Episode Summary
Executive Summary: Hilary Greaves argues that strong long-termism is surprisingly robust across moral theories: even non-utilitarians may have reason to prioritize the very long-run future because the stakes can be enormous and lock-in effects matter. The conversation also clarifies cluelessness, moral uncertainty, and a new paper defending the coherence of existence comparativism, while highlighting open research agendas in global priorities research.
Main Topics: Strong long-termism and its scope: Greaves explains axiological vs deontic strong long-termism, why the thesis is about decision relevance rather than vague talk of “the future,” and why long-termism can matter even when extinction risk is not the main issue. Robustness across moral theories: The discussion argues that strong long-termist conclusions can survive changes in population axiology, prioritarianism, and even many non-consequentialist views, because the expected stakes of the far future may dwarf short-term considerations. Cluelessness and uncertainty: Greaves distinguishes simple cluelessness from complex cluelessness, emphasizing that uncertainty about long-run consequences often lacks symmetry and can make probability assignment feel arbitrary without decisively favoring short-termism. Moral uncertainty and decision theory: The episode reviews approaches such as maximize expected choiceworthiness, favorite-theory views, and bargaining/parliamentary models, with Greaves favoring expected choiceworthiness while acknowledging fanaticism and inter-theoretic comparison problems. Existence comparativism and population ethics: Greaves and the hosts discuss whether it can be better for someone to exist rather than not, why standard anti-comparativist arguments may fail, and how this could affect views on extinction and future population. Global Priorities Institute and research agenda: Greaves describes GPI’s slow but growing influence, its role in launching global priorities research, and open problems in fanaticism, infinite ethics, and long-run forecasting.
Key Arguments: Strong long-termism is not just about extinction risk; it can also concern trajectory changes, governance lock-in, and value-system lock-in, including via AI. Axiological long-termism appears robust to many plausible ethical views because future interventions can affect average well-being even when population size is held fixed. Prioritarianism does not automatically oppose long-termism; if the future could be very bad, prioritarianism may actually strengthen the case for long-term interventions. If axiological long-termism is true, it is likely true by a large margin, making it hard for most non-consequentialist theories to override the long-termist conclusion in high-stakes cases. Simple cluelessness is often symmetrical and may cancel out in expected-value reasoning; complex cluelessness is more troubling because rival causal stories can create arbitrary-looking probability assignments. Cluelessness is not clearly an objection to long-termism specifically; it may lower confidence in both short-termist and long-termist claims and should prompt more research rather than immediate retreat to local concerns. The maximize expected choiceworthiness approach remains Greaves’s preferred moral uncertainty theory, despite open problems about inter-theoretic comparisons and fanaticism. Bargaining-theoretic or parliamentary models do not obviously solve the problems of moral uncertainty; in the Nash bargaining version, results look close to expected choiceworthiness. Existence comparativism is defended by arguing that standard anti-comparativist arguments rely on questionable assumptions about relations requiring actual relata and ignore modal/reanalysis possibilities.
Data Points: Future people in expectation: ~10^15 - Greaves offers a ballpark estimate for the expected number of future people when considering the long-run future. Long-run time scale: ~1 million years - Used to distinguish strong long-termism from ordinary political or near-term “long-termism.” Conference dates: November 20–21 - EA Global X virtual conference mentioned in the intro. Paper publication timing: Upwards of 1 year - Greaves notes academic publishing is slow, often taking a year or more even for nearly finished papers. Family/home-office example: 4:30 a.m. - Greaves says she sometimes wakes at 4:30 a.m. and works before family responsibilities begin. Model threshold example: 10^-100 - Mentioned as an example of a de minimis probability cutoff that still runs into problems when probabilities are partitioned differently.
Pivotal Quotes: "If long-termism, then long-termism by a really large margin." — Hilary Greaves: Greaves’s argument that deontic theories will usually not override strong axiological long-termism when the stakes are enormous. "It seems likely that if axiological strong long-termism is true at all, then probably it's going to be true by a large margin." — Hilary Greaves: Explains why non-consequentialist constraints may become decision-irrelevant at very high stakes. "I think it's not an objection to long-termism. And maybe that's for reasons that you just hinted at when you said what people often say to you." — Hilary Greaves: Her response to the idea that cluelessness undermines long-termist reasoning.
Implications: Listeners should treat long-termist reasoning as broader than extinction prevention and not dismiss it due to non-utilitarian ethics or cluelessness. The episode points to major open research areas in forecasting, moral uncertainty, and population ethics that could materially shape future priorities.