Risks analysis

To evaluate a "risk " means, from existing data, to compute the probability that a similar event might occur in the future. It might be, for instance, the breakdown of some compo-nent, of a human error, a severe earthquake, and so on. We distinguish between two types of risks: ordinary (usually covered by insurance or guaranty) and extreme. 1. Ordinary risks If the risk deals with the failure of an industrial object, typically the questions are: How many tests are necessary in order to evaluate the number of returns, after the product has been put on the market? How much will the guaranty cost? What preventive main-tenance is necessary, and how frequent should it be? These are "ordinary risks " in the sense that one has to balance the number of occurrences and the cost of each occurrence. Our specialty here is the critical analysis of the models. Quite often, indeed, in most commonly accepted situations, the probabilistic laws are oversimplified. People admit that the variables follow a Gaussian law, or a uniform distribution. Or, quite commonly also, people take the data gathered in some situation and use them in another, which is not correctly related. The same way, in many cases, people consider as independent th