A moral dilemma behaves differently in a group than in your own head. Alone, you settle it in ten seconds. Put it in front of six people and it stops being about right and wrong. It becomes about the six people.
That is the good part, and where these evenings fall apart. Groups agree too fast, or one person tells a story and the room goes quiet, or a question lands on something somebody at the table is living through.
Below are 32 questions in four tiers, ordered by what they cost the room up to the last tier, then the mechanics: group size, turn order, and when to stop.
Sort by cost, not by topic
Most lists group by theme. Fine for reading, bad for running. What matters in a room is how exposed a question makes people, and that rises steeply. Open with something that costs nothing and people are talking by the third question. Open with the hardest and you get polite half-answers for an hour. Cost rises through tier 3, and then tier 4 trades it back: those questions are harder to answer but they stop being about anyone at the table, which is why they work late in an evening and in a classroom.
It helps to know why sincere people land in different places. According to Moral Foundations Theory, morality rests on a small set of innate systems, listed as of 2023 as care, fairness as equality, fairness as proportionality, loyalty, authority and purity. Cultures and individuals build different virtues on top of them. That is why one table can hold opposite verdicts, both sincere.
Tier 1: everyday ethics
Low stakes, high disagreement. The warm-up.
1. Is it wrong to fully recline on a six-hour flight when the person behind you is very tall?
2. A neighbour's parcel lands at your door by mistake and it is something you have wanted for months. How long before staying quiet becomes theft?
3. Is it acceptable to say "I am five minutes away" when you have not left the house?
4. Is it wrong to pass a friend's streaming password to your sister?
5. A restaurant clearly had one bad night. Do you still post the honest one-star review?
6. Is it wrong to bring your own snacks into a cinema that survives on snack sales?
7. Someone cuts in front of an elderly man who says nothing. Do you have a duty to speak?
8. Is it acceptable to lie about your child's age for the cheaper ticket?
Tier 2: money and loyalty
Now people have something at stake, and answers reveal habits rather than opinions.
9. Your employer overpays you by a month's salary and payroll has never caught an error. Must you report it?
10. Is it wrong to accept a generous gift you know the giver cannot afford?
11. A friend asks you to lie to her landlord so she can sublet. She has nowhere else to go. Do you?
12. You are self-employed. Is rounding expenses in your favour a small habit or a small crime?
13. A shop assistant gives you a discount you are not entitled to. Do you say anything?
14. A close friend pays a cleaner well under the going rate and the cleaner is happy. Is it your business?
15. Someone you love is in debt over a habit they will not name. Do you keep lending?
16. Five friends split a lottery win. One did not chip in that week, though he always does. What is fair?
Tier 3: work
Handle with care if colleagues are present.
17. Your company's product is legal and mildly harmful. Is taking the paycheck a moral choice or just a job?
18. A colleague is struggling and their work keeps landing on your desk. When does covering stop being kindness?
19. You are told to interview a candidate for a role already promised to someone else. Do you tell them?
20. Is it wrong to use a job offer you would never accept to get a raise?
21. A supplier sends you a large gift in December. Where is the line between a gift and a bribe?
22. Is it wrong to take a sick day when you are not ill but are close to breaking?
23. A rival is about to hire your best teammate, and you know the rival is nearly out of money. Do you warn her?
24. Your best hire lied about a degree they never finished. Two excellent years later, do you act on it?
Tier 4: the hard ones
Abstract, which makes them safer than tier 3 in a mixed room. Save them for late.
25. If a button ended one stranger's life and cured a disease that kills thousands, would pressing it make you a murderer?
26. Is there a real moral difference between letting someone die and killing them, when the outcome is identical?
27. Do you owe more to one person in front of you than to a hundred you will never meet?
28. Should anyone be punished for what they were always going to do, given their genes and their history?
29. Is it wrong to bring a child into a world you believe is getting worse?
30. If a promise was made to someone who has since died, does breaking it harm anyone?
31. Is it possible to do something genuinely wrong that harms nobody?
32. Should someone who has truly changed still owe a debt for who they used to be?
Curious how strangers split on cases like these? BeJudge hands you one, takes your verdict, and shows the global vote.
How many people, and what breaks above eight
Four to seven is the working range. Below four you get a debate with two sides and no surprises. Above eight, quiet people stop contributing and loud ones start performing.
The second problem is the expensive one. In a Science study of 699 people working in groups of two to five, Woolley and colleagues found that groups where a few people dominated the conversation scored lower on collective intelligence than groups with a more equal distribution of conversational turn-taking, at r = -0.41 in the subset of groups measured with wearable badges. Collective intelligence was not strongly correlated with average or maximum individual IQ.
With ten or more, split into two pods, run the same three questions in each, then compare. The comparison usually beats either one.
Turn order matters more than the questions do
The thing to design against is drift. Research on group polarization, published in Personality and Social Psychology Bulletin, describes it as a change in individual preferences from before to after group discussion, moving in the direction the group's average preference already favoured. If your first two speakers lean the same way, you spend the night measuring the drift instead of the disagreement. So take answers privately first, reveal at once, then discuss.
Two more mechanics earn their keep. Rotate who reads, so nobody owns the agenda. And on any sharp split, ask one person to argue the side they did not pick.
For a version with verdicts and scoring attached, the AITA format does that job and needs no setup.
Keeping it off the people in the room
Somebody answers a tier 2 question and everyone realises they are describing their brother. Two rules cover most of it. Keep answers in the third person until the question is finished, so people reason about the case instead of defending themselves. And never build a question from something someone present is going through.
Skitka, Bauman and Sargis, writing in the Journal of Personality and Social Psychology in 2005, found that stronger moral conviction predicted greater preferred social and physical distance from people who disagreed, lower good will and cooperativeness in mixed groups, and a greater inability to generate procedural solutions to the disagreement. When a question hits a real conviction, the room does not get sharper. It gets further apart.
Shared vocabulary helps. The Markkula Center for Applied Ethics publishes a framework for ethical decision making built on six lenses: rights, justice, utilitarian, common good, virtue and care ethics. It says plainly that different lenses may lead to different answers, each still offering insight. With that vocabulary, "you are wrong" becomes "you are using justice and I am using care," which is survivable.
Some splits trace further back than the lens, to where people grew up: see cultural differences in moral judgment.
When to stop
Three stopping rules, in order.
Stop the moment someone's real situation surfaces. Not the question, the round. Acknowledge it, move on, come back another night if they want to.
Stop when an argument repeats for the third time. Nobody is changing their mind; they are restating for the scoreboard.
Stop on a split, not a resolution. Pushing until the group agrees produces the worst version of the night. A room that ends 4-3, everyone slightly surprised, is a good result. Forty-five minutes and six questions beats three hours and thirty.
The point is not to win these. Hearing a friend calmly defend a position you find unreasonable changes what you think that friend is like, usually for the better. That effect is the subject of building empathy through gaming, and it is the real return on the evening.
If you would rather not host it, BeJudge gives you a case, takes your verdict, then shows how the rest of the world ruled. A good way to find out how unusual your instincts are.
FAQ
How many people do you need for moral dilemma questions?
Four to seven works best. Fewer than four and the group rarely splits, so there is little to discuss. Past eight the quieter people drop out, and at ten or more you are better off splitting into two pods and comparing answers afterwards.
What makes a good moral dilemma question for a group?
Both answers have to be defensible by a reasonable person, and the facts concrete enough that nobody can redefine the question halfway through. If everyone answers the same way in two seconds, it is a quiz.
How do you stop a moral dilemma discussion from turning into a real argument?
Keep answers in the third person until the question is finished, never build a question around someone in the room, and end the round the moment somebody's real situation surfaces. Committing privately before anyone speaks also stops the group clustering around the first opinion voiced.
Should the group reach a verdict at the end?
No. Chasing consensus pushes people to overstate their positions and flattens the disagreement you came for. A group that ends divided, with everyone clear on why, has done the exercise properly.
Are these questions suitable for classrooms or teams?
Tiers 1 and 4 are, being low stakes or abstract. Tier 3 is risky with colleagues, because the questions map onto decisions people at the table may have made. Pick tiers to match how much the group can afford to reveal.