If you want to maximize the impact of research software on science, Open Source is essential but not sufficient. You also need to provide high quality documentation, workflow descriptions, demonstrations, and tutorials.
Rob, Jason and Willem are at the strategic heart of developing the eScience Technology Platform – An online platform that aims to stimulate collaboration and the reuse of software and knowledge. Rob: “2016 was a special year for the eScience Technology Platform – with active involvement of the whole eScience Center team we designed, developed and launched a new website”.
What is the eScience Technology Platform?
Rob: “The eScience Technology Platform is not just a place where we upload the software that we develop. The goal is to maintain a website that contains all our software, workflows, documentation, demonstrations and training resources. It is an essential instrument to make the knowledge and software that we develop in our scientific projects available to the scientific community.”
Willem: “It is important that we find ways to stimulate the reuse of the software that we develop, and simply making that software Open Source is far from sufficient. Without good documentation, tutorials and promotion, the developer is likely to be the only user.”
Without good documentation, tutorials and promotion, the developer is likely to be the only user
Jason: “Every year we initiate around 15-20 projects in collaboration with researchers. Those projects start with a research question, for example ‘What is the relation between politics and media during and after political pillarization in the Netherlands?’ Together with the project team we explore possible directions to apply or develop the software that is needed to answer that research question. At the end of a project, the researcher will have a tailored solution to his or her disposal to continue this research. But it shouldn’t stop there.”
From left to right: Willem van Hage, Rob van Nieuwpoort and Jason Maassen. Photography by Elodie Burrillon.
What should happen next?
Willem: “We want all researchers to benefit from the knowledge that we gain and the software we develop in our scientific projects. But we can only do around 50 research projects at a time. We learn many new things in the process of tailoring existing and developing new software. How does specific software work? Where can it be applied, and how? What is the latest achievement in Natural Language Processing and is that useful for this sociology project or for that humanities project? Or maybe not?”
Rob: “This knowledge and experience that Willem talks about is crucial when you want to reuse software or workflows in your research. And that is why Open Source is essential but not sufficient. You need to provide high quality documentation, workflow descriptions, demonstrations, and where possible training resources accompanying those Open Source software.”
That sounds like an ambitious effort
Rob: “It is. And eventually it will have a huge pay off. Imagine the benefits for other researchers if they can get detailed information on how we translated specific scientific questions to the application of specific software, why we chose that software instead of something else, and most importantly how we tailored and applied this software to the specific needs of the researcher. We want to share that accumulated knowledge, so that researchers don’t have to acquire all of it from scratch with the start of every new project.”
Jason: “Another significant fact to mention about software is that it is not domain-specific. We regularly apply software across domains – astronomers, climate- and medical scientists for example all make use of image recognition software. Point cloud software to render 3D models of environments or objects can be used by archeologists, medical scientists, climate scientists and likely many more. And you can of course imagine that software to visualize data is of interest to any scientist.”
We regularly apply software across domains – astronomers, climate- and medical scientists for example all make use of image recognition software
Willem: “It does come with an additional challenge. If we want software to be applicable for other researchers and across domains, it is important to strategically develop software and provide knowledge in the form of documentation and tutorials so that it can easily be reused in future projects. We refer to that step as ‘generalization’.”
As a researcher, can I download and use your software, guided by your documentation and tutorials?
Jason: “You do need some knowledge of scientific programming if you want to use and tailor our generic software for your research – the users we have in mind for this platform are Research Software Engineers who are involved in research projects across scientific domains, such as the Humanities, Climate Science or Chemistry. Our generic software will be valuable for them in the same way as it is valuable for us when we re-use it across projects and domains.”
What was the biggest achievement in 2016?
Rob: “Our biggest achievement was to align the whole team behind the effort. We are all very dedicated to the effort, and we frequently discuss how we can achieve the most impact on scientific research with the eScience Technology Platform. Also, we attracted the interest of several partner organizations to whom we presented our progress. That positive feedback tells us that we are on the right track, but we also see many opportunities and challenges in further developing the platform.”
We frequently discuss how we can achieve the most impact on scientific research
What is your biggest challenge?
Willem: “Finding ways to structure all that software and knowledge, and making it easily accessible and user-friendly. How do we make sure that Research Software Engineers find what might be useful for their research as fast as possible? That relates to another challenge, which is how to make Research Software Engineers and domain scientists aware of the eScience Technology Platform. We will focus on those questions in 2017.”
Jason: “What is also challenging is measuring the impact. What is the best way to measure whether and how researchers are using our software? It is a question faced by research software developers in general. In 2017 we will start an alliance with CWI to address that question.”
What does the future look like?
Rob: “The long-term ambition for the eScience Technology Platform is to become an integral part of the national e-infrastructure next to, for instance, the hardware provided by SURFnet and SURFsara. On the short-term, the platform will contain our core expertise, knowledge and software. On the longer-term we might be thinking of joining forces with other organizations that develop research software, and the platform might grow into a national hub for research software. In any case, we have to realize that software is nowadays at the heart of scientific discoveries and the current eScience Technology Platform is an essential step in re-using high quality software and maximizing its scientific impact.”
We have to realize that software is nowadays at the heart of scientific discoveries