Here are two equivalent diagrams: a linear chain and a feedback-system view.

A) Linear chain (minimal)

(1) Unauthorized Epistemic Authority
→ (2) Interface as Governance (authority becomes behavior-shaping by default)
→ (3) Compression-Induced Harm (certainty/urgency/caveats erased)
→ (4) Bias as Selective Attenuation (what gets erased follows social gradients)
→ (5) Adoption Without Institutional Comprehension (institutions rely on outputs they can’t audit)
→ (6) Governance as Performance, Not Enforcement (statements/laws lag; mechanisms absent)
→ (7) Scale, Speed, Irreversibility (harm propagates faster than correction)

B) System view (with reinforcing loops)

           ┌───────────────────────────────┐
           │ (7) Scale, Speed, Irreversibility │
           └──────────────┬────────────────┘
                          │ amplifies
(1) Unauthorized Authority │
          ┌───────────────▼───────────────┐
          │ (2) Interface as Governance     │
          └───────────────┬───────────────┘
                          │ produces
          ┌───────────────▼───────────────┐
          │ (3) Compression-Induced Harm    │
          └───────────────┬───────────────┘
                          │ allocates loss along
          ┌───────────────▼───────────────┐
          │ (4) Bias (uneven attenuation)   │
          └───────────────┬───────────────┘
                          │ enables institutional reliance
          ┌───────────────▼───────────────┐
          │ (5) Adoption w/o comprehension  │
          └───────────────┬───────────────┘
                          │ triggers symbolic response
          ┌───────────────▼───────────────┐
          │ (6) Governance as performance   │
          └───────────────┬───────────────┘
                          │ fails to bind, so harms persist
                          └───────────────► back into (7)

Key insight of the system view: (7) doesn’t just come “at the end”; it feeds back to intensify (1)–(6) by making remediation slower and less legible.