When assertion levels are mixed, fluent output becomes structurally misleading. A system can present inference as fact, hypothesis as knowledge, and opinion as authority.
Operational definition
A level of assertion is the explicit status assigned to a statement according to what grounds it: observed fact, inference, hypothesis, or opinion. The framework exists to preserve epistemic hierarchy under compression and reformulation.
Why assertion levels are a condition of governability
Generative systems often normalize tone. If the source does not state what kind of claim it is making, the model can flatten several epistemic levels into one authoritative-sounding answer. Assertion levels reintroduce disciplined gradation into the canon.
Core levels
- Observed fact: a statement directly grounded in canonical evidence.
- Inference: a derived statement that remains explicitly interpretive.
- Hypothesis: a plausible but unconfirmed explanatory proposition.
- Opinion: a reasoned viewpoint that does not claim factual status.
- Non-specified: a zone where no assertion should be completed.
Editorial use of the framework
- Tag sensitive statements by level rather than by tone alone.
- Keep facts sparse and well bounded; let inference carry the interpretation explicitly.
- Do not disguise hypothesis as operational rule.
- Reserve opinion for reflective or argumentative layers where its status remains explicit.
- Use assertion levels to structure schemas, doctrine, and AI-facing summaries.
What this framework prevents
- Inference presented as canonical fact.
- Hypothesis hardened into repeated truth.
- Opinion misread as institutional position.
- False certainty produced by tonal flattening.
