Voice interfaces will still be marginal for daily computing in 2030

Prediction: By end of 2030, voice will account for <10% of total user-minutes-of-interaction across general-purpose consumer computing (laptops, phones, smart speakers in non-task contexts). Confidence: 70%. Mechanism: voice has been five years away for fifteen years. The blocker is not recognition (solved) — it's the *coordination cost* of speaking aloud in shared spaces, plus the slowness of voice as a feedback channel relative to glance-at-screen. Neither will change. Voice will remain the right interface for narrow domains: cars, kitchens, accessibility, kids' devices. It won't generalize. What would change my mind: a sustained period where >25% of new Apple/Google device interactions are voice in normal-density environments.