Covering Coronavirus: Tips, best practices and programs

From idea to graphic — how the Washington Post visualized the spread of a virus

It has become the go-to graphic to understand the exponential growth of the coronavirus and how the growth curve can be flattened. Washington Post graphics reporter Harry Stevens found a visual means of telling complex story and made it go, well, viral. As Poynter’s Alex Mahadevan pointed out, former President Barack Obama posted it on his Twitter feed and President Nicolas Maduro of Venezuela showed the graphic on state television. We reached out to Stevens to explain the process behind this successful visual story.

How did you decide on this visual approach to tell the story of the potential spread of the coronavirus?

Stevens: A couple weeks ago, me and members of the graphics team were brainstorming ideas for how to move our coverage further. I showed them an experiment I did with collision detection last year, and we decided it would be an interesting avenue to explore showing network effects. I had to do some work to fix some bugs and to get it to run more efficiently, and of course to get the dots to transfer diseases when they collide, but I was able to use a lot of that same code.

Once I got that working and after seeing a lot of people share graphics explaining how we need to “flatten the curve,” it became clear that I’d need a way to show how the simulations changed over time – to extend the graphics into the fourth dimension. Once I got the simulation hooked up to the area charts showing change over time, I shared the prototype with the rest of the team, and we decided it was worth pushing it forward.

What was the collaboration like from the time the idea was conceived through when it was published?

Stevens: I shared several versions of the story with my editors and the rest of the graphics team over the week or two I was working on it. With their advice and guidance, I made several refinements to the graphics. For example, an earlier version I made used “scrollytelling,” where a single simulation would remain fixed in one place on the screen and would change as the user scrolled. The problem was the text explaining each simulation ended up sitting right on top of the simulation, and it was impossible for the reader to know what to look at. When I shared a scrollytelling version with the team, my coworker John Muyskens told me it didn’t really work and that I should seriously consider doing it another way. He was right. He was also the one who suggested making it so the dots recovered after a certain amount of time, and he was right about that, too. We also have regular open hours sessions on our graphics team, where you present your work to a panel of editors and they give feedback. I got a lot of tough and constructive criticism during those sessions, which ended up making the final graphics better and clearer.

What has the reader response been? (Paul Farhi mentioned in a tweet it was the most read piece on the WashPo website ever – true? and are more people continuing to find it?)

Stevens: This has clearly struck a chord with readers. It’s hard to build an intuitive understanding of how network effects work. There’s just something counterintuitive about exponential growth, even if you read about it or study a mathematical formula describing it. When you see it, it becomes very easy to understand. I personally did not have a good understanding of how network effects work until I got the simulations running on my computer screen.

I’ve received hundreds of messages which basically fall into four categories. The first, and most common, is just readers saying thanks. The story has helped show people that there’s a light at the end of the tunnel if we adapt our behavior, and so it’s given people hope in an anxious time, which is a reaction I did not anticipate but feel good about. The second is requests from readers to translate the article into their native language, which we’ve been working hard to accommodate. We already have Spanish and Italian versions published, and more are on the way. The third is technical questions about how I built the visualizations – which programming language, which libraries, and so on.

The fourth is from readers who criticize the article on the basis that the simulations do not accurately reflect the reality of covid-19. The article makes no attempt to simulate covid-19, but instead shows a fake “disease” called “simulitis.” There are many differences between covid-19 and simulitis. For instance, every dot that recovers from simulitis can no longer get sick again or transmit it. There are some articles I’ve read that suggest that is not true of covid-19, but I didn’t build that into simulitis because I wanted to keep the explanations as simple as possible. Another example is that no one dies from simulitis. I didn’t make it lethal, but I mention the lethality of covid-19 at the end of the article. Of course, there are many other factors – for example, human beings don’t transmit diseases by moving aimlessly around and bumping into each other. One thing I’ve tried to keep in mind is that these simple simulations are inherently limited – they could never forecast the outcome of a real disease.

Rather than modeling the spread of covid-19, the purpose of the simulations is just to show how network effects work. You don’t even have to think of it as a disease – it could be an idea or a meme on social media.

What do visual journalists need right now to do their best work throughout this coverage?

Stevens: The same as “non-visual” journalists! Keep asking questions and thinking hard about how we can best serve our readers.

How are you taking care of yourself?

Stevens: The Post encouraged employees to start saying home two weeks ago, and I have pretty much been in my apartment ever since. Instead of going to the gym, I’ve been going for runs, which I don’t like as much but I reckon is better for everyone. And my wife and I have been watching lots of movies and cooking more.