a Tower of Assumptions” (2022), which challenged critics to specify where in the stack they get off.
• The tower has five tiers; the author addresses the bottom three (Foundational Assumptions, Less Basic Assumptions, Cause Prioritisations), on the grounds that the top two (Specific Projects, Institutional Credibility) only matter if you’ve already bought the lower ones — as with debating denominations before accepting Christianity.
• Disagreement, not condemnation: EA has plausibly caused more giving and saved lives in the developing world; the world is better for it. This is “I won’t throw myself behind it,” not “raze it and salt the earth.”
• Explicitly declines to use FTX as ammunition — all movements attract rogues, FTX deceived far more sophisticated parties than charities, and the collapse says nothing about the philosophy.

Tier 1 — Foundational Assumptions: full agreement

• We should help other people — unobjectionable under almost any moral code.
• We should consider how much effort we devote to this — less widely held, but yes; worth properly examining at least once in a life.
• Some methods of helping are more effective than others — trivially true.

Tier 2 — Less Basic Assumptions: the core disagreements

”Donate 10% of income or take an altruistic career”

• Why 10%? EA’s answer is psychological: “give everything” produces guilt and paralysis, so a clear discharge point produces more actual giving. The tithe precedent in Judaism and Christianity is cited — though Islam’s zakat is 2.5%, so the figure is hardly universal.
• As psychology this is hard to argue with; the objection is to its status as a moral claim.
• The welfare-state argument: the historical/religious tithe precedent predates modern taxation. UK tax is ~35% of GDP, the highest in over 70 years, and the state now funds what charity once did — poor relief, medical research, overseas aid. That fact must enter the calculation.
• Conclusion: in a highly taxed Western nation, further charitable giving is supererogatory rather than obligatory — praiseworthy, above and beyond duty. The author argues supererogation is a badly underrated category, without which morality collapses into “mandatory or forbidden.”
• Specific obligations do exist, however, and are real duties rather than optimisation problems:
  • underfunded causes your government neglects (e.g. dementia vs cancer research);
  • repaying or “paying forward” benefits received (mentoring, educational opportunity);
  • commitments to communities and organisations you belong to;
  • bonds of kinship and friendship.
• Central charge against EA: it has no room for these particular obligations, substituting a general utilitarian duty for them.

Utilitarian calculus (QALYs)

• Granted at the base level: if you’ve decided to fight malaria, you should pick the charity that saves most lives per pound. EA’s charity evaluation work — and the competitive pressure it exerts on charities — is “inarguably valuable.”
• Measurement problems: bednet lives are countable; scientific research is not. Proxy metrics (citations, publications) can distort the research they measure. Should a possible malaria vaccine in a decade — or never — beat lives saved today? Anti-corruption, policy influence, and institution-building are harder still to measure yet may matter more long-term.
• The deeper objection — QALYs are the wrong maximand:
  • What about art, music, culture? What about the natural environment?
  • The British Museum case: 18th-century bequests founding Britain’s museums and galleries were made amid tremendous poverty. Six million people visit the British Museum annually. Would we prefer Hans Sloane’s wealth had gone to the contemporaneous rural and urban poor? Prosperity came from economic progress, commerce and democracy — charity’s distinctive role is “creating and preserving that which makes life valuable.”
  • The environmental case: rainforests could always be bulldozed to save lives. Do we want our children in a world rich in nature, or one holding the maximum number of humans?
  • Quotes Betjeman’s “The Plantster’s Vision” — clearing away churches and cottages for “workers’ flats in fields of soya beans” — as the image of what QALY-maximisation logic implies. This is the chief reason the author cannot wish EA to become dominant.
• On obligations again: doing good is about fulfilling commitments and repaying debts, not only maximising outcomes. Helping a brother thrown out of his house outranks buying bednets.
• On rationalist transcendence of “sentiment”: EA thinks it has moved beyond such ties; the author says it has moved below them — invoking the N.I.C.E. scientists in C.S. Lewis’s That Hideous Strength, who got “beyond” justice. Ties of relationship, community, kinship and reciprocity are what turn isolated creatures into civilisation. “We do not progress by ignoring these ties; rather, we regress."

"Wackiness”

• Caveated as unfair to press too hard — EA is young and every movement attracts eccentrics.
• Still treated as diagnostic that the community seriously entertains: annihilating predators to spare prey, insect suffering, Roko’s basilisk, arguments justifying infanticide.
• Read as a sign the framework is at best prone to malfunction as a reasoning tool, at worst carrying deeply immoral axioms.

”Helping the 3rd World beats helping the 1st”

• The tier the author has most time for.
• But he is an aid sceptic: countries that escaped poverty did so via good economic policy and strong civic institutions, not aid. Aid may be merely ineffective or actively counterproductive — distorting economies toward the public sector, diverting talent into aid-maximising work, and entrenching corrupt elites.
• Aid that does work: disaster relief; disease elimination and epidemic control (smallpox, polio, ebola); large-scale reconstruction like the Marshall Plan or EU accession work in Eastern Europe (though only states can do that last one).
• Bottom line: if you accept the utilitarian frame, you probably can do more good per pound abroad — and saving lives is good regardless of long-term effects.

Tier 3 — Cause Prioritisations

Cause	Verdict
Global health	Accepted. Saving lives is good in itself; health gains may also aid development, though he’s sceptical how large those knock-on effects are without functioning institutions.
Animal welfare	Rejected — not top three, not top ten. Opposed to needless cruelty, but comfortable with a natural-law view that eating animals is fine. No moral weight to wild-animal suffering (a lion eating a gazelle). Conservation matters because nature is good and beautiful in itself, not on welfare grounds. Rejects any calculus trading cows — let alone fleas or mosquitos — against human lives.
X-risks	Middling. Human extinction would obviously be very bad, but unsure how much an individual can do, or that EA’s chosen routes are effective. Still, nuclear disarmament campaigning, AI safety and asteroid deflection aren’t unreasonable pursuits.

Which X-risks

• Nuclear war — unfashionable but underrated. The US and Russia still hold thousands of warheads; an India–Pakistan exchange could kill hundreds of millions; the Cold War’s list of near-misses is alarming; Putin has reminded everyone war isn’t banished. “As long as we still have thousands of these things sitting around, we might use them.” Unclear what a private individual can do, but valid.
• Pandemics — needs little justification post-COVID. Natural extinction unlikely, but gain-of-function research is an eminently sensible worry. Even a non-extinction pandemic killed tens of millions and did lasting economic damage; there are concrete tractable actions here, from biomedical research to lobbying governments that remain woefully unprepared.
• AI — where he is most sceptical. Doubts fast take-off and recursive self-improvement without hitting constraints; sees no basis for treating superintelligence as “an infinite Charisma score.” But humans are dumb enough to deploy killer robots without alignment or write a rapidly evolving agentic virus, and the modern technological world is fragile. Notes Scott Aaronson’s ~2% extinction estimate as a hard-to-argue-with reason to spend some effort on it.

Conclusion

• Systematically locating where you diverge from a movement is more constructive than diffuse unease — and the post is offered to others who sense they disagree but can’t place why.
• Not a personal attack: “I think the effective altruists are genuinely good people,” and both individually and collectively they do a great deal of good. Even the odder causes do no real harm at EA’s current level of political power — a caveat he says would change if EA gained real influence or major parties adopted its priorities.
• Overall: agreement with Tier 1, substantial disagreement at Tier 2, mixed at Tier 3 — but “there’s a lot of good that needs doing in the world, and a lot of space for different people to approach it in different ways.”