There is a fundamental disconnect between the process by which research is designed and the process by which software is “programmed.” Developing software for research is fundamentally an iterative process, sometimes even a process of successive approximation. The typical instruction to a software engineer is “Please do X, Y, and Z.” Weeks later, researchers receive and try it only to discover that Z was not correctly understood or is not right for what they want to do. So, the researchers say, “Please change Z to W.” Weeks later, they try it again and it crashes or does something bizarre because the engineers did not anticipate the effect of the change through out the code. Back to the engineers it goes ? and so on and so forth until deadlines and budgets have been trashed and the researchers are angry with the programmers and programmers are angry with the researchers.