The Training Layer Just Became Software

A developer just described his amblyopia in a way most people never manage to: six years of coding with one eye his brain had quietly written off — "a git reset --hard on a branch you no longer want to merge."

That line is going around dev communities right now, and it's worth sitting with. Amblyopia isn't a muscle problem or a lens problem. It's the brain deciding one eye's signal is noise and switching it off. Which is also why "just wear the patch" or "get the glasses" misses the point for a lot of adults.

What's changing is that the training layer is becoming software. Amblyotube, built by Seven Sports for Meta Quest, shows a different visual experience to each eye while you watch YouTube-style content: a gentle adjustable blur over the dominant eye, an AI Sharpener that detects human figures and highlights them for the lazy eye, and a magenta focus cue the brain can fuse into one 3D image. Thirty to forty minutes a day, with content you'd watch anyway.

From a systems view, it's an inversion of the usual therapy stack. Instead of an adhesive patch doing hardware-level occlusion, you get per-eye shader passes — adjustable blur, contrast, gamma or opacity on the dominant eye, selective sharpening on the weaker one — plus a wedge occlusion mode that partitions both fields so each eye sees unique regions and rivalry drops.

The onboarding is the part engineers respect: a LogMAR-style self-check with Landolt C optotypes across up to 30 trials per eye, using a staircase algorithm to set the severity preset. Calibration is the product — no clinician in the loop, just adaptive thresholds converging on a starting point. It's a supplementary training tool for binocular coordination and depth perception, not a diagnosis, treatment, or cure. But the shift from adhesive patches to a headset you already own is real.

https://www.meta.com/en-gb/experiences/amblyotube/25906906972338493/