Overview
Interpreter Training Program is a short experience in which you play as an interpreter trainee, completing deterministic tasks involving a set of unfamiliar symbols. The game takes place through a diegetic interface, the player’s workspace, consisting of a CRT console, a custom keyboard, a training manual, and a display panel. The player is introduced to the environment and completes tasks that treat arbitrary sequences of symbols as words and grammatical components but hold no real semantic grounding. The space is mechanical, utilitarian, and ultimately meaningless. By the end, the player completes the training and begins receiving tasks that require interpretation and meaning, only to realize the absurdity and cyclical nature of the system. The machine then takes over control to answer and generate meaning for you, leading to the collapse of the system.
This piece draws on several thought experiments and fiction, including the Chinese Room, the Infinite Monkey Theorem, Newspeak in 1984, and Papers, Please. The Chinese Room is a thought experiment in which a person who does not speak Chinese is put into a room with a manual. By following the instructions in the manual, the person can construct a believable response in Chinese despite not understanding anything. A Chinese speaker outside the room can have a valid conversation with the person inside and will not notice that they cannot speak Chinese. This raises the question of understanding without comprehension, which is exactly what I wanted to create through the symbols: words, grammar, parts of speech, and devices such as particles. The system presents sequences with real, convincing grammatical and linguistic rules and trains the player through tasks that resemble a gamified language-learning app. The player learns to understand the system, yet there is no comprehension. In the ending sequence, the machine begins to answer the questions for the player. Not the deterministic, lookup-table-like tasks, but questions about the player, personal, and intimate even. This creates the irony that the player is completing deterministic tasks a machine can easily accomplish, while the machine becomes the one that generates meaning and expresses “itself”.
This is what we see today with the development of generative AI tools. In my Artwork #3: Cogito, ergo sum, I embodied an LLM by speaking only what the AI tells me to. LLMs do not understand the meaning of the text they receive or generate. They “speak” and “think” by guessing the most probable next token based on the training data. They do this so well that the text becomes convincing, just as the responses generated by the operator inside the Chinese room. In this game, the player essentially becomes the machine, performing operations by following deterministic instructions without understanding any of the words they generate. I observed that when players are abruptly asked to “interpret” – for example, by describing the characteristics of a chair – they inadvertently return to the menu and look for adjectives or sequences of symbols that resemble a chair. When asked to enter their favorite color, they often recall their response to “the color of the sky” that was supposedly “correct”. The truth is, these questions were never evaluated by the program. Pass/fail is only determined by a pseudo-random number generator, which, ironically, is also deterministic by nature.
Throughout the experience, the player does not make meaningful decisions. They are guided throughout the training, and the system does not care who they are or what they do. Unlike Papers, Please, where you have only a limited time each day and have to earn money for your family, the system never rushes you (except in the Memorize task). You can always quit and resume later, and you can be away for a long time, and the system will still be there, waiting for your action. The environment is neither hostile nor welcoming – only indifferent. Obedience isn’t forced, but voluntary. To amplify this, I designed a “maintenance break” around halfway through the experience: a full minute of nothing but waiting. I found it interesting that most players would just sit there and stare at the progress bar, like an idling machine waiting for the next instruction. Perhaps it is because the maintenance reflects a sense of meaning, showing that their actions in this arbitrary system have an impact.
Iterations & Documentation
The first iteration focused on solving logical puzzles that are closer to the Chinese Room. The tasks involve either processing the sequence to generate output or categorizing it according to rules. While this aligns more with mechanical and procedural gameplay, it distracts the player from the central message: those who are not used to solving logical puzzles find it confusing and intimidating, while others become too focused on the challenge and enjoyment to understand the experience. The interface design was also confusing to use, though I personally liked the 3D low-res style.
To address the issues, I decided to go with 2D, which allows better organization of the components and improved readability. I simplified the tasks, which is when I realized I could draw inspiration from language-learning apps, which helped shape the final set of tasks: Translate, Parts of Speech, Fill in the Blank, and Memorize. Through the playtest, I noticed that the new interface is very intuitive for the players to pick up without any explanation. The tasks are much less “fun”; however, they align better with the theme and do not distract the players. I would say the interface design has been one of the most challenging parts of this project, especially with the relatively minimalist style. For example, one piece of feedback I received was that players often do not notice when their task has changed, and sometimes they only read part of the console messages, missing crucial details. I ensured that each task had a unique appearance so they were easily recognizable, but the most effective modification was adding a one-second transition when the screen changed.