System engineering practices have reduced the rate and effect of errors in complex, heterogeneous computing systems, but there are still areas in which they are insufficient. Space systems and other research systems have unique problems that preclude the use of conventional techniques, since the requirements are not all available until after the system is nearly finished. For these systems, the requirements are discovered along the way. That requires a more explicit account of the design decisions and rationales than is usually written down, and a much more explicit connection between the requirements, the design decisions, and the appearance of the code. These conditions also need to be flexible enough to change as new requirements are discovered. In this paper, we show how our Problem Posing Programming Paradigm supports this flexibility of connection, and how some related verification techniques can be used to address the problem, and point towards further development.