How to Do the Most Good

Episode Summary

Executive Summary: Ezra Klein interviews Holden Karnofsky about the evolution of effective altruism from rigorous charity evaluation to higher-risk, worldview-driven philanthropy and long-termism. The discussion covers GiveWell’s evidence-based approach, Open Philanthropy’s diversified ethical buckets, animal welfare, pandemic preparedness, and why Karnofsky increasingly thinks AI could make this century uniquely consequential.

Main Topics: Origins of GiveWell and evidence-based charity (Priority: 5/5): Karnofsky explains how GiveWell emerged from wanting a 'Wirecutter for charities' focused on proven, cost-effective, scalable interventions rather than overhead ratios or vague charity prestige. From GiveWell to Open Philanthropy (Priority: 5/5): The conversation traces how working with major donors like Dustin Moskovitz and Cari Tuna pushed the organization toward higher-risk, hits-based grants and broader philanthropy beyond top charities. Philanthropic success stories and leverage (Priority: 4/5): Karnofsky argues that some of the biggest philanthropic wins came from funding innovation, especially technologies that scale broadly, like the Green Revolution and the pill. Worldview diversification (Priority: 5/5): Open Philanthropy intentionally splits giving across incompatible ethical frameworks instead of forcing one moral theory, allowing support for humans, animals, and long-term future causes simultaneously. Animal welfare and moral uncertainty (Priority: 4/5): They explore the case that factory-farmed animals may matter far more than current norms assume, and that uncertainty about moral weights justifies hedging through dedicated funding. Long-termism and existential risk (Priority: 5/5): Karnofsky increasingly takes seriously the idea that the 21st century may determine humanity’s long-run future, especially through AI, climate, and pandemics. AI, digital people, and a radically different future (Priority: 5/5): The later part of the interview sketches a world where AI accelerates science and growth, potentially enabling digital minds, post-human civilization, or catastrophic misalignment.

Key Arguments: The overhead ratio is a poor way to judge charities; what matters is impact per dollar and whether an intervention is proven, cost-effective, and scalable. GiveWell’s best-supported early recommendation was bed nets for malaria prevention because there is strong randomized evidence and very low cost per life saved. Effective philanthropy often requires looking beyond your own neighborhood or country because money can do far more good in poorer countries. Philanthropy can succeed by funding innovation and leverage, not just existing services; some of the biggest historical successes were technology-like breakthroughs. Open Philanthropy uses 'hits-based giving' for speculative causes because a small number of huge wins can justify many failed bets. Worldview diversification avoids the trap of assuming one moral framework is definitively correct; it lets the organization fund according to multiple plausible ethical views at once. Animal welfare may be a top-tier moral issue if future generations judge factory farming harshly; even low moral weights for chickens can imply major philanthropic value. Long-termism argues that actions affecting many future generations could dwarf present-day impact, so existential risk deserves serious attention. AI may make this century unusually important by accelerating science and technology, potentially collapsing long timescales into decades. The greatest risk is misaligned AI or a race dynamic in which the least careful actors shape the future before society reflects adequately.

Data Points: Cost to avert one death via malaria bed nets: $3,000-$5,000 - Karnofsky cites GiveWell’s estimate for insecticide-treated bed nets as a benchmark for highly effective charity. Cost of insecticide-treated bed nets: About $5 each - Used to illustrate why bed nets are an exceptionally low-cost intervention. Coverage per bed net: Often covers 1-2 people - Explains the leverage of each net distributed. Duration of bed net effectiveness: Several years - Supports the claim that bed nets are durable and scalable. Alternative foods / farm animal welfare: Largest funder in the world - Open Philanthropy says it is the largest funder of farm animal welfare work. Cage-free corporate pledges: Most major U.S. grocers and fast food companies - Karnofsky says Open Philanthropy helped speed and globalize the cage-free movement. Open Philanthropy long-termist allocation proposal: At least 50% of available capital - Referenced from a 2018 update on cause prioritization. Economic growth assumption: ~2% per year - Used to explain how exponential growth can become physically implausible over very long horizons. Future horizon example: A billion years - Used repeatedly in discussing long-termism and the possibility of radically unfamiliar futures. Green Revolution impact: Credited with saving a billion people from starvation - Presented as a major philanthropic success story funded by Rockefeller-supported crop research.

Pivotal Quotes: "if you want to do the most good, in some ways, the worst place you can start is your neighborhood, your friends, your country" — Holden Karnofsky: Explaining why GiveWell broadened the moral circle to prioritize global poverty interventions. "worldview diversification" — Ezra Klein / Holden Karnofsky: Core concept discussed for handling uncertainty across incompatible moral frameworks and future scenarios. "the most important century" — Holden Karnofsky: Describing his view that the 21st century may determine humanity’s long-run trajectory.

Implications: The interview suggests philanthropy should be both evidence-driven and morally pluralistic, while society urgently needs better AI governance, biosecurity, and reflection on how future technologies could reshape civilization.