I find social deduction games one of the harder types of tabletop games to design and in this blog I’m going to explore some things I’ve learned in analyzying and designing them. Once upon a time, I had a hard time designing a hidden traitor game and had to go through many different ideas before I got one to work. While I would argue hidden traitor and social deducation are different genres, they are adjacent and share design concepts.
I was initially hesitant to revisit social deduction but came back to it because one of my designer friends had me look at a social deduction game he designed that didn’t seem to be working. In trying to help him fix it, we came up with a new concept which I hope to bring to the table soon for testing. Fingers crossed; hopefully, this new idea will work.
Designing Information
The tricky part of designing social deduction/hidden traitor is that you have to design the information that is available to the players. The information shouldn’t be too obvious or else the bad guy will be easy to deduce. It also shouldn’t be too random that the good guys will have no idea how to figure out who the bad guy/s are. The information should be at a sweet spot between these two extremes which is easy to say but hard to do.
A Visual Framework for Information
One way to imagine the possibilites over the game’s play time is a funnel. At the start of the game the funnel is wide because anything is a possibility. As events in the game happen, the funnel gets smaller and smaller as more and more information is revealed until finally the funnel collapses unto the truth.
The goal of the good players is to collapse the funnel as quickly and as accurately as possible while the goal of the bad players is to keep the funnel from collapsing. If the bad guys are especially good, they can even collapse the funnel to somewhere outside of the truth to really mislead the good guys.
Events and Scenarios
Events are in-game happenings that give information to the players. Events are explained by Scenarios so for example Event A can be explained by possible Scenarios B, C, D, etc, of varying probabilities. So it’s possible that Scenario B is a more likely explanation to Event A as compared to C and D but it does not rule out C and D completely.
Events collapse the information funnel because when an event happens in well-designed social deduction games, some ecenarios become less likely or even can be outright eliminated. As more and more events happen, the funnel narrows down.
Practical Design Tips
Here are some things I’ve discovered designing social deduction and hidden traitor games:
Don’t design Events with 1 or 2 Scenarios
Every event in the game should have more than two scenarios on why it occured. If there is only one explanation, the game is solved and we don’t want that. However, if there are only two explanations, it becomes very easy to collapse the probability funnel.
If there are only two scenarios, it becomes easier for players to compare the two and pit them against each other directly. For example, if an event can only be explained by scenario A or B, it can be easy for players to debate A vs B and find that A has a 40% chance of occuring while B has a 60% chance of happening. Compare that to if you add a scenario C and it becomes harder to directly compare A and B because you also now have to compare A and C, and B and C. (I’m not sure if there’s a psychologic explanation about analyzing binary decisions and if someone knows please comment it below).
Avoid Events with too many Scenarios
On the other hand, if there are an infinite number of explanations to an event, the funnel does not collapse at all and doesn’t move the game forward. Too many scenarios will make the players will feel apathetic to deducing the truth. They won’t feel like the fun of the game is worth the effort to solve it.
How many scenarios per event is just right?
This is a hard question to answer and it probably depends on your design. You’ll only really find out once you playtest. My gut feeling is around 5-8 would be a good starting goal then temper it. Also, the number of scenarios per event isn’t a hard rule that you have to keep track of. (Please don’t count the actual scenarios per event in your game unless you really have to deep dive to find what’s wrong). It’s more of a mental model that helps you guide your design and diagnose if there’s a problem. If the game is too easy to solve, muddle the information more. If the players feel too stressed trying to figure things out and don’t want to play, then tighten the information.
Some considerations: If your social deducation game is short, then the scenarios should be less. You want the probability funnel to collapse faster (but not too fast). The number of scenarios also depends on your target player. More casual games probably should have less scenarios so it’s easier for the funnel to collapse while more hardcore games can have more nuanced and hard-to-analyze events so that the players have more fun analyzing each event and figuring out possible scenarios.
Events are better if Player Choices hint at their Roles
Events in social deduction games have to help people judge the role of a player/players. I think the best way an event can do that is by giving a player a Choice that leads to scenarios which can then be explained by their role and possible intentions. The best kinds of events ask the question “Why did the player CHOOSE to do that?” as opposed to just “Why did the player do that?” which can be driven by chance, the board state, or something else not within the player’s control.
As a lame example for a decision driven by chance (just to illustrate the point), let’s say a deck has three kinds of cards with these abilities:
- “If you are good, discard this card facedown. If bad, keep it.” – 10 copies
- “If you are bad, discard this card facedown. If good, keep it.” – 5 copies.
The player draws a card and keeps it which gives information about their role and hints they are good. However, the Event did not give the player any agency and decision making. It gave information on their role, but not on their intentions. There is no way the player may try to prove they are good or pass as good if they are bad which is no fun for the player and isn’t fun for the other players who are judging them either.
To expound on the lame example, what if the abilities were:
- “If you are good, discard this card facedown. If bad, you may keep it or discard it.” – 10 copies
- “If you are bad, discard this card facedown. If good, keep it.” – 5 copies.
This adds the possibility of choice and intention. While still having an element of luck, now the player’s intentions are a possible explanation to the event. The other players will also have a more fun time dissecting the player’s intent.
Player choices are viable no matter what role
When a player makes a choice, it has to be a viable option to do whether that player is good or bad. If a choice only makes sense to do if you are good, then there’s only one scenario you’d do it which means your role will be too easy to deduce. Same is true but change good to bad.
One thing you can do is review the actions in your game and imagine yourself playing. Imagine you are a good player, are there multiple scenarios to choose that action outcome? List those scenarios. Now imagine you are a bad player, iare there also multiple scenarios to choose that action outcome? List those scenarious too. If your list for either role is not very long or one list is way longer than the other, then the scenarios may be limited or skewerd towards one role which makes it easy to deduce.
Multiple Levels of Obfuscation before revealing Information
One way to muddle the information of an Event is to make it made up of many sub-events before the information is revealed. A good example of this is in Secret Hitler when the President draws cards from the deck, discards one, then hands them to the Chancellor to choose. A lot of steps and scenarios can explain the final result. Was it the President who was a Fascist and filtered the original cards drawn? Was it the Chancellor who chose the result? Were the two working together? Did the President have no choice because they were liberal but drew 3 Facist cards? Etc.
Sequencing sub-events together is a relatively easy way to blur the truth because each of the sub-events can be relatively simple and, on their own, have very few scenarios. For example, if in Secret Hitler the rules were that the President draws 3 cards and then they chose to enact a Fascist policy (skipping the Chancellor part), there are only two possibilities: The President is a Fascist and chose it or the President chose only drew Fascist policies. Based on our previous discussion this is a binary decision which collapses the funnel too quickly. But by sequencing multiple sub-events together without revealing information, the number of scenarios goes up dramatically.
Don’t make information hard to remember
Unlike in other genres of games where you can have reminder cards and where the rulebook can be referenced, in social deducation often the act itself of referencing the rulebook or a reminder card can be suspicious. Information has to be kept as simple as possible to help the bad team in their lies. If they are caught forgetting information, they will be easily outed!
Some Design Spaces to play around with
Here are some design spaces in a social deducation game you can play around with. I made this list designing my own social deduction game and it’s not exhaustive but hopefully helpful.
Roles
Roles can be static like The Resistance or Secret Hitler. They can also be dynamic like in One Night Ultimate Werewolf and Blood on the Clocktower meaning that a player’s role can change (with or without them knowing about it).
Roles can have abilities (like in One Night Ultimate Werewolf or Battlestar Galactica) or not.
Do the people in the teams know their teammates? Usually bad guys know each others and good guys don’t.
Roles can even be chosen mid-game like in Cuba de Mafia.
To be honest, I didn’t explore Roles as a design space too much because in both games I designed the theme required them to just be static and ability-less. Comment below if you have any tips on how to utilize dynamic roles and roles with abilities!
Group Action Results
Often social deducation/hidden traitor games have a group action like a Vote or a result that involves a subset of the players (like missions in the Resistance, Avalon, and Crisis <I forgot if this is the right term> in Battlestar Galactica).
Some design spaces you can play around with for Group Actions:
- Who is part of the group action? Is it all the player or just a subset?
- A subset group action can be a very interesting way to find information about players because multiple subsets can be tested for different results. However, you usually need to hide the individual contributions in each subset (like in shuffling result cards in The Resistance).
- Who picks the subset and why would they do that? Can they be trusted?
- Is the result muddled in a way? Some games shuffle the results to hide individual contributions to the action while others have it out in the open.
- Does the group pick the result (like in a vote) or does the group pick a representative who will then be the one to pick the result (like voting in the President in Secret Hitler)?
- Games where you pick a representative often have a mechanic which punishes the good team for not being able to pick a representative which gives the bad team an additional strategy of just confusing the discussion.
Manipulating Information to catch lies
Players can manipulate the board position not directly to help or hinder their team but just to gain information on other players. For example, in One Night Ultimate Werewolf, there are many cards that allow you to peek and move roles. Just be careful in doing this, make sure there are still multiple scenarios that can explain events if information is outright revealed. For example, a player may say they peeked at a role and saw them as a bad guy, but maybe the player themselves is lying.
Actions that both Help and Harm the good team
One trick I did in my hidden traitor game was that most of the actions players can do both help and damage the good team at the same time based on the situation. Situational awareness would then be the way for players to deduce if the player was actively trying to help or harm the good team. For example, let’s say the good team uses cards as resources but when the cards run out, the good team loses if they haven’t yet won the game. You can have an action where you draw cards. This helps the good team by giving more resources but hastens the clock to the lose condition. Based on the board state, either can be more helpful, harmful, or even both at the same time based on the player’s justifications of why they chose to do that action.
Conclusion
I hope this helps you design a social deduction or hidden traitor game. If you have any other suggestions or violent reactions, let me know in the comments!
