What Is DILEMMIX? The Online Dilemma & Prediction Game for Friends

DILEMMIX in 30 seconds

The core loop is deliberately simple. You do not need a long rules explanation before the first round.

• ◆
  Choose — answer the A/B dilemma for yourself.
• ◆
  Predict — guess what the other players chose before you see their answers.
• ◆
  Reveal — see everyone’s choices and which predictions were correct.
• ◆
  Discuss — find out why somebody chose the answer you never expected.

The choice is only half the game

Most A/B conversation games focus on one question: what would you choose? DILEMMIX keeps that question, then adds another one before the reveal: what do you think they chose?

That small change turns a preference into a social prediction. You might know that one friend hates uncertainty, that another usually chooses experiences over money, or that your partner has argued for the same value a dozen times. Then a new dilemma appears and all of those assumptions have to survive one actual choice.

Sometimes they do. Sometimes the friend you were certain about picks the opposite option. That gap between expectation and reality is often where the most interesting conversation starts.

How a DILEMMIX round works

The steps below describe a multiplayer round. Solo keeps the A/B choice and comparison with global statistics, without predicting another participant. With two players, predict the other person's choice; with three or more, predict the group's majority.

• ◆
  Choose for yourself. Everyone receives the same dilemma and commits to an answer without seeing the group.
• ◆
  Read the people around you. Use what you know — or think you know — to predict their choices.
• ◆
  Reveal the answers. The group sees who agreed, who surprised everyone and which predictions were correct.
• ◆
  Ask why. The vote tells you what someone chose; the explanation tells you why.

How is DILEMMIX different from Would You Rather?

Would You Rather and DILEMMIX share a useful starting point: one question, two alternatives and a choice that is easy to understand. The difference is what the game does with that choice.

Classic Would You Rather usually ends with the answer and discussion. DILEMMIX inserts a prediction stage before the reveal, lets players score accurate reads of the group and can compare choices with global statistics.

So the simplest distinction is this: Would You Rather asks what you would choose. DILEMMIX asks what you would choose and what you believe the people around you will choose.

Is DILEMMIX a “How well do you know your friends?” game?

Partly, but not in the trivia-quiz sense. DILEMMIX does not ask whether you remember someone’s birthday, favorite film or a fact they told you years ago.

Everyone answers the same hypothetical choice. Your challenge is to predict how another person will respond now. That makes the game less about remembering facts and more about the picture you have built of someone’s preferences, priorities and values.

A friend you have known for ten years can still surprise you. Someone you met recently can sometimes be unexpectedly easy to read.

This is not a prediction game about the future

Many prediction games ask you to predict what will happen next: a result, an event or a future outcome. DILEMMIX asks you to predict something much closer to home — the choice another person is making in the same dilemma you just answered.

You are not trying to forecast tomorrow. You are testing the mental model you have built of someone else. That is why the same dilemma can feel completely different with close friends, a partner, coworkers or people you have just met.

Why predicting friends can be harder than it looks

Knowing somebody well does not make every new preference perfectly predictable. We fill missing information with memories, assumptions and sometimes our own answer.

Research by Tal Eyal, Mary Steffel and Nicholas Epley found across 25 experiments that simply trying to imagine another person’s perspective did not consistently increase interpersonal accuracy. In their final experiment, directly getting the other person’s perspective through conversation did improve accuracy.

DILEMMIX was not part of that research, and a game result is not a psychological measurement. The useful connection is smaller: making an assumption about somebody and then hearing their actual reasoning are two different sources of information. That is why the explanation after the reveal can matter more than the point.

What kinds of dilemmas can you play?

DILEMMIX is not built around one mood. The public experience moves between icebreakers, moral dilemmas, relationships, psychology, friendship and harder impossible choices, so the group can choose the kind of conversation it wants to have.

A lighter category can work when people are still warming up. A more difficult one can put loyalty, honesty, freedom, security or relationships into direct conflict once the group is comfortable disagreeing.

Full DILEMMIX packs contain 50 dilemmas. The Journal only previews a small number of real questions so the choices and reveals inside the game still have room to surprise you.

Who can play DILEMMIX?

The mechanic stays the same, but the experience changes with the people in the room.

• ◆
  Close friends — when confidence makes every wrong prediction funnier.
• ◆
  New groups — start with lighter dilemmas that give everyone something concrete to react to.
• ◆
  Couples — compare your own answer with what you assumed your partner would choose, without treating the result as a compatibility test.
• ◆
  Parties and game nights — the rules are quick to understand, so the complexity comes from the people rather than a rulebook.
• ◆
  Long-distance friends — the browser-based game can be combined with a voice or video call.

How to start in each game mode

Use the mode selector on the home screen, then press Play now. The game runs in a browser/PWA. Here is what to prepare for each of the three modes.

Choice, prediction, result: the game screens

Captured from the public web game on 2 October 2026 with the English interface, shared across all language versions of this guide. Demo A and Demo B are fictitious participants configured for this guide. Click an image to open its original size.

One round explained: a vote is not a prediction

In the captured round, A is Coffee and B is Tea. Here is exactly what we entered for the two fictitious participants.

What DILEMMIX is not

A DILEMMIX answer is not a diagnosis, moral score, personality profile or relationship verdict. It is one response to one hypothetical situation.

Mood, wording, recent experiences and the details a player imagines can all change the answer. The game works better when choices are treated as openings rather than labels.

Instead of saying, “Your answer proves you are selfish,” the more useful question is, “What made that option worth choosing for you?”

What makes a good DILEMMIX dilemma?

A strong dilemma does not need to be shocking. It needs genuine tension between the options. If one side is obviously wonderful and the other obviously terrible, there is very little to discover.

The better questions make you give something up whichever side you choose: comfort or adventure, truth or protection, freedom or security, loyalty or fairness, certainty or possibility.

The wording matters too. The two options should be comparable in tone and perceived cost so that the choice measures a real trade-off rather than which answer was written to sound more admirable.

How we build and review a DILEMMIX pack

A DILEMMIX pack is not assembled by dropping a long list of prompts into a template. Our working editorial checklist looks for balanced A/B trade-offs, repeated ideas, tone, category fit and whether both options are plausible enough to divide a real group.

The same check matters during localization. A literal translation can preserve the words and still lose the dilemma, so phrasing is adjusted when an idiom, cultural reference or difference in intensity would make one option feel artificially stronger.

Before a pack is treated as complete, we also check the intended pack size, the public sample dilemmas and cross-pack duplication so the preview represents the pack without giving away the full set.

• ◆
  Balance — both choices need a believable reason to be chosen.
• ◆
  Originality — repeated or near-duplicate dilemmas are flagged across packs.
• ◆
  Localization — meaning and tension matter more than word-for-word translation.
• ◆
  Preview integrity — public samples are checked separately from the full pack.

Editorial workshop: turn a label into a real choice

An editorial exercise created for this guide. It demonstrates a rewrite, not a claim about user testing or the history of a published pack.

Frequently asked questions