Jacob Krüger

• Home
• Research
• Team
• Publications

Research

I am a software-engineering researcher specializing on the development and evolution of variant-rich software systems. Variant-rich systems exist in different customizable variants and have become the most widely existing form of software systems in all areas of society. My goal as a researcher is to improve our foundational understanding of how the respective software evolves, focusing on the interplay of human cognition and software quality. Evolving variant-rich systems is particularly challenging for developers, and during my PhD studies I have found that developers’ comprehension of their system is the most important factor to engineer high-quality (variant-rich) software. These findings have fascinated me to follow a scientific career on this topic, for which I provide a conceptual sketch in Figure 1.

Typically, developers start to develop and evolve a single system (1). When this system becomes successful and has to be adapted for new customers, the developers start to create clones (e.g., via forking on GitHub) of the system, thereby creating a variant-rich system (A), that can be easily modified (2)). With the increasing number of cloned variants, the developers involved easily lose their understanding of the system (B), which, in turn, can amplify quality problems that propagate through all variants (C). During my PhD, I was particularly concerned with moving from such a fork-based variant-rich system towards a product line (3), which represents an integrated and configurable platform of all relevant variants to ease future evolution (4). With my PhD, I contributed to tackling long-standing problems around developers comprehension of variant-rich systems (e.g., ) and the involved economics (e.g., ) to decide whether moving to a product line is beneficial for an organization (3). To obtain this data, I have extensively collaborated with industry (e.g., Axis AB). Other companies have also been interested in and used the data for their decision making, too (e.g., Danfoss). Importantly, while my work on the economics of variant-rich systems has confirmed and systemized many common beliefs, it has also refuted and challenged established assumptions researchers made about the benefits and drawbacks of cloning compared to product-line engineering (A).

After my PhD, I have continued to intensify my research on the interplay of developers’ cognition (B) and software quality (C) during the evolution of variant-rich systems (A). For instance, in our recent works on developers (B), we explored how the COVID-19 pandemic impacted developers that were involved in developing contact-tracing () and certificate-verficiation apps (). Among others, we found that the involved developers faced new challenges due to the emergency nature of the situation and public demand for their app, suggesting directions for applied software-engineering research to support developers in other emergency situations. Moreover, we have contributed a large meta-review in which we connect different areas of software-engineering based on a concept we call software-change intentions (). A software-change intention captures what the developers actually wanted to achieve when evolving their software. This notion has been inherent to various research topics on software evolution, but this concept has not been studied explicitly before. I try to transfer my insights on developers’ cognition by engineering prototypes that can support developers in comprehending their software. Most recently, we have proposed a number of visualizations to make sense of variant-rich systems that are based on forks and their evolution (). Currently, we aim to extend our current prototype into a useful tool for practice.

Regarding the consequent impact on software quality (C), we have conducted several studies on quality problems that may arise from developers’ misconceptions and their potential impact. My focus here lies on architecture smells that challenge future evolution and system vulnerabilities that can cause harms. For instance, we have recently interviewed developers at ASML to derive a catalog of unintended dependencies in their software (). While an unintended dependency does not mean that the software fails, developers may not anticipate that it exists, which challenges their comprehension and can later result in a smelly system. We have further studied to what extent the quality of variant-rich systems can be improved by reusing the contributions of other developers (). Specifically, we have studied whether the test cases within specific forks could be used to improve the quality of other forks.

Besides these core topics of mine, I have also published papers on improving scientific methods and practices. For instance, we have studied the reliability of literature search engines offered by computer-science publishers when we noticed during an own literature review that these returned different and inconsistent results (). Such publications are a byproduct of my core research.