Abstract
Framed by a patterns of history of robust statistics advantages and disadvantages of the robust processing data are explained and put in the proper context. The discussions are illustrated by series of tables demonstrating the real state of situation, not repeating the statistical folklore e.g. about the loss of efficiency of the robust estimators.
Examples of accompanying diagnostic tools and modifications of the ''parent'' method of the least weighted squares for the various situations, e.g. of broken orthogonality condition, are also presented.