The manslaughter convictions of seismologists in Italy offer a timely reminder of science's core functions to both scientists and non-scientists (Nature 490, 446; 2012).

Over time, combining existing scientific data and new methods has allowed scientists to predict possible outcomes and associated probabilities. But the accurate prediction of certain events is, and will continue to be, challenging — particularly in economics, climate research, disease pandemics and natural disasters. Science can predict the probabilities of events occurring under a given set of circumstances, but not the events that will occur.

The public may not fully appreciate probabilities and risk assessment, but scientists must work to change this. Researchers need to be able to present information about uncertainty, prediction and probabilities in simple terms, and to convey information to the public and to government that is accurate, consistent and clear.