Assumption Conversion Prompt¶
Reflection prompt — instantiates Knowledge-Warrant Audit
A prompt that converts "we know" statements into explicit assumptions when the evidence chain is weak or missing.
Assumption Conversion Prompt is a single reflective question, applied to one claim at a time, that forces the speaker to trace the claim back to its warrant — and, when the trace comes up short, to rewrite the claim from a known into an assumption. Its whole reason to exist is the moment a sentence that begins "we know…" cannot answer the follow-up "…because what?" The prompt does not grade evidence on a scale, does not standardize how the claim is later communicated, and does not test a hypothesis; it does exactly one demotion — it takes a belief currently wearing the costume of established fact and, if no chain holds it up, strips the costume off and re-labels it as an assumption or an open unknown. It is the archetype's conversion step (step 7 of the audit loop) made into a portable habit anyone can run in the middle of a conversation.
Example¶
A software team is scoping a redesign of their signup flow. Someone writes on the whiteboard: "We know users abandon signup because the form is too long." The facilitator applies the prompt to that one sentence: What is the actual evidence chain behind "because the form is too long"? The room traces it. There is a real, observed fact — signup completion is 41% (illustrative). But the cause — "because the form is too long" — turns out to rest on one designer's recollection of a usability session two years ago and a competitor's blog post. The observation is warranted; the causal claim is not.
So the sentence gets rewritten in two pieces. The observed part stays a known: "Signup completion is 41%." The causal part is demoted and re-labeled: "Assumption: long form length is a primary driver of abandonment — untested." A third line is added for what nobody can see at all: "Unknown: where in the form users actually drop off." The redesign proposal now reads honestly, and — more usefully — the assumption line is now a cheap experiment waiting to be run, rather than a fact quietly steering a quarter of engineering work.
How it works¶
The prompt is a fixed three-move script applied per claim:
- Ask for the chain. Point at a confident claim and ask "what warrants this — observation, measurement, a source, or inference?" The demand is for the link, not for more confident restatement.
- Judge chain vs. no chain. If a genuine evidence link is produced, leave the claim as knowledge (its strength is someone else's job to rate). If the chain is missing, broken, or turns out to be "we've said it a lot," the claim fails the test.
- Convert and mark. Rewrite the failed claim in explicit assumption language, and split off any part that isn't even assumable into an explicit unknown. The rewrite is the deliverable — a demoted, honestly-labeled claim that action can no longer silently lean on.
The distinguishing discipline is minimalism: it changes the epistemic status of one claim and nothing else. It is a scalpel, not the whole audit.
Tuning parameters¶
- Trigger sensitivity — how confident the language has to be before the prompt fires. Apply it to every "obviously" and "clearly" and the meeting slows to a crawl; reserve it for load-bearing claims and you catch the ones that matter without pedantry.
- Chain bar — how much of an evidence link counts as "enough" to stay a known. A high bar demotes aggressively and risks paralysis; a low bar lets weakly-warranted claims keep their fact costume.
- Conversion verbosity — whether the rewrite is a terse tag ("assumption") or a full restatement with the missing evidence named. Fuller rewrites teach the room what would resolve the claim, but cost words.
- Unknown-splitting — whether to bother separating "assumable but untested" from "genuinely unknown." Splitting sharpens what to go find next; skipping it keeps the prompt fast.
When it helps, and when it misleads¶
Its strength is catching reification in the act — the slide by which a repeated assumption hardens into a fact nobody re-examines.[n1] Because it operates live and on a single sentence, it fits inside normal conversation and turns a vague unease ("is that actually true?") into a concrete, cheap correction: a relabeled claim and, often, an obvious next experiment.
It misleads when it is run as a rhetorical weapon rather than an honesty tool — demanding an unbroken chain for every claim is a recipe for indecision, and a hostile prompter can demote an opponent's well-founded claim by feigning that the chain isn't "enough." It also stops at demotion: it tells you a claim is only an assumption, not how strong the surviving evidence is or how much the decision can bear the uncertainty. The guarding discipline is to apply it to load-bearing claims, accept a proportionate chain rather than mathematical proof, and hand the demoted assumptions onward to be rated and tracked rather than treating the relabel as the end of the work.
How it implements the components¶
Assumption Conversion Prompt fills the narrow demotion slice of the archetype — the act of moving a claim across the knowledge/assumption line:
assumption_vs_knowledge_separator— this is the mechanism's core act: it decides, per claim, whether a "known" survives as knowledge or must be rewritten as an assumption.evidence_chain_trace— the prompt's first move traces the claim to its warrant; the trace coming up empty is precisely what triggers the demotion.uncertainty_and_unknown_marker— the part of a claim that can't even be assumed is split off and tagged as an explicit unknown rather than left buried inside a confident sentence.
It does not standardize how the demoted claim is later communicated to others (communication_labeling_standard — that is Epistemic Status Labeling), nor rank surviving warrants by type and strength (warrant_type_taxonomy, support_strength_rating — that is Evidence Ladder Labeling).
Related¶
- Instantiates: Knowledge-Warrant Audit — the prompt is the audit's conversion step made into a live, one-claim habit.
- Sibling mechanisms: Belief-Warrant Matrix · Epistemic Status Labeling · Evidence Ladder Labeling · Claim-Confidence Warrant Review · Source-Independence Cross-Check · Unsupported-Certainty Red Flag · Update-Trigger Checkpoint · Warrant-Decay Review
Editorial Notes¶
Form Classification¶
Form family: Interface, Display & Cue
Rationale: A prompt that converts 'we know' statements into explicit assumptions when the evidence chain is weak or missing, making its operative form a user-facing perceptual surface, prompt, label, or affordance that shapes attention or action.
Independent corroboration: The frozen evidence defines Assumption Conversion Prompt as 'A prompt that converts 'we know' statements into explicit assumptions when the evidence chain is weak or missing', so its operative form is Interface, Display & Cue.
Review outcome: Independent reviewer agreement; high confidence.
Origin Attribution¶
Primary origin: Philosophy
Origin pattern: Cross-disciplinary synthesis
Present-day reach: Universal
Rationale: Epistemology asks what warrants a knowledge claim and requires unsupported belief to be marked as assumption or unknown.
Related originating lineages:
- Education & Pedagogy — Socratic questioning turns warrant tracing into a portable reflective prompt.
- Organizational & Management Science — Facilitation applies the prompt to planning and product claims.
- Rhetoric — Claim-evidence-warrant analysis separates observation from causal assertion.
Review resolution: The knowledge-versus-assumption distinction is agreed to be philosophical. Reflective pedagogy, facilitated organizational inquiry, and rhetorical warrant analysis materially shape the live prompt. Statistical testing is a downstream way to investigate the demoted assumption, not part of the conversion prompt's origin.
Encyclopedia synthesis: The exact catalogued form synthesizes established practice rather than reproducing a single standard historical label.
Review outcome: Reconciled after independent review; high confidence.
Notes¶
[n1] Reification — treating an abstraction, inference, or assumption as if it were a concrete, established fact. The prompt is a direct countermeasure: it re-opens the question of whether a "known" ever earned that status. ↩