Experiences in evolvability research

Many technical products and systems nowadays have functionality that is largely determined by software, so called software-intensive systems. The requirements for software-intensive systems change over time, causing the system to evolve. We define evolvability as the ability of the system to respond to such changes. Improving evolvability of zsoftware-intensive systems was the goal of the Darwin project. The vision of this project consisted of four cornerstones. In this paper we share the obtained experiences, insights, and results. We have collected some evidence that three of the vision’s cornerstones, which are about knowledge, i.e., extracting knowledge, representing knowledge, and economic decision making, improve evolvability. The representation of knowledge in A3 architecture overviews is the result with the most evidence that it is useful in practice. 2012 Elsevier Ltd. All rights reserved.