Our mission can be pursued in many ways, each shaped by different assumptions about what matters morally and about how change happens.
While our funds can be supported individually, some donors ask for our advice on allocating their money across our portfolio. This involves judgment calls on questions that don’t have clear or universally accepted answers, such as:
These judgment calls involve more than just moral considerations; they also involve different empirical beliefs and different strategic approaches to impact. Some approaches emphasize tight feedback loops and measurable results, while others focus on shaping the long-term trajectory of scientific progress or reducing risks that could affect future generations.
We call these overarching sets of assumptions worldviews. Because we find several worldviews plausible and don’t think any single one is clearly all-things-considered much better than the others, we practice worldview diversification — dividing resources across several different worldviews.
In practice, we apply this strategy when allocating large amounts of funding across cause areas. For smaller donors, or those with greater conviction in a particular worldview, we think it can make sense to focus within a single area. But at our scale, our uncertainty about which worldviews are most promising — combined with our capacity to meaningfully support several — makes dividing our efforts appealing.
Four broad worldviews currently guide our advisory and grantmaking work. They differ not just in terms of moral priorities but also in their theories of change, the kinds of evidence they rely on, and the timescales over which they expect to see impact.
These worldviews represent different, though of course not exhaustive, answers to the question of how impact happens and what counts as doing good:
Each rests on reasonable but contestable assumptions about values, how change happens, and how to model impact over time. The question we face is how to act when we find all of these worldviews plausible at the same time.
If we had perfect information, we might discover that one worldview points to opportunities far better than the rest. In that case, the right strategy would be straightforward: put all our resources behind that worldview, rather than diluting our impact. By analogy, an investor who knows for sure which stock will outperform wouldn’t diversify — they’d put everything in that stock.
In principle, the same could be true in philanthropy: we could commit to our “best guess” worldview, or the one with highest stakes, and accept the risk that we might be wrong. That kind of risk tolerance makes more sense in philanthropy than in financial investing. There’s no philanthropic equivalent to going bankrupt, so to help others the most we should aim for maximum expected impact, not simply hedge our bets. This difference from finance highlights an important point that is often overlooked: philanthropists can afford to take bigger risks in pursuit of greater impact.
But our uncertainty over worldviews runs far deeper than an investor’s uncertainty about which asset will perform best.
When moral weightings differ dramatically, they can flip the ranking of causes entirely. For example, if you think animal suffering matters even a little, work to reduce factory farming could seem vastly more cost-effective than global health programs; if you think it doesn’t matter at all, those same efforts look worthless. The ranking of causes depends almost entirely on those moral assumptions, not on new empirical data.
It seems unlikely that all moral frameworks would rate such causes as equally effective. If we knew which value system was “correct,” one approach would clearly dominate. But values aren’t empirical facts, and thoughtful people with the same evidence can reasonably disagree about which kinds of good matter most.
These disagreements can’t be resolved simply by gathering more information. We’re uncertain not just about facts but also about values and methods: how to compare very different kinds of good, and what counts as good in the first place. In that context, we can’t simply pick the “highest-expected-value” worldview and go all-in. Doing so would require judgment calls we can’t make with confidence, since it’s unclear that any single worldview wins on neutral grounds.
We practice worldview diversification because it’s the most coherent way we’ve found to act under deep uncertainty. There are several overlapping arguments for it — some philosophical, some practical.
We believe the most compelling case for worldview diversification comes from moral philosophy.
When people disagree about values, there’s plausibly no neutral way to decide who is right. The worldviews we use differ not only in emphasis but in the scope of moral concern — whose welfare counts, over what timescales, and by what measure. In those cases, there isn’t a clear way to say that one worldview is “more important” or “more correct” than another.
These disagreements go deeper than uncertainty about facts. Worldviews are often incommensurable, meaning there’s no obvious shared scale on which to measure them. One worldview might place enormous moral weight on preventing animal suffering; another might focus on improving human wellbeing today; another on safeguarding the long-term future.
You might think such views could be reconciled by adopting a single “common currency,” such as the value of saving a human life today, but we don’t think so. A worldview that prizes present-day reciprocity and shared social bonds differs in important ways from one that measures good by aggregating welfare across all sentient beings or over millions of years. Someone who treats animal and human suffering equally on utilitarian grounds may, in the eyes of others, be missing what is distinctively bad about human suffering. These frameworks don’t just disagree on how much something matters. They disagree on what kind of thing matters at all.
We recognize that the idea that moral frameworks can’t be compared on a single scale is a controversial philosophical stance and we don’t claim it’s obviously correct. But it leads us to generally allocate resources across worldviews based on how credible or well-supported we find each one overall, rather than by comparing them through a single shared metric. That means that we will fail to maximize impact by any single standard, but we see that as an acceptable trade-off, given the depth of moral uncertainty we face.
Even if we could somehow translate all moral values into a single scale, there would still be practical reasons to diversify. Worldview diversification could still be the right choice even on purely expected-value grounds.
Each worldview has a limited number of highly cost-effective opportunities; beyond that, marginal impact can decline quickly. If we concentrated all our resources within one worldview, we’d eventually fund weaker opportunities while leaving strong ones in other worldviews untapped. From an expected-value perspective, spreading resources across several worldviews can increase total impact whenever diminishing returns set in faster than our confidence that one worldview is clearly superior.
This argument has less force when one worldview looks overwhelmingly stronger than the rest. But given the deep incommensurability we described above, we don’t believe that’s the situation we’re in.
Worldview diversification has other practical benefits that make us a stronger organization.
Option value: If our beliefs shift, or if new evidence changes what seems most promising, we’re better positioned to redirect resources quickly. We’ve already seen our views evolve significantly over the last decade. Working across multiple worldviews means we’re prepared to scale up in whichever direction the evidence leads.
Cross-pollination of ideas: Lessons from one area often sharpen our reasoning in others: insights from global health can inform our thinking about biosecurity, and our understanding of which institutions are useful in mature fields helps us support similar institutions in newer or riskier areas. We don’t always know in advance which lessons will transfer, but working across causes helps us notice patterns and learn.
Organizational resilience and internal culture: Working in a variety of causes helps others understand that we’re guided by a consistent concern for helping others as much as possible, even if the implications differ across worldviews. It makes Coefficient Giving a more engaging place to work, attracting staff who bring diverse perspectives and values. This range of perspectives also guards against intellectual blind spots: reasoning from multiple starting points keeps us honest about our assumptions and broadens what we can learn from both our successes and our mistakes.
Worldview diversification reflects the depth of our uncertainty — empirical, moral, and methodological — and our effort to act responsibly within it.
We don’t claim to have a precise formula for balancing worldviews. Reasonable people can and will weigh these uncertainties differently, and our own views will likely continue to evolve. If we someday come to believe that one worldview overwhelmingly dominates, we might choose differently.
For now, we see strong, reasonable arguments for several distinct ways of doing good, and no coherent way to collapse them into one. Diversifying across those worldviews lets us move forward responsibly without pretending we’ve settled these difficult philosophical questions.
This page draws on material from an article published in 2016.
Philanthropists tend to choose causes based on personal experience, geographic proximity, or emotional appeal. This approach is understandable — personal connection is what motivates many people to give in the first place. But some causes offer far greater opportunities for impact than others: the same donation might save ten lives in one place but a thousand in another.
One of our core values is our tolerance for philanthropic risk. We’ve seen that the biggest philanthropic wins often come from ideas that seem unlikely to succeed, so we’re open to funding a lot of work that could fail in order to find a few transformative successes.
Measuring cost-effectiveness helps us compare very different opportunities and direct funding to where it can do the most good.