Internet Culture

Dataville

Clock in as a data labeler and watch small choices rewrite a one-room dystopia.

Dataville website homepage screenshot
Site Stumble field capture of Dataville, reviewed October 12, 2026.

Dataville is a short branching visual novel about the people hidden inside the AI supply chain. You label data for a facial-recognition company, balance rent against conscience, and watch the world outside your apartment answer back.

What is Dataville?

Your new gig arrives dressed as flexible work with a respectable mission. Then the labels get thornier. Dataville puts each decision on the same scale as groceries, rent, workplace ambition, solidarity, and the fate of extraterrestrial refugees caught in a surveillance system. The story stays inside one apartment, but it does not feel still: the computer delivers the work, the television reports the consequences, and the window quietly changes as your choices accumulate. That tight frame turns an abstract labor pipeline into something domestic, immediate, and difficult to shrug off.

What you can do there

  • Label fictional training data
  • Make branching workplace choices
  • Watch the apartment reflect accumulated decisions
  • Pursue different endings

Why we picked it

The game makes the invisible human layer of machine learning playable without turning it into a lecture. Its best trick is spatial: one room becomes both workplace and moral weather station, registering decisions that might otherwise feel like harmless clicks.

How to get the most from it

Play the browser build with a keyboard and mouse, read the job language closely, and resist choosing only for the cleanest outcome. A second route should reveal how rent pressure, company status, sabotage, and collective action pull the story in different directions.

Good to know

This is still labeled an alpha and in development, and the page explicitly says mobile play is unsupported. Large desktop downloads are available if the browser build struggles. The unsigned macOS package may trigger a system warning, and live accessibility, save, audio, and performance behavior were not tested.

Who made it, and when?

Brent Bailey, Arnab Chakravarty, Henry Baum, Lifei Wang, and Ian McNeely is the credited creator or organization. The earliest supported launch date we found is November 20, 2024. A documented update was recorded November 22, 2024.

Creator’s official page
Reviewed by Site Stumble editorial

Last editorial review: October 12, 2026. Our notes combine direct observation, first-party information when available, and independent research.

Read our methodology
Open Dataville