Store evidence as independent nodes with methodology — not as decorative citations
Store evidence with full methodological metadata (sample size, control conditions, limitations) as independent nodes rather than as decorative citations on claims, to enable proportionality assessment and multi-argument reuse.
Why This Is a Rule
Most knowledge systems treat evidence as decoration on claims: "Working memory holds 4±1 items (Cowan, 2001)." The citation adds credibility to the claim but contains no usable information about evidence quality. Was the sample 12 undergraduates or 12,000 adults? Was it replicated? What were the limitations? Without this metadata, you can't assess whether the evidence actually supports the claim at the weight you're giving it.
Storing evidence as independent nodes — separate from the claims they support — solves two problems. First, each evidence node carries its own methodological metadata (sample size, control conditions, population, limitations), enabling proportionality assessment: is this claim supported by strong or weak evidence? Second, a single piece of evidence can support multiple claims across different arguments. "Cowan 2001" might support claims about working memory limits, cognitive load, and chunking — but only if it's an independent, reusable node rather than a parenthetical footnote attached to one claim.
When This Fires
- Adding a research citation to a note or document
- Building a personal knowledge base where claims need evidential backing
- Evaluating the strength of your own beliefs or arguments
- Any time you write "(Source, Year)" after a claim
Common Failure Mode
Treating the citation as sufficient evidence. "(Kahneman, 2011)" after a claim signals "someone smart said this" but provides no information about how the claim was established. Was it a controlled experiment, a survey, an anecdote, or a theoretical argument? Each warrants different confidence levels, but the decorative citation treats them identically.
The Protocol
When capturing evidence: (1) Create a separate evidence node (not an inline citation). (2) Include: source, year, study type, sample size and population, control conditions, key findings, stated limitations, and your assessment of strength (strong/moderate/weak). (3) Link the evidence node to every claim it supports. (4) When reviewing claims, follow the link to the evidence node and assess proportionality: is the claim's confidence level justified by the evidence's strength? Claims with only weak evidence should be held more tentatively than claims with strong evidence — and you can only make this distinction if the methodology is recorded.