Homophily
Similarity breeds contact — and ideas spread inside closed circles before they cross groups
Homophily (Everett Rogers) is the tendency for communication to occur more often among people who are similar — in status, education, discipline, risk posture, or identity. Rogers treats it as a structural driver of the diffusion of innovation curve: new ideas circulate quickly inside homogeneous clusters and cross group boundaries slowly, producing the familiar S-shape over time.
For this corpus, homophily is not only a marketing or adoption fact. It is an analytical lens for epistemic failure: when truth is unwelcome, homophily explains why evidence stays inside the wrong cluster, why a correct minority cannot get a hearing, and why the same claim reads as obvious to one tribe and absurd to another without either side examining tier honestly.
Rogers’s use
In Diffusion of Innovations, homophily accelerates spread within a group and impedes spread between groups. Innovations often appear to “stall” not because the idea failed on merits, but because channels between dissimilar actors are thin. Heterophily — contact across difference — is what can bridge that gap; heterophobia is what blocks it.
Epistemic misuse
Homophily becomes destructive when similarity substitutes for evidence:
- A claim is dismissed because the messenger is junior, from another function, or “not one of us” — not because the tier was examined (The buried finding).
- A warning is ignored because the bearer does not look like past successful prophets — Cassandra syndrome.
- Organisations only audit ideas that arrive through familiar channels, so correct out-group insight never reaches a commitment boundary.
These are Innovation Divide-shaped failures plus social topology: appetite masquerades as judgment, and homophily decides who is allowed to speak.
Corpus stance
A2 — working context: Rogers attribute; use homophily to diagnose why truth fails to move, not as excuse to abandon tier discipline. Pair with explicit epistemic uncertainty and authority weight so similarity bias is visible in the record.