Category: Gut Check

Gut Check

I was a programmer and designer on this game made with Isabella Djurovic and Jordan Jones-Brewster.


Play on

Gut Check is an audio-based trivia quiz show game in which a questionable AI forces the player character to answer questions about a randomly assigned topic (subtly hinting at the player’s situation) and listens for their spoken answers. The game was made in Unity with Windows 10 Speech Recognition and tells the story of a rogue AI who captures humans to play her twisted trivia game show with the hopes that they will win and take her place as the captive host AI. I programmed the game, created the art, and participated in the design and writing with Jordan Jones-Brewster and Isabella Djurovic. Our design goal was to create feelings of intrigue, mystery, and frantic uncertainty in players. We wanted a game whose narrative was the player’s driving force of progression – that players keep playing because they must know more about the game’s world and central conflict. We also put a technology constraint on ourselves: we wanted a mainly audio experience. Over sixish weeks we designed and iterated a quiz show system that fit the limitations of a regular, at home player but still felt like a grand game show experience. Due to the format of the trivia show, we knew players would often fail and a usual playthrough would probably not be very long. We identified this challenge and addressed it by creating a highly replayable game in which failing is fun enough for player to want to play through the game multiple times to “unlock” all the death scenes while still being rewarded for successfully answering a question. Technology limitations and question difficulty curves led us to change the quiz show format to better suit the needs of the game’s design goals.

Voice recognition tech shaped the way we could design the story for the player. Initially we wanted to use the Amazon Alexa as the basis for the quiz show. Since Alexa already has a question/answer convention when people use it, we thought the format could serve as an affordance for players so we could focus on telling the story instead of teaching the player to use the tech. Alexa could also serve as a central character that players would already be introduced to. Making Alexa the antagonist would therefore be an interesting relationship for the player to develop over the quiz show. However, after researching the costs of hosting the app online and gauging the time it would take us to learn the programming language required (new to all group members), we decided to use a similar tech that we knew we could develop for. Instead of Alexa, we used Unity’s Windows 10 Speech Recognition software in order to make the game faster and iterate more. The speech recognition system could easily pick up on phrases and was simple to develop for, however it did not come with a recognizable character like Alexa. We decided to keep this antagonist structure however, and gave her a computer voice to mimic the original idea of a rogue AI. Now using Unity, we had the option to include visuals along with more sound files. We used this to our advantage by creating a flashy, animated landing screen for players to look at as they play. We went with a game show sign and button to let players know immediately what kind of game they were playing. However, we still considered Gut Check to be an audio-based experience. We integrated sounds and sound mixing into the antagonist AI’s sound bites to “animate” them. We found this gave the character a sense of omniscient power – that this is her world and she has total control over it. This occured for players during the split seconds it took them to listen and decode the sounds they were hearing – seemingly all coming from the antagonistic AI.

Changing to Unity also allowed us to easily use the engine to create a more responsive, replaybale game. We knew that players would die often by getting questions wrong. We wanted this experience to be fun – a secondary reward for playing. To achieve this we introduced two elements of randomness to the game: the question topics and the endings. There are six topics each with four questions. We kept the number of questions low to increase the player’s chances of winning by reducing the failure space. We wanted them the keep playing, so we randomly assigned each playthrough a topic. Every time the player plays, they have a significant percentage chance of getting a topic and series of questions they haven’t encountered before, keeping them interested. The ending system functioned similarly. There are 15 endings from which the game randomly chooses each time the player dies. The chance of the player getting the same ending twice in a row is slim, so each ending feels unique. These randomness systems gave players a fresh set of content to experience, while also adding to the world design. With her (perceived) breadth of knowledge about many topics and her more than a dozen ways to kill and torture the player character, the central AI seems more powerful and at the center of the world. The player learns through these randomly assigned topics presented in a repeated way that although the AI is a piece of tech that must follow rules, she is on the verge of break them and knows a great deal about how this situation will go down. She is in control, not the player.

Our difficulty curve of questions helped us design a satisfying sense of pacing in both gameplay and narrative. In general, the questions we chose for each topic do not have obvious answers. We wanted the fun of the experience to be in failing and being rewarded with a plethora of randomly distributed, gory endings. Replay was important, but not repetition, as players quickly bored of our first overly long introduction sequence. To give players a break, we decided to give the player thirty seconds to answer a question. If they called out a correct answer, the game automatically progressed. If the answer was wrong, they could continue to guess until the time was out. This allowed players time to ask a friend or frantically attempt to google an answer. We felt like succeeding at this during the time limit was an accomplishment that we should reward, and enhanced the reliance on tech and its evils theme. When players got an answer correct, they were given a short audio response that hinted at the antagonist AI’s motivations and how the “world” of the Gut Check universe operated. We designed these hints to create more questions than answers in order to be a satisfying and intriguing reward to propel the player to want to get the next answer correct. We also recognized that these moments were the only moments players got to rest during the game, which when placed in between each question in a predictable pattern, created an interesting rhythm of tension and release/reward for players. This cycle was also a contributing factor to the replay value of the game. Players liked the ebb and flow and would play the game again trying to stay in that state longer.

Voice recognition tech (and adapting to its limitations) and the difficulty of our puzzles deeply influenced how we created Gut Check as a trivia show experience that gives players a sense of mystery and suspense. We used insights into the technology to drive the format of the game and how it was presented to players. This informed how we structured our narrative in parallel with the quiz show layout; we used the pacing of questions and feedback (conventions of the genre) to introduce questions in the narrative that players wanted to progress to answer. We emphasized the emotions players felt from the gameplay – frantic uncertainty and triumphant, hesitant relief, with more subtle narrative based emotions – mystery, compulsion, the hunger to understand and control knowledge of a situation. We used both emotion sets to propel to the player towards the ending goal and keep them interested in attaining it. We wove gameplay and narrative into a cohesive system that took advantage of trivia show conventions to provide players with a compelling narrative game. Through our two randomness systems, topics and endings, we also addressed the shortcomings of the format – that the high failure rate of typical players during quiz shows leads to ultra short playthroughs. The randomness of the system rewards replays by giving players fresh content in a known system. The random topics and endings also provide the player with more information about the world, more juicy death scenes, and more insight into the rogue AI antagonist. Replay value, volume of content, and the tension and release cycle of the way the questions and antagonist motivation are paced are all narrative-mechanical devices Gut Check uses to innovate on the quiz show format.


Gut Check – Computer Narrator and Design Pivot

I am a programmer/animator/designer on this game that I’m working on with Isabella Djurovic and Jordan Jones-Brewster.

We recorded and mixed the sounds for the narrator and I added a landing screen with animation:


I found a bug that caused players to only be counted wrong if they ran out of time (currently this isn’t told to the player, but will be in an upcoming build) instead of either running out of time or saying a wrong answer. After discussing we decided to keep this as a feature because of the nature of the “trivia quiz show” format made our questions/riddles rather specific and this gave players more wiggle room and a higher chance to progress in the game.

Gut Check Tech Test

I am the programmer for this game that I’m making with Jordan Jones-Brewster and Isabella Djurovic. 

Gut Check is an audio only trivia game in which the player must escape an AI’s trivia game by answering a series of questions correctly. Originally the team wanted to produce this for the Amazon Alexa, but learning the tech was too much for this short (5 week) project. Instead I am using Unity with Windows 10 speech recognition. I’ve used this before and gotten good results, and this can recognize reasonably specific phrases so it’s doable. (It can recognize Tom Hanks, for example). This influenced how we designing the game structure and what kind of questions/riddles we are writing for the game.

In this tech test I’ve implemented the basic structure of the game: the player is randomly given a topic (one out of 4) with four questions/riddles. If they get an swer wrong, they die and get a randomly pulled ending (one of 8). If they win they get a hint at what the AI is and its motivations (an audio clip array that is always accessed sequentially.) Answer all four correctly and the player wins.