Building Bridges: Effective Mixed Teams of Software Professionals and Research Academics

Researchers develop some of the most conceptually complex code in the world, but good software engineering practices are typically absent. Both software engineers and researchers share a responsibility because they are both typically concerned with differing aspects of software development; engineers are about quality and process, whilst researchers are about the science and rarely do the two see each others point of view. Software is a means to an end for researchers and is treated like a stepping stone to the final result. Very lightweight software development methodologies have the best uptake. Techniques that focus on the output are most successful, for example unit testing. Researchers have the domain expertise and consequently do the majority of the software development, they also cannot be coerced into using specific software techniques. The secret to working together is to start with small specific code techniques that have immediate benefit. Any improvement in software development should be seen as a win. Once the group has been working together the researchers may have enough trust in the software engineer to try less short term techniques, for example design documents. Researchers only write code because they have to and want to minimise the time they spend coding. If the software engineers can show they can help researchers do this, researchers will take software engineers advice.