Breath of Fresh Air – Diversity Heroes: Mary Sheeran
As a community, we embrace our diversity; diversity makes us better, stronger. We cannot do enough to applaud all of our heroes in their diversity. They are people who are ACM members, volunteers or experts in their field. Starting from June 2020, we have been reaching out to a number of heroes about their tech career journey, about their perspective on intersectionality and reflect on initiatives for equality.
This month’s guest is Mary Sheeran (in white, standing in the front row), who is a professor in the Computer Science and Engineering Department at Chalmers University of Technology. She has pioneered the use of functional domain-specific languages in hardware design and verification, and in resource-aware parallel programming. She is a founder member of IFIP Working Group 2.8 on Functional Programming. A couple of years ago, her university asked its staff “how would you like to change Chalmers?” and the answer was obvious to her: increase gender equality, aiming for excellence. The resulting proposal was funded – approximately 30 million euros – and work on the Genie initiative began in 2019, with Mary as the Vice Director. Thank you, Mary, for writing for us.
Improving gender balance in academia: a computer scientist’s view
Computer Science and Engineering (CSE) is a growing and varied field that plays an increasing role in our future and that of our planet. There is a huge demand for our graduates. Here at the Chalmers University of Technology in Gothenburg, Sweden, we have attractive global employers like Ericsson, Volvo Cars and many others right on our doorstep. Volvo Cars, for instance, has a long and successful history of working with diversity and inclusion, both on the factory floor and in top management. Yet, the company still struggles to find female talent (anywhere) in technically advanced fields, and we at Chalmers are not managing to feed the pipeline. A recent report from the Swedish Information Technology (IT) and Telecom Industries (1) highlights the IT Competence Shortage (of 70,000 people by 2024) and its threat to Sweden’s future in this age of digitalisation. Several of the proposed measures involve getting more girls into tech and taking steps to retain them. The message from the European Union is much the same. We need to educate more people, and particularly women, in computer science and engineering. A few years ago, I discovered that we had 7% women in the first year of our degree course in computer science and engineering. In a civilised country like Sweden! Having been brought up in a family of engineers, and with the idea that becoming an engineer is a great choice, both for oneself and for the world, this broke my heart. And my heart sinks, not for my own sake, but for all the lost opportunities, when I see that I am the only woman on the list of 30 PhD examiners at our department.
In my journey to being a researcher, I started out as a masters student in the Programming Research Group at Oxford. I stayed on to do a doctorate and even (briefly) as the first female member of faculty in computing science. I look with nostalgia at this 40-year-old photo of me as a fresh-faced masters student among all the men, with Professor Tony Hoare presiding. But the trouble is that I still see pictures like this, again and again. We need change.
Nearly everywhere, computer science remains largely a white, male, middle-class discipline. We seem unable to educate, attract and retain anywhere near enough female computer scientists, despite the pull from the companies that employ our graduates, and the fascinating theoretical and practical problems we face. I have documented my route to becoming a gender equality practitioner in a talk at Code Mesh (an alternative developer conference) (2), croaky voice and all. That talk contains pointers to further reading. Here, I try to convey what I have learned so far, as well as my hopes for the next steps.
At Chalmers, we are now in the third year of Genie, our Gender Initiative for Excellence. I co-wrote the application for funding from the Chalmers Trust of 300 million Swedish kronor (about 30 million Euro) over ten years with Pernilla Wittung-Stafshede, a full professor from our Biology and Biotechnology Department. She leads the initiative, I am vice-leader, and we form a close-knit team with project coordinator Maria Saline, whose research background is in Biophysics. In theory, I spend 30% of my time on this, though, in practice, it takes most of my energy.
We work with all 13 departments at Chalmers, and it has been a shock to discover how different from each other the departments are, in their culture, how they are run, the real power structures, the current gender balance in the department and associated research fields, level of awareness of causes and effects of poor gender balance, and much else. This means that much of the work on improving gender balance needs to be done at the department level or even in divisions or research groups. Nonetheless, some University-wide advice on best practices can be given.
Four steps towards gender equality in academia
1 Raise Awareness (with the help of experts, both theoreticians and practitioners)
Even though I have spent most of my career as the only female full professor in my department, I was until fairly recently shockingly unaware of the effects of societal norms and unconscious bias. Confusingly, one can simultaneously be privileged (as a full professor with good access to research funding) and subject to what I call anti-mentoring, and thus part of the problem as a victim. Anti-mentoring is a term that I have simply made up to describe my own experience and that of others whom I have spoken to in academia. For me, anti-mentoring happens when I get treated in ways that make clear that what I am doing is somehow not seen or appreciated. So I am hoping for support or a deep discussion in which I am listened to, and instead, I feel dismissed and demoralised. I hope to be lifted and encouraged, but instead, I feel put down. Despite my privileged position and seniority, this still happens to me much more often than is comfortable. But I suspect that young women in academia, and others who don’t fit into the norm, are the most usual victims.
My best advice for educating yourself and your students and colleagues is to start with the film Picture a Scientist (3), now available on Netflix. It is powerful and moving, building on the stories of victims, yet firmly based on love and respect for science.
At Chalmers, Genie aims to raise awareness with a steady stream of events, panels and seminars (4). We have deliberately formed an International Advisory Board that covers the range from theoretical to practical. For example, one member, Prof. Liisa Husu, a top researcher on gender equality in academia, has helped us to prepare a page of recommended reading on and around the topic (5). Another member, Paul Walton, a Chemist from York University who has a long history of working successfully with gender equality in academia (6), provides much needed mentoring to the Genie team, as well as advising heads of department and other faculty as they work for change. Our Advisory Board was formed partly through personal contacts, but also making use of contacts gained by attending the European Conference on Gender Equality in Academia.
2 Gather and publish the data
Know the current numbers and how they are changing over time (7). By this I mean that one should keep track of gender-divided data, for example about retention, for every level from undergraduate students to full professors. And the data needs to be made visible. At Chalmers, the five year engineering undergraduate programmes with the worst gender balance are electrical engineering and computer science and engineering, with 15% and 16% in the first year in 2020. The IT programme had 23.5% women in the first year, and Automation and Mechatronics 24% – rather better, but still much too low. Note that Civil Engineering had 50% female students! At Gothenburg University, our Masters in Applied Data Science had 42% female students entering in 2020, and fully 55% in 2021. In my department (CSE), we have two female full professors, one of whom is now vice rector of the University, while we have 25 male full professors. Among our employed doctoral students, just under 30% are female, but among our industrial doctoral students (who are employed by companies) only one of 20 is female. Overall, in the faculty at the department, about 18% are female, skewed towards the junior ranks. We have work to do.
And just waiting is not going to be enough. The Seventh Edition of Informatics Europe’s report on Informatics Education in Europe (Dec. 2019) (8) presents alarming data spanning 2012-2018 across many European countries; The percentage of female students in Informatics Bachelor’s programmes (first year) goes as low as 5-6% (e.g. Belgium) and generally hovers around 15-20%. Notable exceptions are Bulgaria, Estonia and Romania, with 30-40%. It would be good to learn from these good examples. This is not just a European problem, but a global one. In the US, in 2018-2019, the percentage of female Bachelors graduates was around 20% in most CS disciplines, with Computer Engineering and Software Engineering considerably lower (see Table 4b in the eighth annual ACM-NDC study) (9).
3 Examine and strengthen the academic environment (including recruitment and promotion procedures)
For recruitment to academic positions, there is evidence that broader topics and the announcement of several positions simultaneously lead to larger, more diverse fields of applicants. We certainly experienced this in our recruitment of 10 assistant professors at Chalmers in ten different broad topics, including one in ICT, in 2018/19. From over 1000 applicants, we ended up appointing 7 strong female top-ranked candidates, and three male, and Genie stepped in to fund 5 more strong female candidates who had placed second. This cohort of assistant professors can be seen as holding the key to the future of Chalmers, as it is both strong and diverse. Overall, we now have 55% female tenure track assistant professors at the University.
When working with promotion and recruitment, it is important to understand the cumulative effect of unfair or inaccurate evaluations — and evaluation is at the heart of much that we do in academia (10). We believe ourselves to be fair, but we are not!
4 Create and support a community that works for change
We are many who are working quietly for or just longing for change. Some of us feel alone or overwhelmed by the enormity and complexity of the problem, the slow progress, and the difficulty of combining this work with everything else that our jobs demand of us. Part of the solution is to support each other and spread information about good practices. We need to look out for each other! There is an active Facebook group on Gender Equality in Academia. Perhaps we need a subgroup just for computer science?
I would assert that these actions are all necessary for progress in gender equality in academia. But, unfortunately, I don’t believe them to be sufficient, especially in computer science and engineering. I have been so busy listening to experts, reading the literature and helping to set up the actions described above that I have forgotten to think like a computer scientist! For example, take my department (CSE at Chalmers). It is a network of communicating individuals, organised into groups in various ways. There is a hierarchical management structure with a head of the department and a leadership group, with responsibility for personnel (including faculty recruitment), budget, our research and teaching etc.
But there is a whole other system at play. Every faculty member makes a stream of decisions daily about what to focus on, which research topics to prioritise (if any), where to publish, who to collaborate with, how to do teaching, how to supervise students, where to look for resources, whether to stay or go or maybe have a startup, whether to give up on research, how to support which colleagues, and (at least for doctoral students and postdocs) who to hire. These communicating actors form a sort of substrate in the department, influencing each other and determining the academic culture of the department and of sub-groups. The management structure may have a smaller influence than we might expect on individual decisions by academic staff, but it is important as the link between the University level and the faculty. Of course, in a well-functioning department, the formal and informal systems work in happy symbiosis. In any case, to understand and influence the academic culture of a department, we need to observe and understand the behaviour of individuals and groups, their reasons for making decisions, and how both the culture and outside influences determine gate-keeping decisions that guide career trajectories.
If we want our actors to change behaviour, we first need to understand them better! A department or research group is indeed a complex system. We must model it, somehow, at least informally. Worse still, we need to better understand (and influence) the environment in which the system works. This includes the University and how it interacts with the department through the University management structure and other ways; if we want change, are the necessary incentives in place? Are those with power held accountable? It also encompasses our external environment, including flows of people feeding, and leaking from, our pipeline. If we are not careful, the list of four actions above will perhaps make us all feel better but will have no long term effect. Adding a few female actors while leaving everything else unchanged seems to me to be doomed to failure.
I am convinced that we will only be able to change the system if we understand it much more deeply than we currently do. And I think the work needs to be done for each department and the University as a whole. One of the great dangers of gender equality or diversity work is that it may turn into a bureaucratic box-ticking exercise. Here, I plead for quite the opposite: an honest, serious, deep, and possibly painful analysis of individual academic departments and groups and how they really work (or don’t work). Vitally, this analysis needs to be done by the students and academic staff themselves, perhaps with the help of former staff and even external experts. No amount of online training or obligatory courses on unconscious bias is going to cut it.
This line of thought results in a task that I think we must tackle if we are to get anywhere with gender balance in computer science and similar departments.
5 Understand your academic culture by talking about it openly and continuously
There are many questions to ask, at many levels. What kind of support is provided to newly hired researchers and teachers? What kinds of behaviours are valued? What kinds of people (if any) get put on a pedestal? How do we decide in what fields to recruit faculty and how do we search for new recruits? Are there insiders and outsiders in the department? If so, listen carefully to the outsiders! How does communication work generally? Is there a strong feeling of community in the department? How are collaborations initiated and preserved? Do we emphasise true research quality? What makes individuals decide to pursue research at an international level, and where are the blocks? How well does meritocracy work in your research sub-field? What characterises excellent supervision in your group? How do we choose research problems? What do we do to recruit and retain more female students? How do we interact with funding agencies and with the industry? Can we build a shared research vision with other groups or departments? And a myriad more. High and low. Deep and shallow! I am convinced that we need to tackle academic culture and not just culture in general. And it is not only about studying organisational culture in the abstract but also about studying our own concrete organisations, their complicated power structures, and their many secrets.
This task is much more onerous than it looks at first sight. It is tough to have these kinds of discussions; we are trapped in a system that encourages us to describe only our successes, and indeed to compete with each other! Of course, we are not going to agree. In these last few years, I have learned that academic meritocracy is a surprisingly shaky foundation and one that we must question. We have been lured into becoming box-tickers ourselves! Do you have such discussions? Do tell me! I am very keen to hear from you, especially if you see yourself as an outsider.
My, possibly naive, hope is that opening the discussion will free us to imagine and then actively pursue a better, more supportive, yet still excellent, academic culture that allows more of us to thrive as teachers and researchers, supplying the world with the diverse expertise that it needs. In Genie, we are currently running a course on how to work with culture change for representatives of all departments, so we are taking the first steps in a long term experiment, on which we will continue to report. However, if we end up with all talk and no action, we will have failed.
We all need to change, and we need to work hard on observing and eliminating anti-mentoring. We need to change our power structures and norms so that mentoring, rather than anti-mentoring, becomes the normal, natural, usual behaviour. It is not about changing the women; it is not about changing the men; it is not only about changing individuals! But we won’t be able to change our systems and power structures without some deep self-examination. We must avoid stagnating as many previous efforts to improve gender equality in STEM settings have done.
We should remember, though, that lots of great work are going on in computer science departments all over the world. Borsotti’s paper on an empirical investigation of socio-cultural barriers to female participation in the Bachelor of Software Development at the IT University of Copenhagen (11) is a delight to read. It documents concrete, effective interventions. For example, a free, optional workshop before the semester start for CS students with no prior coding experience introduced the basics of programming in a stress-free environment. It was followed up with a peer-to-peer support activity run by experienced students working as tutors. Activities such as these levelled the field, which is important since female students are less likely to have extensive previous coding experience. The project worked on perceptions and prejudices about who is suited for a career too, through various outreach activities, with the aim of persuading more girls with high grades to apply to study computer science. The initial results were promising, and the head of the CS department reports that they have increased the percentage of women students in BSc Software Development from around 6% in 2015 to around 24% last year and 21% this year. In BSc Data Science, the percentage of women is 37% this year. A good reaction to reading this blog post would be to read Borsotti’s paper and do something similar in your department! Or, if you are in the industry, you could volunteer to help at a nearby CS department!
Much of my own research has been about the problem of designing hardware circuits so that they are guaranteed to meet a specification, rather than coming up with a design and then trying to prove it correct after the fact, as is the current practice. This is an important problem both in theory and in practice, and neither I nor anyone else has solved it fully, yet, though many clever people are working on it both in academia and in companies like Intel and Google. But still, I have to say that it is a very easy, well-defined problem compared to figuring out how a department like mine works, in Chalmers and in the world, and how to make it all work better with any certainty. It is tempting to retreat to the lovely, abstract world of research in hardware design, but I can’t, or at least not entirely! I am faced with a complex, fascinating and important problem, so how can I resist? And the future of our field is at stake! Please join me on this quest.
[Feel free to email me at mary dot sheeran at chalmers dot se.]
(2) Mary Sheeran: Let’s get more women into computer science, Code Mesh LDN 19
(3) Picture a Scientist https://www.pictureascientist.com/ also on Netflix at https://www.netflix.com/se-en/title/81303549
(4) For example, the seminar by Frank Dobbin (Professor of Social Sciences at Harvard) on “Do Faculty Diversity Programs Work? Evidence from 600 Universities over 20 Years”.
(10) Valian stresses this point, see for example the seminar “Inclusive academic science: achieving diversity and excellence” in the series celebrating 20 years of the NSF advance programme, March 2021.
(11) Barriers to Gender Diversity in Software Development Education: Actionable Insights from a Danish Case Study, V. Borsotti, track on Software Engineering Education and Training, Int. Conf. on Software Engineering (ICSE), 2018.