FAKE WARNING: PLAYING WITH IMPRECISE PREDICTIONS

Probabilistic models play a major role in risk assessment and prevention of disastrous consequences of extreme events such as volcano eruptions, earthquakes and tsunami. Probabilistic models play a major role in risk assessment and prevention of disastrous consequences of extreme events such as volcano eruptions, earthquakes and tsunami. The prognoses issued according to these models offer, as a principal result, early warnings concerning the approaching disaster. Nevertheless, false warnings may also appear, leading to unnecessary panic and waves of painful emotions which all deprive society of essential resources. While the necessary scientific background is acquired during the university years through specialized courses in Probability and Statistics, there is an a priori drawback to the use of this theoretical basis. In fact, each model relies on statistical data collected for similar major events already present in the region under study. The estimation of the parameters is thus deeply influenced by particular characteristics. These in turn determine the bias function which controls the difference between the estimated value of the parameter and its real unknown value. Inspired by the complexity of warning systems, we have designed our present game with the goal to enhance the understanding of some important notions which govern the prognoses based on statistical tests: biased and unbiased estimation as well as confidence interval for the estimated parameters. Included in an e-learning game, these notions are embedded into a scenario, enabling the passing of our hero from the initial “ocean level” in which he gets a secret tsunami warning, through the “lab level” in which he analyses oceanic earthquakes, to the actual “tsunami apparition” game sequence.