Abstract
Software development has been often considered as a "standard" manufacturing activity, whose actions can be sequenced and optimized quite like the production of cars. From this the "Waterfall Model" of software production was defined. But, like most human activities, even what people consider a "simple" production of a Cappuccino, cannot be represented as such, and software is definitely more difficult than making a Cappuccino; in particular, in software three major problems occur: irreversibility, uncertainty, and complexity. This has lead to the emergence of novel approach to software production that acknowledges such problems and reverts upside-down the idea of formalizing software processes with a waterfall approach. However, such acknowledgment has not yet been fully received in the area of empirical and quantitative software engineering, where software processes have been still modeled usually with standard statistics and other traditional mathematical tools. We advocate that the usage of Computational Intelligence can translate into empirical and quantitative software engineering the paradigm shift implemented by the new, emerging process models in software development.