← Back Tastbaar

Writings on the wall

About

Writings on the Wall is an exploratory project that started the moment I realized every transcript from every meeting of the Flemish Parliament is public. I put them all into a database, meeting by meeting, along with all the information attached to each one, and started going through it: classifying the data, identifying the subject of each meeting, the emotion behind it, how people spoke to each other, how long they spoke, and who mostly talked about whom.

The core idea was to break that data apart and visualize how these conversations constantly talk over one another. Someone says something and immediately needs a response; everyone tries to talk over everyone else, chasing clickable soundbites while really just wanting the last word.

The project works on several levels. First, all the data was classified using a range of AI systems, programs, and emotion detectors, all run locally on my own machine. That database was then visualized in several ways, starting with a standard chart where each element sits side by side, using color to indicate emotion, and shape and length to indicate other properties.

That eventually evolved into an interactive installation. Alongside a sonification of what's being said, every statement and every speech by every politician generates a small sticker, hung inside a virtual restroom, like graffiti on a bathroom wall: "I was here," "yeah, but so was I," each one plastering over the last, following the logic of sticker culture. Each sticker has its own visual language: the sharpness of the star shape on it signals how angry someone was, a soft, wavy, non-star shape signals someone happy and optimistic, and a sharply pointed shape signals real anger. How much space is filled inside the sticker reflects how long that person spoke. The first sticker to start a conversation is always placed by the party that opened it, which makes it possible to read the general tendency of the room at a glance.

The project isn't only about showing people talk over each other, it's also about the sheer speed of these conversations. There is no real time to decipher everything or keep the same complete overview at every moment. What you get instead are broad strokes: how angry each party is, and who brings up the most. Fully dissecting a single conversation is nearly impossible, and that's the point, because while it's happening, other conversations behind you are going unheard.

Youtube video

Details

Year — 2025

Client — Self (LUCA)

Tags — Data visualisation, Interactive, projection

Tools — Processing, Touchdesigner, ollama, python