How I verify.
Every story on Nexus Signal ships under one promise: rather right than first. This page is what that promise means in practice. It is the sequence every story passes before it is published, written down so that anyone can hold me to it.
State the claim plainly.
Before anything is checked, the claim is rewritten in its barest form. Distortion hides in elegant sentences. A plain one has nowhere to hide, and most errors surface at this step, before a single source is opened.
Two independent sources.
Load-bearing claims need two credible sources that do not depend on each other. A dozen outlets quoting one wire story is one source wearing twelve outfits. Circulation is not confirmation.
The disconfirming check.
The step that feels unnatural: go looking for what would prove the claim wrong. Search against the story, not for it. A claim that survives an honest attempt to break it has earned its place.
Name the weakest link.
Every story has one. It gets named, in writing, before publication. If it is load-bearing and will not firm up, the claim is labelled as developing or it is cut. The reader is told which claims are settled and which are still moving.
Know when to stop.
Verification runs against a hard time limit, because a one-person operation that verifies forever publishes never. When the limit is reached and a claim has not confirmed, the claim does not run. The deadline bends to the standard, never the other way around.
What the sources line means.
At the bottom of every story is a sources line, and it is written to show the chain, not the most impressive name in it. If I read a wire story through republications rather than at the original, the line says so. Aggregators establish that a claim is circulating, never that it is true, and they are treated accordingly. Where evidence is still developing, the story says that too, in plain words.
When verification fails.
Some stories are written, formatted, and good, and one load-bearing claim will not confirm. Those stories do not run. Not later, not softened, not hedged into publication. Including on their own publishing day, the feed simply shows nothing where they would have been. No reader has ever noticed one of those silences. That is the point.
When I get it wrong.
Verification lowers the odds of error. It does not abolish them. When a published story is found to contain a factual error, one of two things happens. If the story's core claim survives the correction, the story is fixed on the page and the fix is acknowledged: a dated note saying what changed and why. If the error breaks the story's core claim, the story comes down. What never happens is a silent edit. A story that reads differently today than it did yesterday will say so itself.
This method is not a guarantee against error. It is a guarantee that error will never come from skipped work. The stories it produces are at nexussignal.org.