RedFlag

The 54% test.

Eight famous public statements. Each was later either confirmed or contradicted by the documented record — a confession, a court finding, or a full vindication. Watch each clip (or just read the setup) and make the call a viewer at the time had to make.

The bar: across 384 studies and ~25,000 judges, human accuracy at judging speaker credibility averages 54% — barely better than flipping a coin (Bond & DePaulo, 2006). This page is analytics about human judgment, not a detector of anything.

Your result
White line = the 54% research average. Red = you.

Here's the honest part: nobody reliably beats this by watching — not experts, and not any technology. RedFlag is not a lie detector, and no real science supports one. What is measurable is analytics: the moments where a speaker's voice and face diverge from their own baseline — pitch flattening, blink-rate changes, unusual voicing. RedFlag charts those signals, free, in your browser, and what they mean stays your call.

Try the analyzer

How the scoring works · What the evidence says about detection claims