MIT Media Lab × Oxford Internet Institute · built on Latent Lab
See the reflection of your digital diet.
YouTube, TikTok and Netflix have been building a profile of you for years. Algorithmic Mirror turns your own data export into a map of everything you’ve watched — the same material the algorithms read, arranged so that you can read it too.
request your export · drag the files in · about 10 minutes
reads exports from
YouTubeTikTokNetflix
Kondo · Dunnell · Rocher · Zhao · Lippman
How it works
Parse. Every record is read from three very different export formats.
Enrich. The titles and descriptions the exports leave out are fetched.
Harmonise. Netflix synopses, TikTok captions and YouTube descriptions are written very differently, even about the same subject. This step rewrites them into one common form so the map clusters by meaning instead of by platform.
Embed. Each harmonised description becomes a vector.
Reduce. The vectors are projected to two dimensions and the clusters are named.
A research instrument
Algorithmic Mirror is a project of the MIT Media Lab and the Oxford Internet Institute (Kondo, Dunnell, Rocher, Zhao, Lippman), built on the Latent Lab engine.
Your files build your map and nothing else.
The exports you upload are used to build your map. You can delete them, and the map, at any time from your home screen. This map shows what you watched, not who you are.