Every few years a new face toy takes over the internet: guess my age, find my twin in a museum, which Disney character am I. The tools change; the impulse is old. People have been reading meaning into faces, and comparing them to other faces, for as long as we have records. Here's the lineage, from the reasonable parts to the embarrassing ones to today.
Reading faces is ancient
A treatise on physiognomy attributed to Aristotle's circle argued that character could be read from the body and face, comparing human features to animals: a lion-like face meant courage, and so on. Independently, East Asian traditions developed detailed face-reading systems; in Korea this is gwansang, with its own classic texts and vocabulary. In 18th-century Europe, Johann Kaspar Lavater's essays on physiognomy became a genuine bestseller, with readers checking silhouettes of friends against his plates. The common thread across these traditions: the face was a map of the person, and the job was to decode it.
The dark turn, and the lesson
The 19th century tried to make this scientific and produced phrenology, which read character from bumps on the skull, and later criminal anthropology, which claimed criminals could be identified by facial features. Both were discredited, and both did real harm by dressing prejudice in the language of measurement. The lesson stuck: measuring a face can tell you what it looks like, not what the person is. It's worth keeping that in mind whenever a modern app implies otherwise.
The fun turn
By the 20th century, comparing faces had become a pastime rather than a diagnosis. Lookalike contests were a newspaper and fairground staple, and celebrity impersonation became a profession. (The story that Charlie Chaplin entered a Chaplin lookalike contest and lost is repeated everywhere and has never been verified, which is itself a good example of how much we enjoy this idea.) The point had shifted from "what does your face reveal" to "who does your face resemble", which is a far healthier question.
The internet era
The internet turned resemblance into a shareable product. In the mid-2000s, genealogy site MyHeritage launched a celebrity look-alike feature that matched uploaded photos against famous faces using early face-recognition technology; it spread through blogs and early social networks. In 2015, Microsoft's how-old.net, a demo built on its face API, went viral within days as people posted their guessed ages. In early 2018, the Google Arts & Culture app added a feature that matched your selfie to a painting in a museum collection, and for a few weeks social feeds were full of people next to their Renaissance doubles. Around 2020, augmented-reality filters on Instagram and TikTok asked "which Disney character are you", picking one at random with a spinning animation. The Korean app scene ran parallel, with "AI gwansang" and celebrity-match apps cycling through app-store charts.
What's different now
Two things changed in the last few years. First, models good enough to compare faces became small enough to run inside a web browser, so a test no longer needs to send your photo to a server; see how our matching works. Second, comparison targets expanded from celebrities to anything with a face: anime casts, game characters, dog breeds, Joseon-era portraits. The measurement is real, which the random spinning filters weren't, and the framing is honest: it's resemblance, not revelation.
What hasn't changed
The urge to see yourself reflected in something else, and to show that reflection to friends. Lavater's readers passed his book around the parlour; we send a share card to the group chat. Same impulse, better tools, and this time no one is claiming your jawline predicts your destiny.