Authors: James Hetherington Alan Turing Institute, Ben van Werkhoven Netherlands eScience Center, Robert Haines University of Manchester, Stephan Janosch Max Planck Institute of Molecular Cell Biology and Genetics, Paul Hagan British Oceanographic Data Centre, Samar Elsheikh University of Cape Town, Radek Lonka Norwegian University of Science and Technology
In this article we explain how to get an RSE network or association started in your country, region, community or institution. It first summarises the argument that has worked in other places, and the tone and style to take in persuading researchers of the value of RSEs. After that we list a number of practical steps that are often taken in starting a network.
In making the argument for RSEs, do not emphasise the need for a better, more stable career for people like you! Nobody cares. Honestly. Instead, the argument must be based on how a well-established RSE role will make science better.
The core argument
The argument for RSEs has five core steps. Each needs to be made. Do not miss out any steps!
Firstly, you need to establish the importance of software for research. Almost all science uses software now. This includes data-analysis scripting and experimental control software as well as modelling and simulation.
Secondly, you need to demonstrate that the current practice of scientific software is flawed. Be familiar with the discussions around science’s “reproducibility crisis”. Discuss how each new project often re-invents the wheel, starting again so that we do not stand on giant’s shoulders as often as we should. Discuss how slow-running software doesn’t just waste resources, but also blocks opportunities for exploration and experiment. Suggest that research can be made more impactful through usable user interfaces.
Thirdly, explain that there are well-known techniques from professional software engineering that can solve all these problems. (Version control, automated testing, disciplined-yet-agile project management, parallelisation, web-based interfaces.) Software engineers are trained in these, and can help spread these practices throughout your postdoctoral and PhD student community.
Fourthly, explain that working on software at the frontiers of science and scholarship requires an understanding of the research context. Generalist software engineers without a research background can find it particularly difficult to get to grips with the mathematical aspects of a research domain, its vocabulary, and with the cultural assumptions of the academic community. They can often bring over-engineered “Enterprise” solutions, and create software that is not sustainable by the research community after they are gone.
Finally, explain the RSE role itself: a hybrid role, both fully a part of the scholarly community and an IT professional. Interested in the research context and capable of understanding the scientific literature, but not following their own personal research agenda. Incentivised and measured by code output as much as through traditional academic metrics. Embedded within the research community to allow practices and knowledge to diffuse, but with a professional network of other RSEs. A complement to the existing postdoctoral career structure, for those who want to focus on the software side of their work.
To recap: “Software is important to science, scientific software is broken, software engineering can help, but you need software engineers with scientific understanding, so you need research software engineers!”
The tone to take
It is important to frame arguments in favour of RSEs, and RSE organizations, in terms of how they will support the research community and, indeed, how they will support and improve scientific practice. A key point of focus here is that of support. RSEs are expert software engineers and have a wealth of experiences of using computational methods for research that many researchers, however eminent in their respective fields, don’t have, so it is important that an atmosphere of collaboration and mutual benefit is preserved. No one ever won hearts and minds by being overly critical of someone’s ability or working methods. Support, advise, demonstrate; don’t sneer, mock or be aggressive.
There are enough people in academia and research who give the impression that they are a know-it-all who thinks everyone else is stupid. Don’t add yourself to that group!
Sometimes we can uncover uncomfortable truths in the course of our work. You may offend people by pointing out that they’ve been following bad software development practices for decades. Be gentle - start by helping. Remember that what is bad practice now, may not have been bad practice a few years ago. If you uncover errors in code that has been used to produce scholarly outputs, such as papers, then you are duty bound to flag this to the researcher in question. It is not your responsibility to report the errors to the editors of the publication directly. Leave the decision of whether to retract or update to the research team as a whole; do the errors, once corrected, invalidate the science? That is a call that you have to make with all involved.
A related argument against the involvement of an RSE in a research project might be, “I don’t want to share my software with anyone; you might find bugs in it”!
You will also need to cite evidence for each of these points in the argument. You can find evidence for this in the results of a survey of your research community, (see below for more detail on how to do this), and can use headline figures from the results of previous surveys in other places, and our catalogue of case-studies. For the 2014 UK Research Software Survey see here
This section contains a number of practical step that are intended to help you getting started with creating your own RSE network or association.
Stories and case studies that illustrate the impact of RSE involvement in research projects can be a powerful way to win support. If you join the RSE Slack and ask for examples you will be likely to find plenty of material. This will be even easier in the future as an international RSE “evidence bank” is under construction where we can pool our best resources to help with this.
Have/Attend a workshop at an e-Science conference or two
So where and how do you find people to join forces with? First we are in lucky situation to have running conferences and workshops about research software/research infrastructure. The next big ones are the RSE18 in September and eScience 2018 in October/November. If you go there you might bump into people from your neighbourhood. Check the attendees list. You can also ask the organizers beforehand, if they expect someone from your country. The organizers probably did some outreach and are willing to help you. If you cannot find anyone from your country, don’t worry, you might find someone with the same cultural background from your continent. There must be someone, right? And meeting someone face to face helps way more to build a trustworthy relationship.
A core team
As soon as you are aware of the RSE role you can start looking for people who do this kind of work. You might find colleagues in your field just next door or you may need to leave your field and start exploring unknown territories. Start with a simple question: Is software being used at your place? If yes, who are the users/creators of this software? Inviting these persons to a tea/coffee might work just fine. Sometimes your colleagues travel around, ask them to keep their eyes open. They might be able to help spotting interesting people which you then can contact remotely.
What would be topics you can chat about?
- How important is research software in your context?
- Who are the users?
- How is it created and maintained?
- What does your career look like?
A survey of the research community
Holding a survey is not just there to gather information it is actually also a great way of spreading the word and getting in touch with people that would want to be involved in the RSE association that you are setting up.
There are basically two types of surveys that you could organize targeting different audiences. The first type targets all of research, and is about gathering the evidence or need for RSEs in the first place. A great example of this survey was carried out in the UK in 2014, which showed the very large role that software has in research and the need for people that can do professional software development of research software.
The second type of survey targets RSEs specifically. The goal here is really to outreach to the RSEs, getting a better picture of the where the RSEs are, what they are doing, and if there are any problems that all of them are dealing with in their jobs. Keep in mind however, that this should not boil down to being an employment satisfaction survey. The reason you are doing this is not because you and your co-workers want better career opportunities. Instead, keep the bigger picture in mind of trying to fix the problems with software in science.
Practical advice on setting up the survey is to get in touch with Simon Hettrick and Olivier Philippe about organizing a survey for RSEs in your region or country. There is a repository on GitHub that is used for creating the RSE surveys in different countries including the UK, Germany, Canada, South Africa, the US, and the Netherlands. See here for more information: https://github.com/softwaresaved/international-survey
Adjusting and fine-tuning the survey to your specific region is very important and this will take some time. It’s important to understand local data protection laws as well as knowing what you can and cannot ask depending on your culture.
Spreading the word about the survey is best done through a bottom-up approach. It can take forever if you try to organize this top-down involving different organizations. In our experience, the best way is to ask the people you know to spread the word and try to get people who write newsletters to also include an announcement of your survey. And of course use social media, including LinkedIn and Twitter.
Press and blogs
Wider publicity for the RSE movement through established media outlets is a great way to raise awareness, and drive engagement. Don’t be afraid to turn to friends or colleagues who may be able to give you a plug through their prefered channel, or reach out to specialist journalists who may have a professional interest in promoting your activities.
A feature or blog-post in a prominent publication is obviously a great way to reach a wider audience, but Twitter, Facebook, Linkedin are also great platforms for sharing stories and gaining traction.
Exemplar Initial blog posts:
- The Craftsperson and the Scholar
- The researcher programmer, a new species?
- Is the work of scientific software engineers recognised in academia?
- Introducing NL-RSE
Get an article in a magazine! (Make friends with a journalist):
- Save your work – give software engineers a career track
- Researchers both rely on software - and overlook it (originally appeared in Research Fortnight)
Get someone to give a talk
A great way to make a compelling case for the importance of RSE groups - be they local, regional, or national - is to draft in an evangelist from the existing community to present for stakeholders, either formally or informally, on the virtues of RSE.
You can contact the RSE through any of our online channels to initiate a conversation about how we can help, and find out about the availability of our team to visit in person.
Forming an organisation
Finally, after clearly defining the needs of establishing a national RSE organisation, setting-up its core team, raise awareness of what it can provide to local research communities and individual researchers, it is time to form the organization. This could be through creating a website, mailing list, github repository and a regular national reports to keep researchers updated.
The general advice to get this off the ground is to start small. This means without a legal entity and without any payment fees or formal registration to join the association. In the UK at the start, and also in Germany and the Netherlands, all it takes to join the association is to subscribe to the mailinglist. This lowers the entry barrier and also makes it very easy to setup.
Regardless of the amount of communication on your mailinglist at first, using a mailinglist has been invaluable as a measure of membership. It allows you to quickly see how many people are interested in your cause. Note that every researcher has an email address, and that not everyone uses Twitter or Slack.
The mailinglist itself has become a valuable job advertising resource for UK RSE, which recently moved to weekly digests for job adverts to keep the volume in control. It’s best to use the mailinglist for announcements and use other means of communication, for example Slack, for discussions.
Make your organisation known to the world
Now that you’ve established a national RSE association, it is time to connect your association internationally! Please make your brand new association known to the international network of RSE associations!