Both system acquirers and system developers care about the complexity of systems engineering efforts. Widely believed to cause problems, complexity is somewhat mysterious and undefined. If it could be defined and measured, then less complex solutions can be selected, ideally leading to more cost-effective products and more reliable projects. Dr. Sheard’s doctoral research created a complexity taxonomy, identified potential measures of complexity that would make sense in the domain of system acquisition and development, and obtained measures of those complexity variables for 75 completed projects, along with values of cost and schedule overrun and performance shortfall.
The variables were tested as to which correlated with outcomes. Three variables correlated with all three project outcome measures and an additional twenty variables correlated with at least one outcome. Interesting groupings of complexity measures to outcomes suggest that some complexity measures predicted product outcomes (performance) but not project outcomes (cost and schedule), and other complexity measures worked the other way around. The presentation concludes with suggestions as to how this complexity measurement scheme should be used to assess potential system development and acquisition problems, track complexity reduction, and mitigate risks.