Evidence vs observation: the distinction that makes intelligence trustworthy

Most intelligence workflows treat evidence and observation as the same thing. They aren't. Conflating them is where the quality of analysis degrades — often invisibly.

What evidence is

Evidence is a raw source. A Reuters article reporting that TSMC raised its capex guidance. A 10-Q filing showing days sales outstanding extended from 58 to 71 days. An earnings call transcript where the CEO shifts from "broad-based demand" to "concentrated but durable demand."

Evidence is factual, sourced, and uninterpreted. It tells you what happened. It does not tell you what it means.

What an observation is

An observation is the interpretation of evidence — a structured claim about what the evidence reveals about an underlying system. "TSMC raised capex guidance" is evidence. "Infrastructure investment in AI buildouts remains strong despite near-term revenue softness in individual semiconductor names" is an observation derived from that evidence, plus several others.

An observation is a claim. It can be right or wrong. It requires judgment to produce and judgment to validate. This is why it has more value than evidence — and why it carries more responsibility.

Why the distinction matters

When evidence and observation are conflated, two failure modes appear:

The first is false precision. Treating an interpretation as if it were a fact. "The market is consolidating" stated as established truth, when it is actually a judgment call about a pattern in the evidence. False precision makes it harder to update when new evidence contradicts the interpretation.

The second is untraceable conclusions. If you can't distinguish between "this is what the evidence shows" and "this is my interpretation of what the evidence means," you can't audit your reasoning when you turn out to be wrong. And you will sometimes be wrong — the question is whether you can learn from it.

What good intelligence workflow looks like

Good intelligence workflow maintains a clear separation at every step. Evidence is collected and cited. Observations are derived from evidence and explicitly labeled as interpretations. Observations are validated — by a human who understands the system — before they enter the intelligence base. And every observation remains traceable back to the specific evidence that supported it.

This separation sounds bureaucratic. In practice, it's what makes the difference between an intelligence base you can trust and one that merely looks authoritative.