Epistemics — Standards of Evidence and Labeling

Status: Stable convention (governs the whole wiki). Last updated: 2026-06-08

This page defines how strongly any statement in this wiki is claimed to be true. It is the most important meta-page: a research program toward a universal theory lives or dies on its honesty about what is known, what is inferred, and what is guessed. Every other page is required to use the conventions below.

The cardinal rule: Never let a speculative idea wear the clothes of an established fact. Overclaiming is the characteristic failure mode of "theory of everything" work, and it is treated here as a defect, not a flourish.

1. The five epistemic tags

Every nontrivial claim carries exactly one inline tag.

Tag	Meaning	Test for using it
`[ESTABLISHED]`	Experimentally confirmed to high significance, or rigorously proven inside an accepted framework; broad expert consensus.	"Would the overwhelming majority of working physicists stake their reputation on this?"
`[INFERENCE]`	Well-motivated, widely accepted reasoning, not yet directly confirmed. A strong bet, not a fact.	"Is this the standard conclusion of careful reasoning, while still logically defeasible by future data?"
`[SPECULATIVE]`	Plausible and principled, but lacking decisive evidence; a minority or research-frontier position.	"Is there a real argument for it, but no experiment that forces it?"
`[OPEN]`	Genuinely unresolved; the honest answer is we do not know.	"Is there active, undecided disagreement or simply no answer?"
`[CONTESTED]`	Competent experts actively disagree, with no consensus.	"Do serious camps hold incompatible views?"

Tags may be applied to a sentence, a clause, or a whole paragraph (stated once at the top of the paragraph). When a single sentence mixes a fact and an inference, split it.

Examples (calibration)

General relativity predicts gravitational waves, now directly detected. [ESTABLISHED]
Dark matter is a particle (rather than a modification of gravity). [INFERENCE] — the gravitational evidence for missing mass is [ESTABLISHED]; that it is a particle is the leading inference, not yet confirmed.
Spacetime geometry is emergent from entanglement. [SPECULATIVE]
The correct interpretation of quantum mechanics. [CONTESTED]
The mechanism that sets the neutrino mass scale. [OPEN]

2. Confidence (optional secondary signal)

Where useful, attach a coarse confidence to [INFERENCE]/[SPECULATIVE] items: high / medium / low / very-low. This is a subjective Bayesian credence, not a frequentist statistic. Use it sparingly and never to dress up a guess — a "high-confidence speculation" is still a speculation.

3. Assumption classification (used in `ASSUMPTIONS_LEDGER.md`)

Assumptions are graded on a separate axis — not how true but how load-bearing and how revisable:

Class	Meaning
`fundamental`	Appears irreducible; relaxing it breaks the theory's coherence or contradicts robust data.
`plausibly-fundamental`	Looks basic, but no proof it could not be derived from something deeper.
`historical-artifact`	Adopted for historical/pedagogical reasons; may be a contingent choice, not a necessity.
`mathematical-convenience`	A simplifying technical choice (smoothness, separability, a particular gauge) that buys tractability.
`untested-empirically`	Assumed because it has never been probed in the relevant regime, not because it is confirmed.

A central research activity of this wiki is moving assumptions down this list — turning supposed bedrock into a derivable or contingent feature.

4. Sourcing rules

No fabrication. Never invent a citation, an experimental number, a theorem, or a date. A missing reference is acceptable; a fake one is a disqualifying error.
Real sources only. Cite canonical textbooks and genuinely landmark results. If unsure of exact bibliographic detail, give author + title + approximate year and mark [unverified].
Numbers carry context. Quote measured quantities with their nature (e.g., that the "10^120" cosmological-constant figure is regularization-dependent, see CONSTANTS_AND_SCALES.md). A number without its caveat is a half-truth.
Folklore is flagged. Widely repeated claims that are actually subtle or wrong (e.g., loose statements of Haag's theorem, "energy is not conserved in GR") must be stated carefully, not parroted.

5. Distinguishing four things that get conflated

A recurring discipline here, especially in GAPS_AND_CONTRADICTIONS.md:

Logical inconsistency — two accepted statements cannot both be true. (Rare. Most "contradictions" are not this.)
Domain-of-validity mismatch — two frameworks disagree only where neither is trusted (e.g., GR vs QFT at the Planck scale). Not a paradox; a frontier.
Unsolved-but-consistent problem — no known inconsistency, just no derivation/solution yet (e.g., the value of a parameter).
Conceptual tension — the frameworks are formally compatible but rest on jarringly different worldviews (e.g., block-universe GR vs the apparent dynamical collapse in QM).

Mislabeling a frontier as a paradox is itself an epistemic error.

6. How claims change status

Status is not permanent. A claim moves between tags as evidence and argument accumulate. Every promotion/demotion is logged in CHANGELOG.md with a one-line reason. The wiki is cumulative: we revise pages in place and record why in the changelog, rather than starting over.