... the problem being that for the ordinary building a complete soil-interaction analysis is not being made mainly on lack of knowledge and software power, lack of reliable data able to produce such model and unreliable prediction of the theory and models of the significantly variable behavior of the soils. I agree that the best is the best but only available to some. Not even the codes (the spanish code for buildings at least) are as demanding as your statement implies. It rather builds on past experiences, and at the office level maybe some are doing bracketing for variable conditions of the soil, still rarely the complete soil-structure interaction except for the more relevant structures. This is not to say known aspects of the rigidity of the foundations are not being taken unto account in some simplified manner, but not the state of the art way. That the extant simplified practices may be leading to some oversizing of foundations maybe then not only true, but in fact wished by the code, since it covers, as every code has been doing till today what is not known, whatever the authors or the code say. Furthermore I myself consider the foundations the better place to make some overdimensioning of parts, since from a sound foundation you can cheaperly re-build upwards. To reinforce after failure the foundations is obviously an option but not my preferred, nor I think that cheaper. And not to forget is the fact of the real capacity for proceeding the state of the art way not being in place, but at the top AEC -or specialized- firms. The practices are as they are as a consequence of the practicalities of our available means to design and build, which for no part (not only foundations) do not allow for state of the art, but standard. Standard practice may cover some or even all aspects of whay you say, but in simplified manner. Even in an ACI text of the 90's for mat foundations, the mats are analyzed separately of the structure, congruence being forced iteratively at most. For even simpler structures, it is very usual complete separation of foundation of the structure at design time, and only a guess or data derived value of the allowable pressure on the soil to proceed, and something jsut a bit more smart for piles. Then, come to piles, for some works a size differs of the other by scarce amount in cost, and even if the number of piles is to be high there might be no worth to increase the tightness of their capacity if in doubt.