Prisoner of War
Artist Statement:
Inspired by classic role-play games like Spyfall and Mafia, Prisoner of War is a fast-paced party game of misdirection, secret alliances, and deduction. Designed for 5–10 players, it’s perfect for any setting—from breaking the ice with new people to spending a cozy and fantastic night among close friends. In this game, you might be:
- A Prisoner, trying not to reveal the secret password while figuring out who the Mole is, in hopes of escaping;
- A Mole, pretending to be a Soldier while secretly working to help the Prisoner escape; –
- Or a Soldier, trying to uncover the password before it’s too late—and before the enemy hiding among you strikes.
The Prisoner’s password falls into one category selected from a pool of potential options, giving each game a unique dynamic. Meanwhile, every Soldier must follow a secret behavioral rule, which can help reveal whether a Mole is among them; while also making the process of uncovering the password more challenging.
The game creates a highly engaging environment where every player is drawn in and suspicion runs high, all within a playful theme of war and loyalty. At its core, Prisoner of War isn’t just about escaping captivity; it’s about navigating human dynamics when the stakes are high and no one can be fully trusted.
Concept Map:
Initial decisions about formal elements and values of your game:
“Prisoner of War” evolved from our exploration of social deduction mechanics and our goal of creating fun moments of collaborative problem-solving. Our initial formal elements were shaped by several main design values that guided our development process:
Players & Objectives: We started with a two-team structure separated in different physical spaces where each team had a Prisoner (sent to infiltrate the opposing team), Soldiers (defending their team’s password), and a Mole (secretly working for the opposing team). This created competing objectives that generated natural conflict: Soldiers working to identify the Mole while the Prisoner and Mole collaborated covertly to discover and communicate the password. After playtesting, we discovered that consolidating to a single unified team significantly improved gameplay dynamics and we eliminated the logistical challenges of coordinating between two rooms. This change ensured that all players remained actively engaged throughout the entire game session.
Rules & Boundary Design: We deliberately constructed rule constraints that would force creative communication and strategic gameplay. For Soldiers, we implemented a diverse set of behavioral limitations that created interesting decision spaces. As shown in our game materials, these included some linguistic constraints (“Cannot use the word ‘The'”), speech pattern requirements (“Must start your guesses with ‘umm'”), physical behaviors (“Always tilt your head slightly while asking your question”), or communicative restrictions (“No plurals allowed”). Each rule introduced a meaningful pattern that clever Prisoners could potentially identify while also creating plausible deniability for the Mole to operate.
For the Mole, we established boundaries that created genuine tension so that they know the password category (e.g., “Countries,” “Colors,” “Fruits”) but must communicate it subtly without being detected by the other Soldiers. Our role cards explicitly instructed Moles to “try to show them you are the Mole through the guesses you make – just don’t be too obvious!” This deliberately vague guidance created an exciting challenge where Moles had to judge exactly how subtle or obvious their hints should be based on the social dynamics of each particular game.
These constraints created rich opportunities for creative problem-solving and social observation. For example, in one playtest, a Mole assigned the “Vegetables” category demonstrated remarkable ingenuity by casually mentioning garden-related terms while carefully avoiding the prohibited behaviors that would expose them. Meanwhile, the Soldiers’ varied rule constraints (like “Always ask questions in singular form” or “Speak only in passive voice”) created distinctive patterns that observant players could leverage for deduction. Furthermore, we had a time limit in the game, ranging from 1.5 to 2.5 minutes depending on player count), and the turn-based structure ensured that everyone participated equally while the Discussion/Voting Phase created opportunities for social deduction and emergent storytelling as players debated who might be the Mole based on their observations.
Theme Integration: The prisoner of war interrogation theme serves as the narrative backbone that gives meaning to our core game mechanics. We built each game element to connect naturally with this theme. For instance, the timed questioning creates the real pressure of military interrogations. The voting to remove suspected infiltrators mirrors field judgments during wartime and the limited communication rules reflect how information is controlled during conflict. This theme allows players to understand their roles without detailed explanations and makes it suitable for people to play at their own pace. We hypothesized that soldiers would naturally take charge during questioning, moles would try to send secret signals and show signs of being tense, and prisoners would experience the pressure of answering honestly while protecting vital information. From our playtesting results, we observed that players quickly embraced these roles. Some began to use military style questioning that came naturally from the theme. In fact, some became very analytical in their questioning while others relied more on intuition and social cues. This showed that our war theme was not just decorative but actually shaped how players interacted with each other in meaningful ways. The wartime setting naturally raised the stakes which motivated players to engage deeply with the psychological elements of deception, observation, and trust.
Throughout our design and iteration process, we prioritized mechanics that created genuine social interaction over complex rule structures. We valued approachability (simple enough for new players to grasp quickly), replayability (through varied password categories and Soldier rules), and designing moments of dramatic tension (through timed interrogations and decisive voting). The resulting game creates a psychological battlefield where careful observation and strategic communication determine success. Ultimately, “Prisoner of War” goes beyond just being a game to show real human behavior by revealing how people handle trust, doubt, and deception when they don’t have all the information.
Testing and Iteration History:
For our gameplay, we had 3 playtests from which we derived a great amount of feedback to tailor our game. In each of these three rounds, we mainly had classmates playtest our game in batches of 6-8.
As an overview of the improvements we made from one gameplay to the next:
Round 1:
- Changing from 2 groups in playing in sound-isolated areas to 1 large group
- Clarified category vs. password vs. rule
- Role parameter clarifications
Round 2:
- Introducing timed guesses as opposed to having a limited number of guesses
- Introduced required worldbuilding for people to start playing as their ‘roles’
- Personal rulesheets on each person’s character for faster first-time games
Round 3:
- Adding time limits for guessing and voting
- Aggregate rulesheet for players to refer to as needed
- Sequential voting (as opposed to a general group voting)
In the first gameplay, we split our play-testers into two groups. Each group had 1 mole, 1 prisoner, and remaining students as soldiers. Students were very excited about playing roles but were less clear on the various parts to the gameplay – rule-reading going into interrogation then debate and finally voting.
For our moderator and notetaker, there was a lot of difficulty in managing and engaging all participants when they were split into two isolated groups who were both learning the game in its first form. The interaction points between the two groups were not clear, and groups’ activity periods were not necessarily in sync. For example, one group finished its voting before the other, creating an empty period where the first group had to just sit and wait for the second group to finish deliberation. Additionally, there was confusion around what happened if win conditions were satisfied in both groups at around the same time and without a narrator/moderator role, the two groups truly felt disconnected in their play rather than the intended team 1 vs. team 2 feeling. Due to this syncing and team dynamic issue, we decided to combine both groups into 1 large aggregate group and keep only 1 mole and 1 prisoner for the entire game.
There were additional difficulties with in-game terminology on category vs. password vs. rule and how each of these items added to your goal as a player in the game. We clarified this as having just a password and a team rule, and also clarified that the prisoner is not allowed to lie during questioning.
For the second gameplay, we saw more of the fun shift to the actual gameplay since everyone was playing in one big group. The interrogation went slowly but the debate and voting portion was incredibly lively. This being said, the prisoner had to ‘grow’ into the role through reminders that they could not lie and did not have to answer questions that went outside of the parameters of what they could answer (yes/no/maybe questions).
During this round, we once again saw a period of non-activity from most players while the current guessing player used up their 2 allotted guesses and the prisoner took their time in replying. All players reported that this period was when they felt least engaged and were having the least fun, so we iterated upon this by adding a timed guessing period in which players should go around and try to get as many guesses in as possible. Additionally, we increased the worldbuilding aspect by giving role-cards with character names, backgrounds, and a personal rulesheet (to ease playing for first-timers). This second iteration was where we did the greatest amount of serious development for the game to become real and more professional.
For the most recent (third) gameplay, the fun was firmly shifted to the gameplay, as opposed to the role-based premise of the game. There was again a slight issue of players not necessarily using the right guessing format (asking questions as opposed to naming nouns in the category). We also ran into an interesting problem where the win condition was met in 1 round of voting every time we restarted the game. That being said, this gave the space for moles to learn from each other’s strategies and become ‘better’ moles as each round finished. They started modeling soldier-like behavior at the beginning of guessing so that suspicion could be thrown onto innocent soldiers who were too close to the prisoner’s category.
To consolidate game materials, we introduced a full rulesheet as opposed to personal rulesheets for each player. We also added a time limit on voting so that games could no longer be won in one round – it is harder to get the mole or password every time if you have a rushed period of deliberation. Additionally, to prevent further group collusion, we implemented sequential voting, where every player casts their individual vote and thus cannot attempt group consensus through debate or live group voting.
Through consistent feedback from our 3 rounds of gameplay, we were able to successfully shift the participants’ fun into playing Prisoner of War and to create consistent engagement/dynamic play for all participants as much as possible. We tempered rising issues through the introduction and refinement of gameplay rules, and added worldbuilding to make a casual party game based on deception.
Link to Final Prototype: https://www.figma.com/design/eXhl3Jgu14HTL3OuwZWbwW/Prisoner-of-War–Print—Play-?node-id=0-1&t=GNtjo3elYbac5uSa-1
Video of Final Playtest:
