Lesson 1: The problem of comparison
What’s at stake
Choosing what to work on is one of the highest-leverage decisions you’ll make in your career—and in your life as someone trying to do good. The reason is simple: how much good you do is determined more by which problem you choose than by how well you work on it. A skilled career spent on a small problem can do less good, in expectation, than a competent career spent on a problem that’s a hundred times larger.
Take two real, terrible diseases. In 2021, multiple sclerosis accounted for roughly 0.034% of the global burden of disease. Malaria accounted for about 1.91%—around 54 times more (IHME, GBD Compare). Both are worth fighting. But if you could only meaningfully reduce one, eliminating malaria would avert vastly more suffering than eliminating MS. That’s a 50× difference inside a single domain, available without any philosophical contortion.
Across domains, the gaps are larger and the comparisons are harder. How do you weigh equal rights for disadvantaged groups against reducing the risk of the next pandemic? The suffering of chickens against the suffering of cows? Clean waterways against government transparency? Answering honestly requires wading into questions about values, making subjective calls about facts, and admitting where you’re uncertain. There’s no single correct answer. But there are well-reasoned and poorly-reasoned ones, and some priorities are clearly higher-leverage than others.
This week is about doing that work explicitly enough that you know which of your beliefs and moral commitments are driving your choices, and explicitly enough that you can defend them and update them when better information arrives. You’re not committing to a lifelong cause area. You’re forming a provisional view you can justify, and noticing which inputs would change it.
Before you begin: Record your starting intuitions. This isn’t a test—it’s a baseline to compare against later. Hold onto these answers; we’ll revisit them at the end.
Rank the following cause areas from “most important to work on” to “least important to work on.” Use your gut: global health and poverty, animal welfare, climate change, reducing the risk of catastrophes from advanced AI, reducing the risk of catastrophes from engineered pandemics, improving institutional decision-making.
For your top-ranked cause, write one sentence explaining why you put it first.
For your bottom-ranked cause, write one sentence explaining why you put it last.
Why this matters
Over the past three weeks, you’ve built a serious toolkit. You can evaluate a charity’s evidence (Week 2). You can reason under uncertainty, build BOTECs, and use calibration and forecasting to make better estimates (Week 3). These are powerful skills.
But they leave a prior question unanswered. Before you evaluate how well someone is working on a problem, you need to decide which problems are worth working on in the first place.
This might seem obvious—work on whatever’s most important. But “most important” is doing a lot of work in that sentence. Important to whom? By what measure? Over what timescale?
Consider three people, all thoughtful, all well-informed, all using the same analytical tools:
One concludes that global health is the clear priority: the suffering is measurable, the interventions are proven, and the people affected are alive right now.
Another concludes that animal welfare dwarfs everything else: the sheer number of sentient beings in factory farms makes it the largest source of suffering on Earth, and almost nobody is working on it.
A third concludes that existential risk reduction is overwhelmingly important: if a catastrophe wipes out humanity, all future value is lost—and the expected value of protecting the future is astronomical.
They’re not making different errors. They’re applying the same framework with different inputs—different empirical beliefs about what’s tractable, different moral weights on who counts, different attitudes toward uncertainty. This week is about understanding how those inputs drive conclusions, so you can form your own view with your eyes open.
We’ll learn the ITN framework—the standard tool for comparing cause areas. We’ll apply it. And then we’ll stress-test it, because any framework that claims to compare “preventing malaria deaths” with “reducing the risk of human extinction” should face serious scrutiny.
Why we need a framework
Suppose you have €10,000 to donate, and you want to do the most good. You could fund malaria nets, campaign for chicken welfare, lobby for AI safety regulation, or support clean energy research. How do you choose?
Your Week 2 skills let you compare interventions within a cause area—which malaria charity is most cost-effective, for instance. Your Week 3 skills let you evaluate speculative bets and reason about uncertain outcomes. But comparing across cause areas is a different kind of problem. You’re not comparing two malaria charities. You’re comparing preventing a child’s death from malaria with sparing thousands of chickens from suffering with marginally reducing the probability of an existential catastrophe.
Without some framework, you’re left with gut feeling. And gut feelings are systematically biased in ways that matter:
We over-weight problems that are vivid, emotionally salient, and close to us
We under-weight problems that are abstract, statistical, or far away in time
We gravitate toward problems that are culturally familiar and ignore unfamiliar ones
We anchor on whatever we encountered first
The ITN framework is an attempt to do better—to identify the features that actually make a problem more or less worth working on, and to apply them consistently.