How I started caring about what people actually cared about

Five years ago, I was frustrated at work.

There were many things that I could be proud of. I was leading one of the world’s most successful research groups studying the mathematical models of collective behaviour, studying everything from how birds flock and ants form trails to how racial segregation permeates cities and how society’s values change over time. I had received the prestigious ERC Starting Grant, funding from Human Frontiers Science Programme, Vetenskapsrådet, Wallenberg, SSF and other sources. My research group was culturally diverse, containing women and men from China, India, UK, USA, Italy, Uganda, Tanzania, Chad, Ecuador, Australia, Bulgaria, Ukraine, Russia, Germany, and even Sweden.

Despite this, I felt that our research had very little positive impact on society.

We should have had impact. We were studying how groups, including humans, work together. How the social whole becomes more than the sum of its individual parts. This was a question that was becoming more important than ever. Populations are moving to live together in cities in new ways. Social media is creating new ways of communicating. Society needs to understand collective behaviour.

This might have been naïve, but I became a researcher when I was in my early 20s because I wanted to improve the world. I believed that by acquiring knowledge and skills, understanding the basic mechanisms by which social behaviour worked, I could make a difference.

And here came my frustration. Whereas I had no power and little knowledge as a 22-year-old to effect change, now I had both. But the academic culture I found myself in prevented me from using these to effect change. I felt trapped by what is best summarised as academia’s obsession with long-term narrative.

Let me try to explain the problems with long-term narrative using an example.

All academics complain about writing grant proposals, but have you ever thought about how much time it takes to write a research article? Not just the time taken for the reviewing process (which we like to complain about too), but how long it takes to do the actual writing.

You have done your experiment or you have solved a theoretical problem. You can explain what you have done in a five-minute talk, in a poster or over coffee to friends, but now you have to embed what you have done inside the scientific cannon, make it part of the narrative. The process takes a lot of time and is mentally draining: there is a constant fear that you have missed something that your peers will see, it requires very detailed reading because other academics’ obsession with continuity means mistakes will be noticed. But it doesn’t contribute to the actual things you have found out. Very often, whole fields of research become very esoteric, insular and impenetrable to outsiders, even if the questions they are addressing are actually quite straightforward.

The writing process itself isn’t the problem. It is an example of a symptom. Succeeding as an academic involves investing time in creating a self-consistent, but often very conservative, narrative around what it is you do. This narrative develops over the time scale of years. I recently saw a PhD student write that it took five years from first having an idea to seeing people cite it. Before he got that citation, he had left academia, disillusioned by the lack of feedback.

I decided to stop caring about that narrative and start caring about impact. Not impact in the sense of citation indices or how many conferences I was invited to (these are measures of success in creating narrative), but impact in terms of answering questions that people actually cared about. Real people, not my scientific peers.

This is exactly what I have been doing for the last five years. It started with something we called ‘Fun Fridays’ inside my research group. Every Friday we stopped working on our usual research and did something creative. The aim was to take a personal interest project from start to conclusion with three or four Fridays. We took our inspiration directly from bloggers and hobby data scientists, who thought up an idea, collected data and tested their hypothesis. If they could produce results over such short time spans then so could we.

The plan worked. We created several online games, one of which involved chasing an evolving fish population. We studied polarisation in Ukrainian politics. We looked at the best strategy for getting matches on Tinder, the evolution of language, success and failure of sustainable development goals, the merits of different voting systems, gender bias in academic recruitment and found the country that was the most exceptional democracy in the world. Some of it was pure fun. One post looked at how songs climbed and fell out of the charts according to the rules of gravity. Another made music from data he had collected on cicadas and art from fish tracks.

The other four days of the week, we continued doing narrative-based science. We followed the norms of academia and wrote proper articles. Many of the ideas that started form as ‘fun’ projects later became ‘proper’ research. The majority of people who worked in my research group during this period now have permanent positions in academia, a claim very few research group leaders can make. I like to think that, at least part of their success, can be attributed to being given the time to think widely and freely.

At that time, I also started writing about football. Partly because I was interested in it, but also because I let myself be led by impact. Lots of people are extremely interested in sport, and I used the world’s most popular game to communicate about the power of mathematics. Following the publication of my first popular science book, Soccermatics, I was invited to give a Google and a TedX talk, I wrote articles for leading newspapers and appeared on TV. In September this year I gave the prestigious public lecture in mathematics at Oxford. I found I was communicating to more and more people about the importance of applied mathematics.

I then wrote a second book: Outnumbered. I dissected the algorithms used by Cambridge Analytica to analyse our Facebook profiles. I talked to the data scientists at Spotify, Google, Microsoft and Facebook. I analysed the accuracy of predictions based on opinion polls. I found out how modern Artificial Intelligence works and where its limits lie. I challenged many of the poorly thought through ideas about fake news and filter bubbles.

As I did all this, another thing happened. Something I hadn’t quite anticipated. I realised that the reason I was frustrated before, wasn’t my failure to communicate. It was rather my failure to understand. I found that I was learning more than I was communicating. I learnt about the problems of algorithmic bias, the racism and sexism encoded in decisions made by statistical models. I learnt that many academics, who believe they are working on the cutting edge of theoretical physics and artificial intelligence, are in fact deeply out of touch with reality. I learnt that amateur bloggers often understand more about football than professionals, who are being paid hundreds of thousands of euros per year. I learnt that sometimes undergraduate students are doing better research than their professors.

Public understanding of science is not just about the public understanding of my science. It is also about me understanding what the public knows.

My frustration about my own situation has gone, but it has been replaced by something else.

I now realise that the only reason I could make the journey I have is because of my privileged position. My earlier success in ‘narrative’ research gave me the luxury of exploring an alternative. For work related to football, I received financing from ERC, Vinnova and SSF. But it has proved very difficult to fund the type of impact led research about society I would like to do. The public might be interested in an analysis of Cambridge Analytica, but unless I frame it within a certain academic narrative, there is no traditional research funding to do this work.

If it is difficult for me, who already has the track record research funders are looking for, then it is even more difficult for a younger researcher who wants to engage in the big issues of the day. There are so many young researchers, working on scare resources, but doing amazing work. Like Joy Buolamwini who has identified biases in the AI used at Google, Microsoft and Amazon. Or Julia Dressel who showed that one of the leading US criminal justice algorithms have the same accuracy as people paid $1 to predict reoffending. Journalists like John Burn-Murdoch at the Financial Times and Mona Chalabi at the Guardian who, on small budgets and on extreme timescales, create visualisations that show us the world more clearly. One of my students, Johanna Fyrvald, recently showed how leading AI researchers in industry and academia set aside legal considerations when they get sucked in to the technical details.

The perfect example of how successful impact driven research can be found right here in Sweden, in the form of Factfulness and Gapminder. Ola Rosling and Anna Rosling Rönnlund developed the App which used bubbles to change our perception of the world. The project was started over a family dinner with Ola’s dad, Hans. Perhaps the ultimate fun Friday project.

More than ten years later, do research agencies have a way of funding more Gapminder like startups? Er no, they don’t. Not really. It took Hans Rosling decades before he could communicate his insights to the world. Maybe it would be a smart idea to find the next Hans before he or she is 65? Before they only have a few years of their precious time left to give us.

If we value impact, not the academic narrative, we need to find the young researchers who are capable of changing society using data and science. Give these people the support they need and they will make the difference. It is as simple as that.

Originally published at https://www.tidningencurie.se on October 16, 2019.