Human verdicts
Measurements are checked by people. After watching a moment, a reviewer records what they saw, and Overshoot keeps that label without any name attached.
Did a person watching agree with the measurement?
What it is
After watching a flagged moment a reviewer picks one of four labels: looks ordinary, looks unusual, can't tell, or unusable clip. The label describes the moment, not the person in it. It is saved with the match and tick and an anonymous token, never a name or Steam ID.
Why people stay in the loop
Valve's own review portal, launched in August 2026 according to press coverage, uses human labels to train its detection rather than to ban directly. FACEIT is reported to require several signals before acting. Overshoot follows the same principle: measurements point, people decide.
What the labels are for
Labels let Overshoot check whether its measurements agree with what people see, and where they do not. They are how the measurements improve.
Where it is used
This is in the research fact sheet today. No labels have been collected yet, and it is not in the app yet.
What this cannot tell you
- Reviewers disagree, so agreement between several people counts more than one label.
- A label about a moment never says anything about a person.
- Until enough labels exist, no accuracy figures are claimed.