The case for a smarter emoji keyboard
Scrolling through 3,000-plus characters to find a face is a design failure. Here's what the next generation of keyboards is trying.

The scroll problem
Open the emoji keyboard, scroll to category three, hold for skin tone, swipe twice. The default emoji input has barely changed since 2015, and the catalog has grown more than threefold in that time.
What's being tried
Three approaches are emerging. The first uses on-device language models to predict the next emoji from the surrounding text. The second turns the picker into a search-first surface, with categories demoted. The third reorganizes the grid around recent personal use rather than universal frequency.
Which one wins?
Probably all three, in some combination. The interesting question isn't which approach is correct โ it's whether platforms can agree on enough common ground that muscle memory survives the switch.


