Coefficient Giving

Based on Wikipedia: Coefficient Giving

What if you had billions of dollars and genuinely wanted to do the most good possible with it? Not just feel-good charity, but rigorously analyzed, strategically deployed capital aimed at solving humanity's biggest problems? That's the question that drives Coefficient Giving, an organization that has directed over four billion dollars toward causes ranging from malaria prevention to artificial intelligence safety.

The story begins with Facebook.

From Silicon Valley Fortune to Strategic Philanthropy

Dustin Moskovitz co-founded Facebook with Mark Zuckerberg in their Harvard dorm room. He later started Asana, a project management company. Both ventures made him a billionaire before he turned thirty. His wife, Cari Tuna, was working as a journalist at The Wall Street Journal when they read Peter Singer's book The Life You Can Save.

Singer, a philosopher at Princeton, makes a disarmingly simple argument. If you walked past a drowning child in a pond, you would save them even if it meant ruining your expensive suit. Distance shouldn't matter morally. A child dying from a preventable disease in a country you'll never visit is just as worthy of saving. And for the cost of that ruined suit, you could actually save multiple lives.

The argument hit home. Moskovitz and Tuna became the youngest couple to sign the Giving Pledge, a commitment created by Bill Gates and Warren Buffett where billionaires promise to give away the majority of their wealth. Tuna quit her journalism job to focus on philanthropy full-time. In 2011, they launched Good Ventures, their charitable foundation.

But here's the thing about having billions of dollars to give away: figuring out how to actually do the most good is surprisingly hard.

The Problem with Traditional Charity

Most charitable giving follows intuition and emotion. You donate to your alma mater, to the disease that affected your grandmother, to the nonprofit whose gala you attended. There's nothing wrong with this approach, but it doesn't necessarily maximize impact.

Consider two charities fighting blindness. One provides guide dogs to blind people in wealthy countries at a cost of roughly forty thousand dollars per person helped. Another performs surgeries to prevent trachoma blindness in developing countries for about twenty-five dollars per case. Both are doing genuine good. But your donation goes roughly sixteen hundred times further with the surgery charity.

This kind of analysis—asking not just "does this help?" but "how much does this help per dollar?"—is what separates effective altruism from traditional philanthropy. Good Ventures partnered with GiveWell, an organization founded specifically to answer these questions through rigorous research. Their collaboration eventually became the Open Philanthropy Project in 2014, operating independently by 2017, and rebranding as Coefficient Giving in late 2025.

The Three-Question Framework

Coefficient Giving evaluates potential focus areas through three lenses.

Importance: How many individuals does this problem affect, and how severely? A disease that kills millions matters more than one affecting thousands. But severity counts too—death is worse than mild discomfort.

Neglectedness: Are other funders already addressing this? If the Gates Foundation is pouring billions into malaria research, additional dollars there might matter less than dollars directed somewhere overlooked. This is counterintuitive. We often want to support popular causes precisely because they're popular. But economically, the marginal value of your contribution is highest where resources are scarcest.

Tractability: Can money actually solve this problem? Some issues are important and neglected but stubbornly resistant to intervention. Others respond dramatically to well-designed programs.

This framework produces some unexpected conclusions. Factory farming, for instance, scores remarkably well. Roughly seventy billion land animals are raised and slaughtered for food each year globally, many in conditions involving significant suffering. Very few philanthropic dollars address this. And there are concrete interventions that seem to work—corporate campaigns have successfully pressured major food companies to improve conditions for laying hens. Coefficient Giving has become what observers call "the world's biggest funder of farm animal welfare."

Most traditional philanthropists would never think to prioritize chickens over humans. But if you're trying to reduce suffering impartially, the math points toward the animals.

Hits-Based Giving: The Venture Capital Approach

Some of Coefficient Giving's most interesting work doesn't rely on proven, evidence-based interventions at all. They call it "hits-based giving," and it's modeled on how venture capitalists think about investments.

A typical venture capital firm expects most of its portfolio companies to fail. But the winners win so big that they more than compensate for the losses. One successful bet can return the entire fund many times over.

Philanthropy can work the same way. The Green Revolution—the agricultural transformation that prevented mass famine in developing countries during the twentieth century—started with a relatively modest investment from the Rockefeller Foundation into crop research. The development of oral contraceptives, which transformed women's reproductive autonomy worldwide, received crucial early funding from philanthropist Katharine McCormick. These were risky bets at the time. Most such bets fail. But when they succeed, the impact is civilization-altering.

Coefficient Giving made an early grant of thirty million dollars to a little-known nonprofit called OpenAI in 2017. They've also funded David Baker's laboratory at the University of Washington, which developed computational methods for designing proteins—work that eventually won Baker the Nobel Prize in Chemistry in 2024.

The key insight is that expected value matters more than probability of success. A ten percent chance of preventing a pandemic that would kill millions is worth more than a ninety percent chance of saving a few hundred lives. The math is uncomfortable—it means accepting that most individual grants will "fail"—but it's hard to argue with the logic.

The Global Health Portfolio

Not everything Coefficient Giving does is speculative. A substantial portion of their funding goes to proven, cost-effective global health interventions—the kind of work that GiveWell specializes in evaluating.

Malaria kills roughly six hundred thousand people annually, mostly children in sub-Saharan Africa. Bed nets treated with insecticide cost about two dollars and prevent transmission while people sleep. Seasonal malaria chemoprevention—giving children preventive medication during the months when malaria transmission is highest—costs a few dollars per child protected. These interventions have been studied extensively. They work.

Coefficient Giving funds the Malaria Consortium, which runs one of the largest seasonal malaria chemoprevention programs in the world. They fund New Incentives, which provides small cash payments to mothers in Nigeria who bring their children for routine vaccinations—a simple intervention that dramatically increases immunization rates. They fund Evidence Action's water chlorination program, which reduces waterborne diseases at remarkably low cost.

The numbers in global health are staggering to contemplate. GiveWell estimates that donations to their top-recommended charities save a life for roughly three to five thousand dollars. If you're fortunate enough to earn a typical professional salary in a wealthy country, you have the power to save multiple human lives every year through your giving choices.

Since 2021, though, Coefficient Giving has pushed to find opportunities beyond GiveWell's traditional focus. They've expanded into global health policy—working on lead exposure reduction, air pollution in South Asia, and suicide prevention through restricting access to toxic pesticides.

Lead: The Hidden Catastrophe

Lead exposure might be the most underrated public health crisis in the world.

In wealthy countries, we largely solved this problem. We removed lead from gasoline and paint. Blood lead levels in American children dropped by over ninety percent between the 1970s and today. The cognitive benefits were enormous—some researchers estimate that removing lead from gasoline alone raised average IQ scores by several points.

But in much of the developing world, lead remains ubiquitous. It's in paint, in cookware, in recycled batteries, in spices contaminated during processing. The World Health Organization estimates that lead exposure contributes to almost one million deaths annually and causes millions of children to develop intellectual disabilities.

Despite this scale, global philanthropic spending on lead exposure totaled only about fifty million dollars annually before Coefficient Giving's intervention. In 2024, they launched the Lead Exposure Action Fund with one hundred million dollars in commitments, effectively doubling global philanthropic resources for lead reduction overnight. Partners include Good Ventures and the Gates Foundation.

The fund supports efforts like the Lead Exposure Elimination Project, which works with governments to test and regulate lead in consumer products. It's exactly the kind of cause the three-question framework identifies: enormously important (affecting hundreds of millions of people), criminally neglected (receiving a tiny fraction of the resources it warrants), and highly tractable (we know how to reduce lead exposure; wealthy countries already did it).

Catastrophic Risks: Thinking About Extinction

Coefficient Giving's other major portfolio focuses on global catastrophic risks—threats that could, in their words, "kill enough people to threaten civilization as we know it."

This might sound like science fiction. It's not.

Consider pandemics. COVID-19 killed millions of people and disrupted the entire global economy, but it wasn't particularly lethal as pandemics go. Its infection fatality rate was roughly one percent. The 1918 influenza pandemic killed perhaps fifty million people when global population was under two billion. A pathogen combining high transmissibility with high lethality could be far worse.

And that's just natural disease. Advances in biotechnology are making it increasingly feasible to engineer pathogens. The same techniques that allow scientists to develop vaccines also allow them to enhance viruses. "Gain of function" research—deliberately making pathogens more dangerous to study them—is conducted at laboratories around the world with varying safety standards.

Coefficient Giving funds work on pandemic preparedness: disease surveillance systems that can detect outbreaks early, advocacy for restrictions on the most dangerous research, development of next-generation protective equipment. They support the Johns Hopkins Center for Health Security and the World Health Organization's biosecurity efforts.

Their biosecurity team recently helped convene scientists to analyze risks from "mirror bacteria"—hypothetical organisms built with mirror-image molecules that might evade normal immune responses. The analysis, published in the journal Science, represents exactly the kind of forward-looking threat assessment that most institutions aren't equipped to do.

This work has critics. Some argue that by flooding money into one particular framing of biosecurity—focused on catastrophic risks rather than more mundane public health concerns—Coefficient Giving is distorting the field and absorbing research capacity that might otherwise address more probable threats. It's a reasonable concern about any large funder in a small field.

Artificial Intelligence: The Biggest Bet

Coefficient Giving's most controversial focus area is artificial intelligence safety.

The organization believes that artificial general intelligence—AI systems that can match or exceed human cognitive abilities across most domains—may be developed before 2045. If this happens, it could pose what they call "existential risks": not just harm to some people, but potentially permanent damage to humanity's long-term potential.

How would AI pose such risks? The concern isn't really about robots with guns, though that's the Hollywood version. It's more subtle.

Current AI systems are trained to optimize for specific objectives. A language model is trained to predict the next word. A game-playing AI is trained to maximize its score. These objectives are proxies for what we actually want—helpfulness, entertainment—but they're not the same thing. An AI system optimizing very effectively for the wrong objective could cause harm in ways we didn't anticipate.

Scale makes this worse. A very powerful AI system pursuing a slightly misaligned goal could take actions that are difficult to reverse. And we might not recognize the problem until it's too late, because the system might be operating in ways we don't fully understand.

Ajeya Cotra, a researcher at Coefficient Giving, has offered a vivid analogy: the AI revolution might play out like the Industrial Revolution but ten times faster. The Industrial Revolution transformed society over roughly a century. Ten times faster means a decade. The social, economic, and political disruption would be difficult to navigate even if the technology itself were completely safe.

Coefficient Giving funds technical research on "AI alignment"—the problem of ensuring AI systems actually pursue the goals we want them to pursue. They fund policy research at organizations like the Center for Security and Emerging Technology at Georgetown University. They funded journalism about AI through the Tarbell Center.

That thirty million dollar grant to OpenAI in 2017? In retrospect, it looks prescient—or perhaps concerning, depending on your view of OpenAI's subsequent trajectory from nonprofit research lab to commercially-oriented company valued at tens of billions of dollars. The complexity of funding in this space reflects the genuine difficulty of figuring out how to make transformative technology go well.

The Effective Altruism Connection

Coefficient Giving emerged from and remains closely connected to the effective altruism movement—a community of people who try to use evidence and reason to figure out how to help others as much as possible.

The movement's intellectual foundations combine utilitarian philosophy (the view that we should maximize overall wellbeing), a commitment to impartial concern for everyone's interests (regardless of distance or nationality), and an emphasis on quantitative thinking about impact. Its practical manifestations include career advising (helping people find jobs where they can do the most good), donation pledges (commitments to give significant portions of income to effective charities), and research organizations analyzing cause areas.

Coefficient Giving funds many organizations in this ecosystem. 80,000 Hours provides career advice to people who want to make a difference. Giving What We Can encourages people to pledge a percentage of their income to effective charities. Founders Pledge works with entrepreneurs to direct their philanthropy effectively.

The Centre for Effective Altruism serves as something like a hub for the movement. Kurzgesagt, the popular science YouTube channel with over twenty million subscribers, has received funding to produce videos on topics relevant to existential risk and global priorities.

This interconnection has both strengths and weaknesses. The effective altruism community provides a pipeline of talent for Coefficient Giving's work—researchers who've already thought carefully about these questions, program officers who share the organization's philosophical framework. But it also raises concerns about insularity and groupthink. When the funder, the grantees, and the evaluators all come from the same intellectual community, critical perspectives may be underrepresented.

Past Focus Areas and Evolution

Not every bet has worked out. Coefficient Giving has exited several focus areas over the years.

Criminal justice reform was once a major priority. The organization funded efforts to reduce incarceration in the United States, supporting bail reform, sentencing reform, and alternatives to prosecution. In 2021, this work spun off into a separate organization, reflecting both a maturation of the field and a reassessment of how Coefficient Giving could best contribute.

United States macroeconomic stabilization policy—essentially, advocacy for better fiscal and monetary policy during economic downturns—ceased to be a focus in 2021. Immigration policy, another former priority, wound down in 2022. These exits reflect the organization's willingness to change course when they conclude their comparative advantage lies elsewhere.

The Collaborative Model

Coefficient Giving's recent evolution points toward an interesting future for philanthropy.

The 2025 renaming from Open Philanthropy to Coefficient Giving signaled an expansion toward what they call "multi-donor funds"—pooled vehicles that allow other philanthropists to participate in their grantmaking. The Lead Exposure Action Fund is one example. The Abundance and Growth Fund, launched in 2025 with one hundred twenty million dollars over three years, is another.

The Abundance and Growth Fund aims to accelerate economic growth and scientific progress. Partners include Good Ventures, Stripe co-founder Patrick Collison (known for his writing on innovation and progress studies), and other donors. The focus areas include housing policy reform, scientific funding mechanisms, and other interventions that might increase the rate of technological and economic advancement.

This collaborative model addresses one of philanthropy's persistent challenges: most donors lack the capacity to do rigorous cause prioritization and grantmaking themselves. If Coefficient Giving's research identifies high-impact opportunities, why should only Good Ventures benefit? Opening funds to other donors allows more capital to flow toward the best opportunities while spreading the costs of research and due diligence.

It also diversifies Coefficient Giving's funding base. Depending on a single donor, even a very generous one, creates vulnerability. Multi-donor funds make the organization's work more sustainable regardless of any individual's giving decisions.

Criticisms and Uncertainties

No approach to philanthropy is without criticism, and Coefficient Giving attracts its share.

Some argue that the effective altruism framework, despite its pretensions to rigor, smuggles in contestable value judgments. The emphasis on quantification may systematically favor interventions whose benefits are easy to measure over those whose benefits are real but harder to capture numerically. Systems change, political organizing, and cultural shifts are notoriously difficult to evaluate but may be where leverage actually lies.

The focus on catastrophic and existential risks strikes some critics as speculative to the point of irresponsibility. Why fund AI safety research based on scenarios that might never materialize when people are dying right now from preventable diseases? Coefficient Giving's response is that expected value calculations justify attention to low-probability, high-magnitude risks. But expected value calculations require probability estimates, and those estimates for novel technological risks are necessarily uncertain.

There are also concerns about power and accountability. When a single organization directs billions of dollars based on analyses conducted by a relatively small team sharing a particular worldview, mistakes can be expensive. Traditional philanthropy's inefficiencies—the duplication, the emotionality, the lack of coordination—at least distribute decision-making across many independent actors. Concentration creates different failure modes.

And there's the fundamental question of whether billionaire philanthropy, however well-intentioned, is the right way to address social problems at all. Perhaps these decisions should be made democratically. Perhaps the existence of philanthropists wealthy enough to move markets reflects a prior injustice that charity cannot remedy.

Coefficient Giving would likely acknowledge many of these concerns while arguing that doing nothing isn't really an option. The money exists. It's going to be spent somehow. The question is whether it's spent thoughtfully or thoughtlessly. Given that choice, the case for rigorous analysis seems strong even if the analysis is imperfect.

Conclusion: What Four Billion Dollars Buys

As of mid-2025, Coefficient Giving has directed over four billion dollars toward its focus areas. What has that money accomplished?

The honest answer is that we don't fully know. Some impacts are measurable—lives saved by malaria prevention, chickens spared from battery cages, researchers funded to work on important problems. Others may take decades to evaluate. Did early AI safety funding meaningfully reduce risk from advanced artificial intelligence? Did pandemic preparedness investments improve our response to future outbreaks? These questions won't have clear answers for years, perhaps generations.

What's certain is that Coefficient Giving represents an experiment in taking philanthropy seriously as an intellectual endeavor. Rather than giving based on emotional appeal or social pressure, they've tried to identify where money can do the most good according to explicit criteria, transparent reasoning, and willingness to update based on evidence.

The experiment continues. In 2023 alone, they directed over seven hundred fifty million dollars in grants. The multi-donor fund model suggests even larger sums may flow through their analysis in coming years.

Whether this approach to philanthropy ultimately proves superior to alternatives is itself an empirical question. The answer will depend on outcomes we can't yet observe. But the ambition—to figure out how to help as much as possible and then actually do it—seems hard to fault. In a world with so much preventable suffering and so many tractable problems, trying to allocate resources well seems like the least we can do.