Research
Prioritising problems
The biggest decision in giving is not which organisation to fund. It is what problemProblem. A specific issue within a cause area that reduces the well-being of a target population. to work on. Two donors of equal generosity can end up far apart on what their money changes, and most of that gap is decided here.
Scale, neglect, tractability
Within a cause areaCause. A broad area of concern to donors., we compare problems on three questions. How large is the problem, in lives affected, and how badly? How neglected is it, given everyone already working on it? And how tractable is it: would another rupee actually move it, and what would it take?
This framework is a starting point, not a law. We combine these with other factors as relevant to solving the problem.
Practically, this looks like talking to researchers and experts to answer the questions above and understand the causal factors. As a starting point, we draw on the prioritisation research of expert evaluators that may have already looked into the area, such as Animal Ask, Animal Charity Evaluators, Giving Green, Coefficient Giving, EA Funds, Founders Pledge and GiveWell.
Measured, where measurement exists
Where the data allow, we compare problems with indicators that quantify the problem: disease burden in disability-adjusted life years, CO₂ emissions, income lost or gained, animals affected and the severity of what they experience. Numbers like these are imperfect, and they are still better than adjectives. When two problems cannot be compared on a common unit, we say so rather than forcing one.
We also spread our bets across worldviews. It is hard to weigh a life saved today against a childhood of suffering averted, and the honest response is to prioritise problems that look important under many reasonable views.
The bottleneck, named
Tractability asks how hard a problem is. The sharper question is what, specifically, is holding it back: funding, talent, infrastructure, a proven interventionIntervention. A specific mechanism for change., a missing technology, or a policy window that has not opened. Naming the limiting factor matters, because money aimed at the wrong one buys nothing. A field short of trained people does not need more grants to spend; it needs the training pipeline funded first.
So our cause thesis names the bottleneck. Where the bottleneck is not money, we say so, and look at whether a donor can fund its removal instead. In some cases we may realise it is really not possible to make progress at a reasonable cost, and we explain how we got there.
Global evidence, Indian context
We do not start from zero. Evaluators and research organisations publish serious comparative work, and we read it before adding our own. But evidence generated elsewhere does not transfer to India by default, so every global finding passes through the local questions before it enters a thesis.
- Credible implementers: a programme may be well evidenced, but if a competent organisation with the right capabilities and networks does not exist in the target area, the programme may not succeed
- State delivery capacity: if the end game is government adoption, the administrative machinery an intervention depends on differs sharply from one state to the next, and a programme that runs well where it is strong can stall where it is stretched. It may be cost-effective in a study but might not remain so once it is adapted to be delivered by the state
- Public programme interaction: whether private money adds something the large public schemes do not already do, duplicates them, or quietly pushes them out
- Political durability: who stands to lose if the intervention succeeds, who the actors are and what their incentives might be, whether it can outlast that opposition, and whether the timing is right
What this produces
A thesis per cause, not a list of names
The output of this work is a cause thesis: where the leverage sits, what is already funded, and where the next rupee goes furthest. Those theses drive our funding circles and our advisory. Once a problem is chosen, the next question is what actually works on it: identifying solutions.