Too Little Too Soon? Modeling the Risks of Spiral Development

Abstract

The DoD's evolutionary acquisition policy is directed against project risk, but bears inherent risks of its own. The DoD policy for evolutionary acquisition mandates multiple product releases via spiral (i.e., amorphous & unplanned) or incremental (i.e., defined & deferred) development methodologies for all programs. All amorphous spirals eventually become definitive increments. Incremental development entails the deliberate deferral of work to a subsequent phase. Computational organizational modeling using systems dynamics reveals that this methodology introduces more concurrency during development, and more variety in production. The result is earlier delivery of the first increment, but with later and more costly delivery of subsequent increments than if conducted via a single-step methodology. Curtailments of scope by the exclusive use of mature technology enable more effective delivery of the first increment, further illustrated by two case studies. Duplication, rework, transaction costs, decision backlog and error are causes of inefficiency in the successive increments. Production variety and mixed configurations produce obvious implications for logistical supportability, training, failure causality, compatibility and interoperability, etc. Further, certain attributes of hardware products might help determine the suitability of this development methodology. Products that are nearly immutable, which have binary requirements for key capabilities, require man-rating, or are maintenance-intensive may not be good candidates for incremental development. Mutable products with costless production, continuous requirements, low maintenance, or time criticality are more likely to reap advantages from this development approach. While modular open systems architecture facilitates system adaptation, modularity itself does not necessarily create evolutionary advantages due to relative modular interdependency. Program managers must be aware of the inherent risks of these agile acquisition methods and take additional steps to balance them with appropriate planning and resources, disciplined change-control measures, organizational accommodations and accountability for configuration management.