Count only independent evidence lines for confidence — ten sources sharing one origin is one line, not ten
When evaluating confidence in a belief, count only genuinely independent lines of evidence—sources that do not share origins, methods, or assumptions—rather than total source count, because correlated sources compound confidence on a single foundation.
Why This Is a Rule
Ten articles that all cite the same study are one line of evidence, not ten. The count of sources is misleading because correlated sources — those sharing origins, methods, or assumptions — don't provide independent confirmation. They echo the same foundation. If that foundation is flawed, all ten "confirming" sources fail simultaneously.
Genuine independent evidence comes from sources with different origins (separate research teams), different methods (experimental vs. observational vs. theoretical), and different assumptions (different frameworks reaching the same conclusion). When independent lines converge, confidence is genuinely supported. When correlated lines converge, confidence is built on a single foundation that feels multiplied but isn't.
This is the epistemic equivalent of diversification in investing: ten investments in the same sector feel diversified but correlate in a downturn. Ten investments across sectors provide genuine risk reduction. Similarly, ten evidence sources from the same methodological tradition provide less confidence than three sources from different traditions.
When This Fires
- When multiple sources seem to confirm a belief and you're assessing confidence
- When "studies show" is used to justify high confidence
- During any evidence evaluation where source count is used as a proxy for evidence strength
- When forming beliefs about contested or uncertain topics
Common Failure Mode
Counting echoes as independent evidence: "Five articles agree on this, so it's well-supported." Check: do all five cite the same original study? Do all five use the same methodology? Do all five share the same theoretical framework? If yes to any, the five articles are correlated evidence from a single line, and your confidence should match one well-supported line, not five independent ones.
The Protocol
When evaluating confidence: (1) List all sources supporting the belief. (2) For each pair of sources, check independence: do they share origins (same data, same study)? Methods (same methodology)? Assumptions (same theoretical framework)? (3) Group correlated sources into evidence lines — each line counts as one unit of evidence regardless of how many sources it contains. (4) Count lines, not sources. Confidence should scale with independent lines: 1 line = tentative, 2 independent lines = moderate, 3+ independent lines = strong. A belief supported by 10 correlated sources has the same evidential support as one supported by the single study they all cite.