Algorithm Mirror

Algorithm Mirror is an interactive web-based prototype that simulates the inner workings of social media recommendation algorithms. Users can swipe through short-form video content, with their interactions — including likes, skips, dwell time, and interest selections — being logged and analyzed in real time. Based on this behavior, the system generates a personalized summary at the end of the session, visualizing the algorithm’s “understanding” of the user’s preferences. The project explores how seemingly neutral data tracking can influence self-perception, highlighting the psychological and ethical dimensions of algorithmic personalization. This simulation was developed using HTML, CSS, and vanilla JavaScript, with a backend-ready design for behavior logging. The UI reflects a minimalist Apple-style aesthetic, aiming to be both elegant and immersive.