Sustainability is all that Software Ecosystems need

Abstract:

The last decades have seen a plethora of large-scale software ecosystems (SECOs) developed by distributed teams and organizations dispersed across the globe. Software developed by SECOs dramatically influences our daily lifestyle, from healthcare to research, self-driven cars, and beyond. Consequently, their sustainability is of utmost importance. Inspired by natural ecology, SECOs are complex adaptive systems of interdependent projects working together for a common objective, each with its own road map and work culture. SECOs continuously gather and process feedback from distributors, which customize an SECO product, and end users, for example, in bug reports and contributed features. A SECO can either be proprietary (e.g., Microsoft 365, iOS), open-source (e.g., OpenStack), or hybrid (e.g., Android), and can span any domain, such as scientific (e.g., R), industrial (e.g., Zephyr), and business (e.g., AWS) communities. Despite the central role played by these SECOs, lack of sustainability is a threat that could invalidate all its benefits. Moreover, coordinating and synchronizing the efforts of projects during an SECO's lifecycle is a wicked problem that requires novel strategies to harness SECOs to ensure software reliability, particularly regarding the socio-technical activities needed during a SECO's lifecycle.

This talk explores SECOs' sustainability across major lifecycle phases through large-scale empirical studies of SECO onboarding, micro-collaboration, release synchronization, and cross-community feedback mechanisms. In each stage, challenges, and benefits of current SECO practices are identified in the context of the famous, long-lived OpenStack and Microsoft SECOs based on software analytics, participant observation, interviews, surveys, and focus group studies. Also, I will discuss how my research on the sustainability of SECOs will investigate how foundation models could improve collaboration between human-human, human-machine, and machine-machine collaboration and ensure software's trustworthiness in a SECO context.

Bio:

Armstrong Foundjem is a postdoctoral fellow at Polytechnique Montreal, working on the DEEL project on Dependable and Explainable Learning. His research focuses on the certifiability of safety-critical AI systems under the supervision of profs G. Antoniol, E. Merlo, and F. Khomh. His empirical research follows mixed-methods approaches at the intersection of sustainability of software ecosystems, intersectionality, social network analysis of SECO communities (GNN), release engineering and DevOps/MLOps, and NLP. He defended his Ph.D. at Queen's University in June 2022, advised by Prof. Bram Adams, and his work has been featured in top-tier software engineering venues such as ICSE, TSE, EMSE, and ICST. 

Zoom: https://concordia-ca.zoom.us/j/83488578351?pwd=ZTFPV1Vzaklwa3BPYXBsTzJyTktvUT09