sideproof does not decide what is true for you. It does something narrower and more honest: it finds the checkable claims on a page, gathers what credible sources actually say, weighs that evidence by quality, and lays the balance in front of you. Here is the logic underneath it.
The core idea: weigh, don't count
The internet rewards repetition. A claim echoed ten thousand times can feel settled even when not one source behind it is reliable. sideproof is built on the opposite instinct.
Evidence is weighed, not counted. A single primary source, the original study, the official record, the statement on the record, can outweigh a thousand echoes of it. Conceptually, a claim's support is a weighted balance of its evidence, where every source counts in proportion to how credible and how relevant it is:
A weighted average, not a show of hands. Everything below is this idea, made practical.
Separate claims from noise
Most sentences are not checkable. Opinions, questions, predictions, and turns of phrase have nothing to verify against. sideproof first isolates the statements that assert a specific, checkable fact, the ones with a real answer, and leaves everything else alone. If there is nothing to check, nothing gets underlined.
Gather sources that speak to the claim
When you click a claim, sideproof gathers sources that bear directly on it, not loosely related pages, but ones that address the specific assertion, including the ones that disagree. The aim is coverage of the actual question, not a tidy answer.
Grade every source
Sources are not equal, so sideproof sorts them into tiers and weights each accordingly:
- Primary and official records, original studies, statements on the record, sit at the top.
- Established reporting carries real but lesser weight.
- General sources count for less again.
- Unvetted and social posts count least.
A claim propped up only by the weakest tier is treated very differently from one anchored in primary evidence, even if the weak tier is louder.
Read the claim as written
A statement can be literally true and still mislead: by omitting context, by framing a number to imply more than it shows, or by quietly shifting what is being claimed. sideproof weighs what is actually being asserted, not just whether a matching sentence exists somewhere. That is why a result can come back as missing context or misleading framing rather than a flat true or false.
Show the balance, not a verdict
The card you see is the output of all of this: what is really being claimed, what credible sources say, the other side, and the sources themselves with their quality marked, plus a short bottom line. The label at the top is a summary. The sources are the substance. You are meant to read them and decide, not to take our word for it.
Why context, not a verdict
A single true or false stamp hides the very thing that matters: how strong the evidence is, who stands behind it, and what it leaves out. By showing the weighted balance instead of a verdict, sideproof gives you the one thing a label never can, the ability to judge for yourself. When the evidence is thin or genuinely split, sideproof says so rather than inventing confidence it does not have.
What it does not do
- It does not replace your judgment. It hands you the evidence and the weights; the call is yours.
- It does not manufacture certainty. Weak or conflicting evidence is shown as weak or conflicting.
- It does not watch you read. Detection runs in your browser, and only a claim you click is ever checked. See the privacy policy.
Questions or a correction
Spotted a claim it read wrong, or want to understand a result? Email hello@sideproof.sh. Better evidence, and better reasoning about it, is the whole point.