Created by Alessia La Rocca at SUPSI (Master of Arts in Interaction Design), Replica turns your algorithmic data profile into something you can see, touch, and reshape — letting you manipulate the versions of yourself that live on different platforms.
Every digital action, from a search query to a page visit, silently feeds an algorithmic profile that shapes real-world opportunities, from job screening to credit access, without awareness or consent. This profile, known as the Data Double, is built by cross-referencing fragments of behavior that appear harmless in isolation, and it operates as an invisible gatekeeper long before the user is ever aware of its existence.
Alessia La Rocca
Replica is an ecosystem that makes this data double visible and actionable. A Chrome extension passively observes browsing behaviour and classifies it into six categories, bio-demographic, geographic, professional, economic, socio-cultural and psycho-behavioural, translating raw navigation into a single “algorithmic sentence” that describes, in plain language, who the system believes the user to be. A companion dashboard reconstructs this inferred identity in more depth, showing the weight, age and propagation of each data point, and flagging moments of decontextualisation, when data collected in one context is exploited in another, such as a health search feeding an insurance risk score.
Replica_ Mod 01, the first physical device in the system, is designed for digital workers, individuals whose daily reliance on connected platforms makes them especially exposed to this invisible profiling. It sits on the desk as a quiet, ever-present object, its e-ink screen displaying only the sentence, with no interface to distract from it. Two rotary encoders return control through two opposing gestures: poisoning, which injects noise into a chosen category until the inference collapses and the user becomes unclassifiable, and amplifying, which reinforces a chosen signal to intentionally strengthen it. Both actions live exclusively on the object, never on the screen, so that reclaiming one’s algorithmic identity remains a deliberate, physical act rather than another distracted click.
The device uses Arduino Nano ESP32 as main microcontroller, running a local WiFi HTTP server and OTA updates. Waveshare 2.13″ e-ink display (GxEPD2 driver) for the sentence rendering. Two EC11 rotary encoders (one with push button) for category navigation and action selection. RGB LED for ambient feedback. Power chain with TP4056 charger, single-cell LiPo battery, MT3608 boost converter and USB-C breakout, all in a custom 3D-printed enclosure. Firmware is written in C++ on the Arduino/ESP32 core. Libraries include GxEPD2, Adafruit GFX (with a custom IBM Plex Mono bitmap font), ArduinoJson, WiFi/WebServer, ArduinoOTA. Chrome Manifest V3 extension that passively classifies browsing activity into six data categories via a custom keyword/signal classifier, generating the “algorithmic sentence.” Extension and device communicate over local HTTP. Extension and dashboard communicate through a content-script bridge, avoiding Manifest V3 service worker issues. Dashboard is static HTML/CSS/JavaScript webapp hosted on GitHub Pages, built on a custom design system with dedicated tokens and typefaces (Aeonik, Aeonik Mono).
Project Page | Alessia La Rocca | SUPSI
