The full diversity of the RSE Role

To better understand the current state and problem of RSE, it would be useful to have a clearer understanding of what kind(s) of work are currently identified as such, whether they comprise a uniform kind of work or feature diverse subgroupings with differentiated practices involving perhaps, dominant and minority groups

To better understand the current state and problem of RSE (Research Software Engineering), it would be useful to have a clearer understanding of what kind(s) of work are currently identified as such, whether they comprise a uniform kind of work or feature diverse subgroupings with differentiated practices involving perhaps, dominant and minority groups. From the point of view of policy formation, this would allow people both inside and outside the RSE community to better tailor communication and support, identify bias in RSE literature and funding, and for RSE insiders it allows them to connect to others with similar concerns and to coordinate campaigns to advance their understanding of advanced RSE roles and practices. However, generally the RSE literature, surveys and events only reference or focus on the RSE as organised into centralised teams. While those inside the RSE community are aware of its diversity of forms, those external to the RSE community are not necessarily sensitive to these distinctions and can misunderstand the literature or campaigns (for example, the RSE career path campaign focused on job descriptions but did not make it clear that the job descriptions they were creating were for RSEs in centrally directed teams, so making it confusing for people outside the community when recruiting or managing RSEs).

If there was no diversity in RSE roles, then we would have a monoculture, but the term ‘RSE’ has only existed since 2010, while work in RSE has been ongoing since the 1950s and practised across many disciplines and sites. It is unrealistic to believe that it is a monoculture. It is multicultural.

Monoculturism and multiculturism are terms used in sociology when discussing group dynamics where large scale examples are given for a national or political/religious approach. Ethnocentrism is the practice of framing one’s way of life as natural and valid, and applying that belief system to interpret the characteristics of other cultures. While the RSE community is different to these examples the UK Charity Commission requires that a charity reflects the full diversity of the community they represent so it is safe to assume that multicultural approaches are preferred in the UK and so it is important to understand and recognise the full diversity of the RSE role.

There are distinct difficulties in identifying and understanding the full diversity of RSE roles, many of which result from the job role being on leading edge of the paradigm shift to making computers the 3^rd leg of research, in addition to theory and experimentation are the pre-existing legs of research.

Flexibility versus ambiguity

Flexibility is good during a paradigm shift and for early innovation research especially if there are no constraints (for example ethical or business concerns). Ambiguity can provide flexibility but it can also be a sign of unresolved ideological divides signifying the existence of an emerging profession.

Ambiguity happens when something is open to interpretation and is common in a job role when a new technology alters peoples’ ways of working. Ambiguity provides flexibility but ambiguity is difficult to manage or communicate. To communicate better we need to understand where there is ambiguity and where possible use more descriptive terminology but not to become rigidly fixated on that terminology at the expense of appreciating the variation in the phenomena it is meant to label. Universities and research centres are complex systems and so what works well at one time or place may not at another.

Ambiguity in the role

The term RSE was first identified in 2010 but the role has developed alongside the development of the computer. The role can be very different, on one extreme being generic and so requiring pure software engineering, at another being disciplinary or domain-specific. Here are some illustrations from this spectrum:

• A web expert sited in a university central team managed as a pool who is creating a webservice for medical data input that must be cyber-secure where they need to liaise with the university’s personal data expertise.
• A numerical expert sited in a national research facility supporting an academic research community by developing complex simulation software e.g., for nuclear power production.
• A biophysicist sited in university faculty developing and maintaining an established GUI based application with unique visualization capability.
• A research student who is adding research-relevant functionality to software developed by a research group.

What’s in a name

The grammar of the RSE name can suggest that “research software” is the object of an engineer or that “software engineering” is the contribution of a researcher to a scientific project. It is ambiguous and the meaning of those that identified the term may not be how others especially those outside the community but who have to collaborate with it understand it e.g., research councils or HR work with the term but are unlikely to have the lived experience of several forms of RSE.

The purpose of identifying the RSE job title was to provide a collective identity, allow collective action, improve recruitment, recognition of achievements and career opportunities. However, it is a simplification and so restricts peoples’ motivation to understand or promote the disciplinary specialists and creates a tension between the generalized RSE and the domain specific ones.

Self-Identifying as an RSE

It has become increasingly important to those that want to manage the RSE community to know who is and who is not an RSE. The reasons given for an RSE to self-identify is one of collective professional identity and action but it also validates the existence of those who are tasked with management of   RSE operations. It is important that the collective identity and action include all the people who self-identify, but it does not because the RSE literature, surveys and events prioritise RSE as part of centralized teams. What is the purpose of self-identifying as an RSE if you are not included in or recognized by the RSE campaign. It is a double whammy because to those outside the RSE community it looks like you are being included but you know you are not and have limited means to communicate this.

The illustrations of RSE roles given earlier show that some are in well-established work environments for example a national research facility or faculty while others are in more transient organizational structures within faculty or IT services. In a previous article on RSE whether RSEs have research methods we talked about some RSEs being restricted contractually or through governance of their research activities. An RSE in a national research facility who is prohibited from research activities is very different to one in faculty primarily working as a research student. It is a multicultural environment but without a better understanding of the multicultural nature of the community the community is likely to be led and managed through ethnocentrism.

RSE practices are assumed not investigated

Previously, we considered the need to better understand the research methods of RSEs. While looking at this issue, it became clear that the practices of the RSE are assumed rather than investigated, and we alluded to the idea that there are probably multiple RSE roles which have different practices and research methods. Anecdotally, in the MET office, it seems there are 3 RSE roles, although they are not given the RSE job title, and vary from being mainly software engineering to being an equal member of a research team. In a previous section, we suggested that the RSE cultures cross different organizational structures and contractual agreements, but they also cross closeness to research and use of research methods.

The dynamic nature of computing and paradigm shifts

The RSE role changes as technology changes and as the ideology on how to use computers changes. The dynamic nature of the technology and environment mean that the RSE role needs to be able to adapt, there may be academic researchers who inadvertently become RSEs or RSEs who inadvertently become academic researchers. We think this has already happened with AI experts and is likely to happen with Quantum computing experts in the mid-term. It would be good if these experts could easily transfer from one career path to the other and there was greater parity between those career paths.

There is a current expectation that software and its development will not exist in its current form in 10 years-time. The RSE job roles and descriptions are fixed into our current understanding of what software is, which means the newly defined RSE job descriptions and career path have a limited shelf life (which, again) is probably less obvious to those outside the RSE community.

This means future changes to the role and job description will be needed. Despite the change in what software will be, we expect there to be a continuing need for research technology professionals (RTP), who are increasingly being called DRI (digital research infrastructure) professionals; terms used by UKRI that include the RSE role and could replace the RSE job title. We should aim to make this new job role more generic and less fixed on a particular technology as these can easily become dated.

No clear common purpose

A message is easier to understand if it is simple, but universities and research organizations are complex systems, and the simplifications that have been used to promote the RSE are unlikely to work for all areas of research or all research institutions. The RSE role is complex, and simplifying this into, e.g. a job description means focusing on the dominant group and causes problems for the RSEs in non-dominant groups. If the RSE leaders do not understand the full diversity of the role and culture, then they will have to rely on their own experience and empathy, which is limiting and ethnocentric.

The development of a new RSE profession must include all parts of the community; otherwise, RSEs will be encouraged to use methods that are not appropriate for a project or face workplace repercussions, as discussed in our article on the development of new digital professions.

The RSE spectrum

We gave an illustration of RSE job roles in the section on ambiguity in the role. This illustration follows a spectrum from that nearest to a straight software engineer to that of an academic researcher, which goes from low status to high status in a university setting. Within a university context, there is a boundary in organizational structure between research-active faculty and the RSE. This boundary, in some instances, benefits the RSE and, in some, the academic.

Some examples:

• Credit for research is probably the single biggest issue that shapes the RSE role and the placement of RSEs within the research social order. Academic papers are the established paradigm for the assessment of research to which new research outputs are currently being added as the established paradigm for research evolves. This will take many years, but when it happens, it should reduce this organizational boundary. Currently, research software is not as highly valued as a research output as an academic paper in practice, even if in policy, they have the same value. The Research Excellence Framework (REF) is used to evaluate UK academic research and this is changing to improve the research culture and to include non-traditional research outputs.
• A RSE may be asked to help shape a research funding application, but generally, they are not the lead applicant. In some universities, they may be contractually limited from research activity, including being a research lead, but some may be able to take on honorary academic status in order to apply. This is not to say it is easy for academics, some academics may also be limited in their ability to apply as they may not fit with their organization’s research goals or may not be internally selected to apply. RSEs are asked to design the necessary software and provide a costing and work plan for a funding application. However, as the peer review panels for research funding applications have a very limited number of RSEs even if RSEs were more involved in leading or designing funding applications there is not the expertise in the reviewing pool to recognise their contribution.
• Academics can have very high teaching loads and as teaching is the main income for a university it is not only important to society but also to the university business model. RSEs do not tend to teach educational modules unless they have a honorary teaching role but they often provide skills training, the difference being that there are no exams set by RSE led training, no qualifications and no accreditation process.
• Academic researchers supervise research students, allowing them to run short projects to test new practices and ideas. RSEs, especially those in RSE teams, are generally not able to supervise research students, and as good management of an RSE team means running it at capacity, there is likely to be limited time to test new practices and ideas. This is not surprising because RSEs are not accredited or allowed to teach and assess students and RSE is not taught to students as it is not accredited as part of any degree program, this can only happen when RSE activities are accepted and accredited to a recognized profession. Remember professional bodies both recognize and set a standard for a profession as well as accredit student qualifications and as RSE is not recognized or accredited profession it is difficult to formally teach and issue qualifications for it. However, this limits research students from having supervisors with strong software engineering skills.

Restrictions that stop innovation in practices

There are many difficulties in carrying out the RSE role. Many of these are the result of the role being on the cutting edge of a new paradigm of research, the adoption of computers. This puts RSEs and related professionals in a chicken and egg situation. Which comes first the change in the research paradigm and the consequent change in research organisation or the change in RSE practices?

Removing, or at least reducing, the restrictions that stop innovation in RSE and research practices is a good place to start. This would boost research outputs and improve the efficiency of research organizations.

1. Understanding the diversity of the RSE role and RSE workplace environments to allow better communication and support.
2. Understanding and reducing restrictions on RSE research activity, starting with contractual or governance restrictions.
3. Allowing RSEs time and funding to reflect on, share and publish their research practices as well as their software engineering practices.
4. Involvement in supervision of research students.

This piece was written and provided by Dr Joanna Leng, School of Computing, University of Leeds, UK, Dr Phillip Brooker, School of Sociology, University of Liverpool, Emeritus Prof Wes Sharrock, School of Sociology, University of Manchester

Acknowledgement: This work was supported by the EPSRC grant EP/R025819/1.