A Population Approach to Ubicomp System Design is a five year research programme working towards a new science of software structures.

By treating a software class as a varied and evolving population of instances, we weave together large scale real-world deployments of mobile software, new tools and infrastructure for software design, and advances in statistical, formal and interactional theory. We are developing ways to reason predictively in a principled way about real world structure and use.

Software developers are dogged by the ‘plug-in hell’ that comes when a program has more components and plug-ins than they can handle. We are building tools and infrastructure that will let developers be more innovative; building more complex systems and understanding the niches they are designing for.

Users lack support for understanding how one’s software relates to others’ and what changes to a program’s structure and use will work and what won’t. We will put that help into their hands, and give them rich connections to the developers and analysts of their software.

The overarching population concept makes all this coherent and interconnected. It drives tools for designers, evaluators and users that support adaptation, and principles and techniques that are informed by and can deal with the scale of human and software interaction.

Our large-scale deployments serve both as useful everyday applications for large numbers of users and as testbeds for our new approach. Such real world deployments ground and combine our technological, methodological and conceptual advances, and inspire new work in our core science. In the course of the programme, we plan for 500,000–1,000,000 users.

This project is funded by an EPSRC Programme Grant (EP/J007617/1).