Freelance science journalist Dyani Lewis won the 2021 Finkel Foundation Eureka Prize for Longform Science Journalism for her piece about mathematical disease modelling for Cosmos magazine.

Where did the idea come from for this piece?
Cosmos editors Gail McCallum and Ian Connellan came to me and said they’d really like a piece about mathematical modelling and its role in the pandemic.
Right from the beginning, I wanted to make sure that it was a narrative piece that showed the impact that the pandemic had had on the researchers doing this work, because their lives have been suddenly upended. So I made sure that I asked questions of interviewees, such as disease modeller James McCaw, about, ‘what has this meant for you, have you had a weekend off since this began, when do you have your zoom meetings and phone calls?’
Investigating the historical aspects of the story were also really interesting; the way that people had decided that at one point investigating diseases wasn’t a real scientific field. And someone said, ‘we’ve got to bolster this with something to make it seem a little bit more legitimate; if we could add some maths to it, it’ll have that legitimacy’. I also wanted to talk about the limitations of modelling, because it’s not some magic crystal ball that predicts the future. You’ve got to be mindful of the fact that it’s only as good as the numbers that you put into the model. If those assumptions are incorrect, then you’re not going to get a good reflection of what’s happening in the world.

Where did you start with this story?
First, I had a look at what people would be good to speak to, who were the people who were doing the modelling. Because this was a feature written for an Australian publication, I thought it was quite important to have Australian scientists represented. So very early on, I contacted the Doherty Institute, because they’ve been doing the modelling for the government.
I also went to BioRxiv and MedRxiv – the preprint servers – to see what modelling papers have come out. That’s how I found George Milne at the University of Western Australia. Mikhail Prokopenko at the University of Sydney had also published some work on preprint servers, and they were using these really complex, individual-based models.
Scopus is where I do most of my research to find published papers. I trace back and find where an idea came from? Where were the first people that published on it? Who had the first eureka moment? That’s how I came to Tim Germann, who is a researcher at the Los Alamos National Laboratory in the US. He’d developed these complex models looking at how individual atoms of metal crash together in metals in a car crash. He thought it was a fantastic model, and he was fishing around for people to bring him problems that he could solve with it. It just so happened that there was an epidemiologist who was visiting the area, and they got together, and they modelled a flu outbreak in the US.
It was the first model that, instead of giving a really rough idea of how an epidemic plays out, it was looking at hundreds of millions of people all at once. Germann wasn’t a disease modeller himself, so it was important for me to put in context how someone from a background outside of disease modelling can have a really big impact on that field.
When I was speaking to people, I was really looking for those narrative moments. One of the first questions I asked all of the disease modellers that I interviewed was, where were you when you first heard about this novel coronavirus?
As soon as I spoke to James McCaw, I knew I was going to lead with his story. He and the Doherty team, they were very much going to be doing the modelling for the government very early on, so they didn’t have the luxury of just sitting back and waiting. They immediately had to say ‘what data do we use?’, they’re contacting their colleagues overseas. James said that he had been on these late-night teleconferences at least once a week since the beginning of the pandemic.
So any opportunity that there is to make it more of a story for the readers – to introduce them to a character in a way that they’re not just a name, but they are a person who played a role – then you try to add those in as much as possible.

How was it finding and getting hold of interviewees?
Because I was writing for an Australian audience, I did want it to be more of an Australian story. Which is not to say that Australians are the only ones coming up with all the answers by any means. But again, the Australian researchers were modelling for an Australian context, and that was really important.
It was great to be able to speak directly to someone from the Doherty Institute. It would have made the world of difference to the story if I hadn’t had that access. At that point in time, when I spoke to James, they were just about to release their modelling to show when the peak of the first wave was going to be for us. I think I was very lucky as that access was fabulous.
I didn’t have a huge amount of difficulty getting interviewees. I don’t know that that has always been the case throughout the pandemic, because some of these scientists, they literally don’t have enough time to do their own work, let alone speak to journalists as well.
I always find it rewarding speaking to scientists, because they usually are so giving of their time. Raina McIntyre (Kirby Institute’s Biosecurity Research Program at the University of New South Wales) has been so generous with her time throughout this process. I spoke to Mikhail Prokopenko a couple of times just to clarify things because I had a recording SNAFU.

How do you deal with reading what are often incredibly complex scientific papers?
I often use papers to identify people to interview and then I hand over to them to explain what it is all about.
But when I read papers, I always look for the key take home messages: why what they’ve done is special, and what does it do that previous work didn’t do? What’s the novelty that they are showing? Why is it a big deal?
The second thing that I look for often is I’ll scan through the discussion and conclusions for what it doesn’t tell you, because you want to know why it’s still a little bit half-baked, or why we need to do more work. That’s really important contextual information to give the reader about: does it solve a problem, what sort of problem is it solving, and why is it not the perfect solution?

What’s your planning, structuring and writing process?
I usually go through my transcripts, and I write a summary of the transcript, and I pull out quotes. Anything interesting, I dump it into my Scrivener file so I’ve got all of these potentially useful bits already out of the transcript. Sometimes I put them straight into a Word document, and if I can add a sentence of context or how I think it will go in, then I do that. There were also some key things that I knew that I needed to explain, like how the models worked.
The first bit of it is an anecdote about how James McCaw’s life has been personally impacted, they’re rushing to get these answers for the government, and then the government makes all these decisions that up-end everyone’s life.
And then you get to the nutgraf, which is ‘where did this field of science come from?’ And then it just – I hope – followed this fairly logical path.
It was a reasonably simple structure. It didn’t create so many problems, except that I had all of these different ideas of things to put in. But you’ve got to say, ‘yeah, it would be a cool idea, a nifty little aside’. But if it’s just going to detract from the piece, then you’ve got to kill your darlings, whether it’s yours or your editors.

What did you have to leave on the cutting room floor?
Early on in the pandemic it was all about lives or livelihoods, and I knew that there’d been some interesting economic modelling showing that actually dealing with the virus is the best way to save the economy. I wanted to put that in, but fitting it into the main thread of the piece just didn’t really work. So I put that out in a separate box.

What were the unique challenges with this subject area?
When I was writing this, I had no idea how it would hold up from one week to the next. It was a challenge, because so much COVID reporting is getting the latest information out as soon as you can. The prospect of writing an article for a magazine that’s only going to be published in a couple of months’ time and is going to be read in three months’ time – that’s why I knew it had to have those narrative elements to carry it.
I think discussing the way that modelling isn’t just, ‘when’s the peak of the pandemic, how many people are going to die’ but also, ‘how do we make decisions about what restrictions can be lifted first, or need to be held on for a bit longer’ – I guess that’s one thing that maybe has given it more longevity than it otherwise would have had.

Was there anything that you learned from doing this story?
The same thing I learn every single time – and don’t take to the next piece – is that I need to start writing notes far earlier than I do. I’ll go through and do a whole lot of research and go ‘oh, that’s really interesting’. And then I’ll sit down and I’ll go, ‘why didn’t I already write that down?’

Dyani Lewis spoke to Bianca Nogrady.