Leila Schneps and Coralie Colmez have an interesting piece at The New York Times about DNA evidence in murder trials, the mathematics of probability, and the highly publicized case of Amanda Knox. What good is remembering the math you learned in junior high? If you're a judge, it could be the difference between a guilty verdict and an acquittal.

I remember learning about Ms. Clark’s crib death case, although I hadn’t heard that she died in 2007, how sad. The person calculating the likelihood of two infants both dying due to SIDS in the same home assumed that the two events were independent, which is of course completely incorrect (same crib, same food, same air quality, same genetic predispositions, etc etc). The odds that siblings would die from SIDS is only slightly less than the odds that a single child would die from it. So basically this woman lost her children to a tragic, unexplained syndrome, and was then unfairly blamed and jailed for their murder on top of that. The whole story still haunts me!

As a species, we seem to want to tear apart a member of our pack occasionally, and we’ll come up with the evidence to justify it, no matter how wrong.

In modern forensics probability is frequently critical, and unfortunately the justice system as a whole has absolutely no idea what the word means. More to the point, they’re not trained to ask the right questions, and human intuitions about probability are

really wrong.For me the watershed moment was hearing about the Lucia de Berk case; there’s rarely been a more obvious case of injustice caused by confirmation bias. And poor maths.

Nobody who doesn’t thoroughly understand Bayes’ Theorem – and its implications – should be allowed to answer the question “What are the odds that…” in court.