Centralized execution, decentralized control : why we go slow in defense acquisition

Abstract

The slow pace to field new defense weapon systems is allowing potential adversaries to catch up to the technological advantage the U.S. has maintained since World War 11. Despite hundreds of studies, and a near constant state of "acquisition reform", the problem continues. This research analyzed the defense acquisition process as a socio-technical system, focusing on the source selection process as subset of the Defense Acquisition System (DAS) for modeling purposes to investigate the value of the separation of contracting and program management authorities. Network graphs showed how Conway's law predicted the effect of the separation of authorities on the topographic structure of the source selection process and a high network distance between the separate authorities. An agent-based model was built that showed a 26% cost (112 days) in terms of schedule because of the separation of authorities. The benefit of the separation was investigated by scoring the comments received by the Multi-Functional Independent Review Team (MIRT) for five different source selections and found that less than 1 % of comments had a likely impact on the decision and less than 4% had a likely or highly likely impact on protestability. The results showed that while there is a small benefit to the separation of authorities currently implemented in the source selection process, the cost is very high. Enough data and evidence were generated to recommend taking steps to better structurally combine the two authorities and better integrate source selection expertise into the process.