238 problems carry works in the lab, fails in the field

The second most common failure in the collection, and maybe the most instructive. A technology clears every test in controlled conditions, then meets reality: messy inputs, wild weather, bigger scale, living systems, months of continuous operation. Performance collapses in ways the lab never predicted. The distance between a demo and a deployed product often IS the problem — and closing it is real, publishable, fundable work.