Software engineering still is an ever changing and to some degree immature discipline. To achieve higher software quality and productivity, we must introduce well-understood and tested practices and technologies in practical software development. But what is a "well-understood" technique, and in which contexts can it be applied?

Over the last decade, it has become clear that software development practices and technologies must be investigated by empirical means in order to be understood, evaluated, and deployed; and empirical research in software engineering has made considerable advances to support decisions related to software development. Yet, as software evolves, so must research in empirical software engineering.

The first challenge for empirical software engineering, as for software engineering in general, is that modern software becomes much more a service in an interconnected world, with distributed development and operation, short release cycles, fast correction deployment, and failure models that are very different from traditional, monolithic systems. All these features challenge traditional findings on what makes software development successful, and due to the complex nature of development, require new research methods. Such frontiers were first identified in the Dagstuhl Seminar 06262, "Empirical Software Engineering" (June 2006); it is now time to see how to transcend them.

The second challenge for empirical software engineering is more of an opportunity, but brings problems in itself. As more and more development is fully computer-supported (especially in distributed and open-source systems), there is a wealth of data available that can be exploited automatically. While this opens the door for further observational and ethnographic studies, it also presents several challenges due to the wealth of data involved. At the Dagstuhl Seminar 07491 "Mining Programs and Processes" (December 2007), it became clear that the essential techniques of mining facts from archives as well as relating them at a low level are now well-understood; but the challenge resides in appropriate deep empirical assessment.

In this Dagstuhl perspectives workshop, we will address these challenges by bringing together senior researchers from empirical software engineering, program analysis, software mining, and novel software architectures with the added view of new areas in emerging software systems.

Our main aim is to locate synergies between these communities, to identify future perspectives for the field, and to forge strategies for the future. As a result, we envision a manifesto that will guide and inspire research in software engineering for the next decade.