The five evidence states
Every verified claim on Soval is assigned one of five states. They are not points on a true-to-false line — they describe genuinely different evidence situations. A claim that credible experts dispute is not "half true." A claim nobody has documented yet is not "probably false." The states exist so the verdict can say what the evidence actually says — including when it says "we don't know yet."
Well-supported — high-quality sources agree
A substantial body of high-quality evidence supports the claim, with no credible dissent of comparable weight. Settled empirical facts, well-documented events, and official confirmations land here.
You'll see this when: a claim matches the documented record — a published statistic, a confirmed announcement, an established scientific finding.
This is not: a claim with lots of sources. Weight is judged by source quality, not source count — one official document outweighs ten aggregator posts, and the sources shown on every claim let you see that for yourself.
Contradicted by strong evidence
A substantial body of high-quality evidence contradicts the claim. This is the mirror of Consensus — the same weight of evidence, pointing the other way. A fringe voice in support does not pull a clearly contradicted claim out of this state.
You'll see this when: the claim conflicts with the documented record — a fabricated quote, a misattributed image, a statistic that no credible source supports.
This is not: a verdict we reach lightly for claims about named people or businesses. Those carry a higher evidence bar — if the contradicting evidence isn't overwhelming and unambiguous, we hold the claim at Insufficient evidence instead, and we log every time we make that call.
Credible sources genuinely disagree — both sides shown
Credible, high-quality sources genuinely disagree while looking at the same body of evidence. This is the state most fact-checkers can't express — and the one it matters most to get right. When a claim is Contested, we summarize both sides and show which sources hold each position.
You'll see this when: experts of comparable standing interpret the same evidence differently — an ongoing scientific debate, a disputed economic figure.
This is not: one dissenter against a clear consensus. Contested requires substantial, comparable-quality disagreement — not the existence of any dissenting voice.
Developing — not yet settled
The claim concerns a story that is still unfolding — a breaking event, a newly published study, a situation whose source landscape is actively forming. Thin evidence here is the nature of the moment, not a failure to look.
You'll see this when: the claim is about something that happened hours or days ago and reporting is still developing.
This is not: a permanent verdict. Emerging claims are flagged for re-checking, and every verdict shows the date it was checked.
Not enough quality evidence to judge
We could not find enough quality evidence to settle the claim, and the topic is not actively developing — it is simply under-documented. This is an honest non-answer, and we label it as one rather than picking a side under uncertainty.
You'll see this when: a claim is too obscure, too vague, or too under-documented for quality sources to have addressed it.
This is not: the same as Debunked. "We can't find evidence for this" and "the evidence is against this" are different statements, and collapsing them produces either false debunks or false reassurance. We keep them separate on purpose.
What we never show
No confidence percentages. A score is computed internally for calibration, but it is never displayed. A number like "72% confident" invites precision the evidence can't defend — and it can't tell you whether experts disagree or whether evidence is simply absent. The state, and the sources behind it, are the verdict.
Want the full method — how a claim moves from submission to verdict, and what we refuse to do along the way? How it works →