E01¶


Research	R0002 — Research Standards for AI-Assisted Writing
Run	2026-03-13
Claim	C011
Source	SRC05
Evidence	SRC05-E01
Type	Factual (comparator)

IC Has Calibrated Probability Scale

URL: Not captured — experimental run

Extract¶

ICD 203 defines calibrated probability language for intelligence assessments: Almost certain (95-99%), Very likely (80-95%), Likely (55-80%), Roughly even chance (45-55%), Unlikely (20-45%), Very unlikely (5-20%), Remote (1-5%). This is a formally defined, calibrated uncertainty scale with explicit probability ranges. No comparable scale exists in journalism or fact-checking.

Relevance to Hypotheses¶

Hypothesis	Relationship	Strength
H1	Supports	Strong — confirms journalism lacks calibrated uncertainty scales
H2	Supports	Strong — confirms the specific absence of calibrated scales
H3	Contradicts	Strong — comparator shows what journalism would need to refute the claim

Context¶

This evidence serves as a comparator rather than a direct finding about journalism. The IC's probability scale is a concrete example of what a "calibrated uncertainty scale" looks like. Journalism has nothing analogous. PolitiFact's Truth-O-Meter (True, Mostly True, Half True, Mostly False, False, Pants on Fire) is a rating scale but not calibrated to probability ranges.

Notes¶

The contrast is stark: ICD 203's scale maps words to probability ranges (e.g., "Likely" = 55-80%). Journalism rating scales map to qualitative categories without probability calibration.