On specifying systems that connect to the physical world

Any engineering method is based on the validity of some assumptions about the problem domain and about the artifacts used in any solution of an engineering problem. Engineers have learned, perhaps not perfectly, to lay bare these assumptions and to base their methods on effective use of these assumptions. Software engineers are less good at this, to the detriment of building effective methods. The paper discusses some assumptions that either are false or are not made properly explicit in methods and the detrimental effect on the relevant methods. Some ideas about dealing with these deficiencies are proposed and discussed.