What changed: watching your score over time

AI engines don't agree with each other on the same day, let alone with themselves a day apart. Asking daily would surface a "change" constantly, and almost none of it would be real movement — just the gap between two samples of a noisy process. A monthly re-scan, on the exact same question set, is what turns noise into a trend worth trusting.

A source that entered or left a single question is treated as noise and filtered out — the report says plainly how many changes it discarded and why, rather than pretending it saw nothing. What's left is score movement, questions gained or lost, and sources that started or stopped being read, each one big enough to trust.

If two scans didn't ask the same question set, they are never compared against each other — that would draw "changes" that are really just two different tests, not two points on the same line. The screen says so honestly and waits for a real comparable pair rather than inventing movement.

A free scan is a single measurement, so this screen shows what it honestly is — one data point, with a plain explanation of what would start appearing here every month once a second, comparable scan exists. The monthly re-scan, the alert when something moves, and the recommended work that comes with it are the paid plan.