In many countries, the pattern followed by engineers (and supported by politicians, firms, international organizations and users) rests on two postulates: water shortage and potential water saving in irrigation. It results in expansive hydraulic improvements and leads to understate the importance of low-cost water saving agronomic technology such as direct seedling or laser land levelling. In spite of inefficiency, the narrative to legitimize such practices, including the accusation of "archaic" farmers wasting water, maintains and extends because of a wide consensus that suits many interests. It is not a question here of demonstrating the administration's mistake. Figures are difficult to collect, in particular recycling waters by drains and rivers (IWMI), groundwater refilling or lake evaporation. The point is to understand why uncertainty is not examined by specialists. Consequently, it is rather a question of understanding why engineers favour analytical indicators (plot efficiency of irrigation, conveyance efficiency in canals) meanwhile they dismiss synthetic ones (basin efficiency), which could contradict the water shortage-hydraulic water saving paradigm. Our thesis relies on the structure of the administration, in which specialized skills are bound by a vertical authority perpetually seeking scientific legitimization, the strong compartmentalization of services and high dependence from the political sphere. It is the ideal recipe so that an engineer cannot get all the elements of analysis for an irrigation system. This structure preserves the legitimacy of the high-ranked officials and the ethos or culture of technical expertise required in the country.