Record your answers somewhere you’ll have during the discussion (a notes app, a Google Doc, the spreadsheet copy described in Exercise 3—wherever works) and bring them to the session.
Exercise 1: Moral uncertainty and your priorities
1.1 Identify two moral views you have some credence in (e.g., consequentialist and rights-based, or present-focused and long-termist). Assign rough credences that sum to at least 80% of your total.
1.2 Pick one cause area you’re considering. Rate its choiceworthiness (-10 to +10) on each of your two moral views. Does taking moral uncertainty into account change your importance rating compared to what your favourite view alone would suggest?
1.3 What’s the single moral question that, if you could resolve it, would most change your cause prioritisation? (1–2 sentences)
Exercise 2: Cast a wider net
Before you apply ITN systematically in Exercise 3, spend half an hour generating ideas. Cause prioritisation is bottlenecked on the problems you’ve actually thought about—and your gut, left to itself, surfaces only the cause areas you’ve happened to encounter. The aim here is to deliberately surface more, before narrowing down.
2.1 Initial list (5 min). Set a timer for 5 minutes—actually do this—and write down every global problem or challenge you can think of. Don’t filter; aim for quantity. Anything that strikes you as a real problem worth working on is fair game.
2.2 Expand the lens (10 min). Set another timer and revisit the radical empathy and fringe-ideas frames from the cause area tour. For each prompt below, add new items to your list:
Problems that affect large populations far from you—geographically, culturally, or economically.
Problems that affect populations not usually included in our moral concern: factory-farmed animals, wild animals, invertebrates, digital minds, future generations.
Problems that may emerge or compound over the next century, even if they’re modest now.
Problems that pattern-match to “weird.” Sit with the discomfort and ask why.
2.3 Borrow from others (10 min). Other people have spent years thinking about which problems matter most. Skim some of these lists and add anything new to yours—don’t try to read them properly, just scan for ideas you missed:
Don’t try to narrow yet. The goal is breadth. You’ll narrow in Exercise 3.
2.4 Pick one for the spreadsheet. From your expanded list, pick the cause area that strikes you as weirdest or most outside your usual sphere of moral concern. That’s the one to drop into the “A cause that strikes you as weird” row of Exercise 3’s spreadsheet, where you’ll evaluate it on the same terms as the canonical cause areas.
Exercise 3: Your cause prioritisation
Complete this exercise using the template spreadsheet. The link will prompt you to make your own copy—sign in to a Google account, name your copy (it’ll save to your Drive), and fill it in as you work through the exercise. Bring your copy to the session.
Work through this systematically. There are no right answers, but there are well-reasoned and poorly-reasoned ones.
Throughout this exercise, practise reasoning transparency (Week 1’s skill): show your working, indicate your confidence level, and make your assumptions explicit. Imagine explaining each answer to a thoughtful friend who might disagree with you. If they’d need to ask “but why?” then you haven’t been transparent enough.
The template has three tabs:
Tab 1: Your inputs. Enter your moral weights (how much does a chicken's suffering count relative to a human's? how much does a future person count?) and your empirical beliefs. For AI and engineered pandemics you give two probabilities each — the chance of a catastrophe this century (affecting people alive today) and the chance of an existential catastrophe that permanently ends humanity's future. You also estimate climate change's existential risk (a contested figure — most experts rate the direct risk low, but the column is there for your own view), and the chance that focused work improves the long-term trajectory if we survive. A given figure for the number of future people (~100 trillion) sits there too, for you to scrutinise and change. These are the inputs that drive everything else.
Tab 2: What your inputs imply. This is where it gets interesting. The spreadsheet takes your inputs and applies a deliberately crude formula to each cause area, splitting its importance into two parts. Near-term importance counts the beings alive this century (scale × moral weight × severity × probability of catastrophe) — where severity is how bad the harm is per person, 100 for a death and lower for non-fatal harms like homelessness or poverty (it works like a DALY weight from Week 2). Long-term importance counts future people — about 100 trillion if humanity survives — at your future-person weight, times the chance the cause is existential (for AI, biosecurity and climate) or shapes the long-term trajectory (the "better futures" row, which is long-term only). The two add up to a total. The scale figures (e.g., 80 billion animals killed per year, 600,000 malaria deaths per year) are given; your moral weights and probabilities come from Tab 1. Watch what happens to AI safety, biosecurity and climate as you raise the future-person weight or the existential probabilities — the long-term part can dwarf the near-term one. That is the longtermist multiplier from the lesson, made visible.
The formula is a thinking tool, not an oracle. It shows you what your stated beliefs mathematically imply about relative importance. Sometimes the result will match your intuitions. Sometimes it won’t. Where it doesn’t, something interesting is happening — either the formula is missing something, or your intuitions haven’t caught up with your beliefs.
After reviewing the raw numbers, you rate each cause area on importance (1–10), tractability (1–10), and neglectedness (1–10). Your importance rating doesn’t have to match the formula output — but if it differs significantly, you should be able to say why.
Column C also lists approximate annual funding for each cause area (combining government and philanthropic spending). Use this as an anchor when rating neglectedness: a problem with ~$200M against ~80 billion affected animals is meaningfully more neglected per unit of scale than one with ~$50B against ~1.2 billion people. Funding alone isn’t the whole story (think about whether new resources can usefully be deployed given the field’s state), but it’s a useful sanity check.
Alongside the cause areas from the tour, the table prefills a few everyday, closer-to-home causes — cancer in the Netherlands, homelessness, conflict, poverty in the Netherlands — with rough figures, so you can see how the causes you care about locally compare on the same terms. (The figures are rough and editable, and the severity column lets you reflect that a non-fatal harm like homelessness or poverty counts for less per person than a death.) And if you want to evaluate a cause area that isn't listed — nuclear disarmament, mental health, wild animal suffering, whatever interests you — there are blank rows where you fill in all the inputs yourself. If your cause matters mostly for the far future (digital minds, deep-future people), put your estimate straight into the long-term column.
Do: Research sprint (optional but recommended, 30 mins)
Before filling in your ratings, consider investing 30 minutes in testing your assumptions. Pick one cause area — ideally one you’re uncertain about — and do a rapid research sprint:
List your key uncertainties (5 min). What would most change your assessment? What assumptions haven’t you checked?
Rapid research (20 min). Focus on your biggest uncertainties, not a comprehensive overview. Use the 80,000 Hours problem profiles, Our World in Data, or GiveWell as starting points.
Synthesise (5 min). How has your assessment of importance, tractability, or neglectedness changed?
If you’re very unsure about a cause area’s score, that’s a reason to investigate it more, not less.
Tab 3: The gap. The spreadsheet generates a ranking from your total importance scores (near-term plus long-term). Here you write your actual ranking — the one you'd defend if pressed. The two rankings are shown side by side.
Where they match, your analysis and your judgement are aligned. Where they don’t, there are three possibilities: the formula is missing something important (tractability? neglectedness? something else?), your gut is tracking information you haven’t made explicit, or your intuitions are resisting the implications of your own beliefs. Any of these is worth examining.
3.5 Reflection
In your spreadsheet copy (or somewhere you can find again), record your formula-generated ranking and your actual ranking, and write a few sentences on where they differ and what the gap tells you about beliefs you hold that you haven’t made explicit. Steelman your bottom-ranked cause in 1–2 sentences: what’s the strongest case for prioritising it? Bring all of this to the discussion.
3.6 Crux identification
Identify the single belief that your ranking is most sensitive to. If this belief changed, how would your ranking shift?