Some Fundamentals of Insurance Statistics [Discussion]

This review is directed to an attitude summarized in the paper in the following paragraph "All probability and all applications of statistical data are based on partial ignorance. If we knew just how a pair of dice were imperfect, were held, and thrown, and blown, and how the surface on which they bounced reacted, we could predict from tried and true engineering formulas just how the dice would fall. If we knew more about each insurance risk than we do know or even than it is at all practical to determine, we could rate each risk better, and we could build a foundation of statistics which would enable us to rate each risk still better, until, in the ultimate we could predict the actual event insured against so that savings would replace insurance as a means of mitigating the 'risk', provided, of course, that our understanding and our knowledge were both built up far beyond the present ability of mankind to know, and to use knowledge.