The paper presents a discussion of the difficulties of formulating stable requirements early in complex engineering programs, and the severe consequences on program execution. The problems are caused by the need to seek political and funding support for the program. Formal classical Systems Engineering (SE) and Program Management (CSEPM) methodology is based on the assumption that the knowledge to anticipate all interfaces and create good requirements exists early in the program, and it is only a matter of working out the details to build extremely complex devices such as satellites, aircraft, refineries, nuclear power plants and high speed rail. The paper argues that this works well only for well-understood systems but it breaks down when the knowledge of what needs to be done still needs to be discovered, which is the case with most complex systems. In programs that develop new complex systems, the reality leads to the following Faustian Bargain: "Either develop and anticipate all interactions and requirements early, and allocate them to suppliers when the knowledge is not yet available, then conduct massive, painful, and cost-and schedule-busting requirements changes throughout the program; or delay the subcontracting until the system design is mature, complete and stable, and only then allocate requirements to subcontracting, but then risk the program termination because of the lack of political support and funding." The paper argues that in order to radically change this major deficiency of classical Systems Engineering and Program Management a radical change of the program business model would be needed.