A match probability answers one question — how rare is this evidence? — not the one that decides a life: how likely is this person innocent? Swapping the two is the prosecutor's fallacy.
So count instead. Among ten thousand possible suspects, a one-in-a-thousand test wrongly flags about ten innocent people. Add the one true culprit and a lone “match” is just one voice in eleven — nowhere near proof.
That is Bayes in natural frequencies: begin with how many are guilty before the test (the base rate), see how many the test flags, and read guilt as the true matches over all the matches.
Independent evidence compounds. Each genuinely separate clue — one the guilty share but the innocent rarely do — multiplies the odds, thinning the coincidental matches until, ideally, only the guilty remain.
The trap is real. Sally Clark was convicted on a “1 in 73 million” that was both miscalculated and misread. Numbers persuade — so make them count the right thing.
Something in the simulation stopped unexpectedly — the lesson continues without it. Nothing you did was wrong; you can move on.