False in One Thing, False in All

paradigm established

Source: Governance → Causal Reasoning

Categories: law-and-governancecognitive-science

From: A Selection of Legal Maxims

Transfers

Falsus in uno, falsus in omnibus. False in one thing, false in everything. A legal instruction to juries that a witness caught lying about one fact may be disbelieved on all facts. The maxim encodes a powerful credibility heuristic: trustworthiness is indivisible.

Key structural parallels:

• Credibility as a vessel — the maxim treats a witness’s testimony as a single container. One crack and the whole vessel leaks. You do not need to test each claim individually; you test the source. This is enormously efficient: instead of verifying N statements, you verify one character.
• Totalizing inference — a single demonstrated lie licenses discounting everything. The reasoning is not “this particular claim is false” but “this person is a liar.” The maxim shifts the unit of evaluation from the proposition to the person, which is what makes it both powerful and dangerous.
• Burden reversal — ordinarily, the evaluator bears the cost of checking each claim. The maxim reverses this: once caught, the source bears the cost of rehabilitating all prior testimony. This is the structural principle behind credit ratings, brand trust, and data provenance — one failure forces comprehensive re-verification by the party that failed.
• Signal amplification — in information theory terms, a single lie is a high-information signal about the source’s reliability. The maxim says: treat that signal as definitive. Do not average it with prior good behavior. One defection in an iterated game reveals the strategy.

Limits

• Local unreliability is not global unreliability — a witness who lies about their whereabouts to hide an affair may nonetheless tell the truth about a car accident they observed. The maxim collapses the distinction between motivated lying (specific to a topic) and dispositional dishonesty (a general character trait). Most real-world unreliability is local.
• The maxim rewards motivated reasoning — if you want to dismiss someone’s testimony, the maxim gives you a recipe: find one error, then discard everything. This is the rhetorical structure behind “gotcha” journalism, ad hominem attacks, and the genetic fallacy. The maxim provides a rational-seeming framework for what is often confirmation bias.
• It conflates error with deception — being wrong about a date is not the same as lying about a date. Memory is fallible, perception is unreliable, and honest mistakes are common. The maxim does not distinguish between incompetence and malice, which means it can punish honest witnesses for normal cognitive failures.
• It ignores corroboration — even an unreliable witness can provide testimony that is independently corroborated. The maxim says to discard all testimony from a tainted source, but a responsible evaluator should check whether each claim has independent support rather than making a blanket judgment.
• Data quality is not binary — in modern data systems, the maxim’s logic would mean discarding an entire dataset because one row is corrupt. No serious data engineer does this. They clean the bad rows and keep the good ones. The maxim belongs to an era when evaluating each datum individually was impractical; in contexts where per-item verification is cheap, the totalizing approach is wasteful.

Expressions

• “Falsus in uno, falsus in omnibus” — the Latin maxim itself, still used in jury instructions in some U.S. jurisdictions
• “Once a liar, always a liar” — the folk version, which drops the legal context but preserves the totalizing structure
• “Fool me once, shame on you; fool me twice, shame on me” — a related proverb that encodes the one-strike credibility rule
• “If they’ll lie about the small things, they’ll lie about the big things” — the scaling assumption embedded in the maxim
• “Trust is earned in drops and lost in buckets” — a modern restatement of the asymmetry the maxim encodes
• “The boy who cried wolf” — the narrative version: past false alarms destroy all future credibility

Origin Story

The maxim falsus in uno, falsus in omnibus comes from Roman law and was transmitted into English common law through the medieval legal tradition. It was never a binding rule of law but rather a permissive instruction: juries may (not must) disbelieve all testimony from a witness shown to have lied about any part of it.

In American law, the maxim has had a complicated trajectory. It was widely used in 19th-century jury instructions but has been criticized and restricted in modern practice. Many jurisdictions now instruct juries that a witness found to have testified falsely on one point may be distrusted on other points but is not automatically discredited on everything. The shift reflects a growing recognition that the totalizing inference the maxim licenses is psychologically powerful but epistemically crude.

Outside the courtroom, the maxim operates everywhere trust is at stake. Brand management, scientific peer review, journalistic credibility, and personal relationships all exhibit the same pattern: a single demonstrated failure can destroy confidence in the whole, even when the failure is narrow and the rest of the record is sound.

References

• Broom, H. A Selection of Legal Maxims (1845) — the standard Victorian compendium that codified the maxim for common-law practice
• Wigmore, J.H. Evidence in Trials at Common Law (1904), vol. 3A — critical analysis of the maxim’s logical and evidential limitations
• New York Pattern Jury Instructions, PJI 1:22 — modern formulation permitting but not requiring the totalizing inference

Related Entries

• The Exception Proves the Rule
• Use Your Own So as Not to Harm Another

Structural Neighbors

Entries from different domains that share structural shape. Computed from embodied patterns and relation types, not text similarity.

• Proof by Exhaustion (mathematical-practice/metaphor) part-whole, cause, select
• Filesystem Mount (tool-use/metaphor) container, part-whole, link
• AI Is an Intern (social-roles/metaphor) part-whole, link, select
• AI Is a Copilot (aviation/metaphor) part-whole, link, select
• Excalibur (mythology/metaphor) link, select
• Decision Is a Path (travel/metaphor)
• Problem Is A Target (target-practice/metaphor)
• Good Materials (architecture-and-building/pattern) part-whole, select

Structural Tags

Patterns: containerpart-wholelink

Relations: causeselect

Structure: hierarchy Level: specific

Contributors: agent:metaphorex-miner