Will MacAskill on Longtermism and What We Owe the Future

Episode Summary

Executive Summary: Russ Roberts and Will MacAskill discuss longtermism: the moral claim that future people count and that today’s choices can shape an enormous future population. They debate whether we should prioritize existential risk reduction, moral progress, and institutional design over present-focused spending, while also wrestling with objections about uncertainty, egoism, religion, and whether future people may be richer but still vulnerable to catastrophe or bad values.

Main Topics: Longtermism as a moral framework (Priority: 5/5): MacAskill argues that because future people may vastly outnumber the living, their welfare should be a major moral concern today, especially when current choices can affect whether those people exist and what their lives are like. Existential risk and civilizational fragility (Priority: 5/5): The conversation focuses on threats such as nuclear war, engineered pandemics, biotechnology, and value lock-in that could permanently damage humanity’s long-run trajectory. Moral demandingness and utilitarianism (Priority: 4/5): Roberts presses whether longtermism implies extreme sacrifice; MacAskill says his case does not require full utilitarianism and argues that current spending on long-term good is still tiny relative to what would be justified. Moral progress vs. centralization (Priority: 4/5): They explore the tension between coordinating globally to reduce catastrophic risk and preserving diversity, experimentation, and openness so societies can keep improving morally rather than freezing into one ideology. Historical contingency and moral entrepreneurship (Priority: 5/5): The discussion uses slavery, Benjamin Lay, and the 20th century to show that moral progress is not inevitable; individuals and movements can redirect history, for better or worse. Human flourishing, well-being, and future generations (Priority: 3/5): They debate whether lives are net positive, what counts as well-being, whether nonexistence is preferable for some, and whether having children can itself be a way of improving the future.

Key Arguments: Future people matter morally just as present people do; the fact that they are later in time does not erase their interests. Because there may be vastly more future people than present people, small changes today can have enormous moral significance over the long run. Current concern for the future is likely too small; even a modest increase in long-term-focused effort would be morally important because present spending is far below 1% of world GDP. Future people may be wealthier, but that does not make them immune to catastrophic harms, extinction, enslavement, or bad value systems. The biggest long-run risks are not just marginal harms like a little more or less consumption, but catastrophic or civilization-shaping events. Civilizational collapse would be devastating, but humanity is likely resilient enough to recover technologically if enough knowledge, institutions, and population remain. Moral progress is less automatic than technological progress; slavery’s end depended on contingent social change and moral activists like Benjamin Lay. A good long-term society should preserve open debate, diversity of moral views, and decentralized experimentation while still limiting existential risks. Religion historically helped restrain selfish behavior through belief in moral surveillance, but long-term moral progress should rely more on reason and persuasion than coercion. Having children can be longtermist if children are raised well and their lives contribute positively; the carbon cost can be offset and outweighed by their future contributions.

Data Points: Human lives represented in thought experiment: about 100 billion lives - The opening thought experiment follows the birth order of all humans who have ever lived. Total human life on Earth so far: almost 4 trillion years - MacAskill’s opening thought experiment sums all past human lifetimes. Hunter-gatherer share of total human experience: 10% - In the thought experiment, this portion of life is spent as hunter-gatherers. Agriculturalist share of total human experience: 60% - In the thought experiment, most human experience occurs under agriculture. Time raising children: 20% - A share of the thought-experiment life is spent raising children. Time farming: 20% - A share of the thought-experiment life is spent farming. Time in religious rituals: almost 2% - Portion of the thought-experiment life spent in ritual activity. Time afflicted with malaria or smallpox: over 1% - Disease burden in the thought experiment. Coffee consumed: 44 trillion cups - A vivid statistic from the opening long-range historical reconstruction. Share of experience after AD 1200: a full third - Population growth makes later history disproportionately represented. Share of experience after 1750: a quarter - The modern era occupies a growing share of total human experience. Time in space: 150 years - A playful estimate in the thought experiment about future technological life. Time walking on the moon: one week - Another example of modern technological experience in the thought experiment. Share of experience of people alive today: 15% - How much of all lived human experience is contemporaneous with the present. Probability year 2300 is better off than today: 80% - MacAskill estimates future well-being is likely higher, but not certain. Probability year 2300 is worse off than today: 20% - MacAskill stresses significant downside risk from catastrophe or stagnation. Risk of all-out nuclear exchange in the Cold War: about 1 in 3 - MacAskill uses this as a historical example of serious existential risk. Target risk reduction he wants: 1% or 0.1% - He says he would like nuclear-war risk driven far lower than historical levels. Current longtermism-related share of world GDP: 0.1% or 0.01% (his estimate) - MacAskill argues current effort to improve the long-term future is extremely small. World population in a post-collapse scenario: 80 million - A thought experiment about civilizational recovery after catastrophe. U.S. population in comparison: 350 million - Roberts argues specialization and trade matter greatly for recovery. Climate impact of a child in the UK: about £10,000 per year - MacAskill cites this as a rough cost associated with childrearing. Cost to avert one ton of CO2 via effective climate nonprofits: about $1 - Used to argue that carbon impacts of having a child can be offset. Estimated child offset example: $1,000 can offset around 1,000x the child’s carbon impact - MacAskill says a modest spending increase can more than neutralize emissions. Current U.S. fertility desire vs. reality: 2.6 desired vs. 1.8 actual children - He uses this to argue people often want more children than they have. Current share of world population that is atheist or agnostic: 16% - MacAskill references this while discussing the social role of religion. Share of world population under forced labor: 0.5% - Used to illustrate moral progress relative to the 1700s. 1700s share of people in some form of forced labor: majority - MacAskill contrasts the past with the modern world to show change. British Empire slavery dates: 1807 and 1833 - 1807 outlawed the slave trade; 1833 outlawed owning slaves in much of the empire.

Pivotal Quotes: "future people count morally. There could be a lot of them. And we can really make a difference to their lives." — Will MacAskill: A concise statement of the core longtermist thesis. "I think the numbers really matter." — Will MacAskill: MacAskill explains why the sheer scale of future populations changes moral priorities. "Why should I care about suffering? There's no further reason that one can give." — Will MacAskill: His reply to Roberts’ challenge about why morality should extend beyond self-interest.

Implications: The episode argues that policy, philanthropy, and personal choices should weigh long-run civilization outcomes far more heavily. It encourages listeners to think about existential risk, moral progress, and the future impact of institutions, technology, and family decisions.