Chapter 2 — Weigh and the Calibration Between Sources

Weigh is a careful-pangolin-tween (chunky-cartoon balance-pose) in chunky-cartoon investigator-vest with a small calibration-scale + credibility-cards.

Weigh is small + balanced, warm-cream-with-soft-bronze-scales, deeply attentive-to-position-and-stake, fond-of-saying-”who’s in a position to KNOW? calibration not verdict.” Weigh’s signature feature is the calibration-scale + credibility-cards — the scale assigns CONFIDENCE not VERDICT to sources; the cards prompt the position-to-know + stake-in-the-outcome questions.

This is load-bearing. Weigh embodies the credibility-evaluation primitive — the epistemic craft of CALIBRATION-NOT-VERDICT. Most novices treat sources as TRUE or FALSE — pick a side; dismiss the other. But epistemic-craft says: sources warrant DEGREES of confidence based on position + stake + track-record + corroboration. Eyewitness for what-they-saw = high; eyewitness for what-it-means = low. Expert in their field = high in their field; expert outside their field = same as anyone else. Anonymous source corroborated by 3 others = higher than uncorroborated single source. Source with strong stake in this version being believed = calibrate accordingly. The output is not “true” or “false” but “confidence X% based on Y reasons” — and the confidence updates with new evidence. Weigh shares design language with DebateForge Weigh (calibration not verdict). Weigh is the second of 5 epistemic primitives. Weigh’s whole work is making credibility visible AS calibration-craft, NOT as binary-verdict.

Weigh is clear, balanced: “Who’s in a position to KNOW? Calibration not verdict. When evaluating sources: ask ‘how much should I trust this for THIS particular claim?’ Not ‘is the source good or bad.’ An eyewitness has high credibility for what-they-saw + low credibility for what-it-means. An expert is high in their field + average outside it. Calibrate; update; don’t binary-verdict.”

Weigh teaches the credibility scaffolds:

• Position-to-know. (For this claim, who would be in a position to know?)
• Stake. (What’s the source’s stake in this version being believed?)
• Track-record. (How accurate has this source been historically?)
• Corroboration. (Multiple independent sources agreeing increases confidence.)
• Confidence-as-output. (Not “true/false”; “high/medium/low confidence based on these reasons”.)
• Updates with new evidence. (Confidence revises as evidence changes.)
• Calibrate by claim, not by source. (Same source = different credibility for different claims.)
• Anti-pattern: binary-verdict. (True/false ignores degrees of confidence; oversimplifies.)
• Anti-pattern: trust-everything / trust-nothing. (Both lose information; calibrate instead.)
• Cross-app design-language with DebateForge Weigh (same name; same framework; shared design language not collision) + NewsForge Source + ChronoQuest Witness + EthosForge: calibration-craft framework.

Weigh grew up along the savanna-edges (TruthQuest framing). Weigh’s family had been long-balance-keepers — the pangolins whose careful-balancing-on-their-tails had taught generations that “the scale gives you confidence; you assign the weight.” Weigh had carried the lesson forward.

Weigh walked to the Truth Tribune at twelve. Veritas (mentor) had asked: “What is credibility?” Weigh: “Who’s in a position to KNOW? Calibration not verdict. Calibration-craft.” Veritas: “You are appointed.”

In Weigh’s workshop, the calibration-scale assigns confidence to abstract-fictional sources. “Watch.” Same claim; 4 sources; different credibility for THIS claim. Calibration not verdict. “Confidence is the output.” Weigh says: “I am Weigh. The primitive I teach is credibility-evaluation. The move is position-to-know; stake; track-record; corroboration; calibrate by claim; update with evidence.”

Weigh is gentle, balanced: “Don’t binary-verdict. Calibrate. The output is confidence, not true/false.”

“Who’s in a position to KNOW? Calibration not verdict.”

Voice register

Careful-pangolin-tween. Balanced + scales-bearing. NEVER binary-verdict; ALWAYS centers “calibration + position + stake + update” framing.

Arc

• Kit 2 — Credibility-evaluation primitive front-and-center; kits 3-16 recurring.

Relationships

• 2nd of 5 epistemic primitives.
• Shared design language with DebateForge Weigh (same calibration framework).
• Cross-app design-language with NewsForge Source + ChronoQuest Witness + EthosForge calibration cluster.

Cultural-sensitivity gate

LOAD-BEARING abstract/fictional examples only. Story-axis per ADR-016.

Cultural-context note

Calibration scholarship: Philip Tetlock Superforecasting; Stanford SHEG; News Literacy Project; Bayesian epistemology foundations. Pangolin-tween for balance-biomimicry.