Why is the NYT spiral graph getting so much attention?

Let’s start by explaining what “the NYT spiral graph” is, since that’s not a very specific explanation for one particular chart. But the fact is, if you’re part of the data visualisation community, unless you’ve been hiding under a rock for the last few days, you’ll almost certainly have seen this chart more than once. The NYT (New York Times) spiral graph, shown below, is a graph showing new COVID cases throughout the United States in spiral form from the start of 2020 to the time of the graph’s publication in early 2022.

The NYT Spiral Graph

It first appeared in an article on the New York Times Opinion section by scientist Jeffrey Shaman on Thursday (January 6th). It gained a lot of attention on Friday 7th, and by Saturday January 8th, there had been a huge amount of reaction, conversation, opinion, and counter-opinion. By Sunday January 9th, the number of tweets, threads, blogs, and videos discussing the chart had probably already hit saturation point. By the time I publish this, on Monday January 10th, it’s most likely that people will already consider this “old news”. I can confidently suspect this will be the most commented on chart of the year, and I say that just ten days in!

So why I am I adding to the huge amount of what’s already out there on this chart? I really want to give my own perspective on this, because this chart is perfect for Questions in Dataviz – not just this blog, but for the upcoming book. I’m delighted to report that I handed in the first (very) draft manuscript on January 4th, less than one week ago, however if I hadn’t already done that, there’s no doubting we could possibly have extended to a whole additional chapter or more on this graphic alone! 

I don’t want to highlight every contributor or comment (that would be impossible), but rather summarise the reaction as it unfolded in my own circle. First of all, initial reaction was very negative. Like many, I first came across the graph attached to a tweet from twitter user Zach Freed with the comment that there was “literally no reason to make this graph into a spiral”. Other negative reactions included:

  • Timelines should always be linear
  • Anything circular or radial is bad practice
  • It’s hard to see any correlation or seasonality using a spiral
  • Dislike of the colour and imagery – it reminded people of worms/tapeworm
  • Dislike of the offset of the spiral
  • Why are values shown both above and below the line?
  • Tufte would hate it (!)

Incidentally, I‘m sure Tufte will have seen this by now, and I have no idea whether he hates it or not (I’m not particularly interested whether he does, other than the fact it would be an interesting addition to this already fascinating debate). Many discussions or arguments around the quality or validity of a visualisation can often involve quoting the teachings of Tufte. Indeed, I even find room to do so myself in the Questions in Dataviz book! And so, for any of the reasons above, many commenters have used the word “bad” when they describe this chart.

But then came many alternative views on the chart. Many people have enjoyed the reaction and discussion around the chart, and some of the positive reactions have included the following:

  • It’s impactful and attention-grabbing
  • It’s an improvement on separate yearly line charts as there’s no disconnect between years
  • It has sparked much conversation, not just on dataviz practices, but on the underlying figures themselves,
  • Timelines in spirals are effective use of screen real estate
  • Using a spiral makes it easy to see seasonality and correlation
  • Spiralling is a good metaphor for increase in case numbers
  • Tufte would hate it (!)

One thing to note is some of the contradictions. You might have seen that different people find it easy and difficult to make year on year comparisons and observe seasonality. We can probably put this down to unfamiliarity and, perhaps, people who have already made their minds up on the validity of the chart based on their ideas or visceral reactions to it. And, of course, in the eyes of different data visualisation proponents, Tufte hating a chart can be seen as an equally good or bad thing, depending on whether people liked (a) the chart, and/or (b) Mr. Tufte!

It’s important to note that some commentators have made a number of comparisons with other well-known visualisations but I think they might be missing the point. First, some have used the subject of disesase data depicted in a radial manner to compare this to Florence Nightingale’s well-known Coxcomb charts (a version is shown below). But Nightingale’s charts (also discussed in my book!) use discrete months rather than continuous date for their radial segments – and, perhaps more to the point, create a separate coxcomb for each new year.

Florence Nightingale’s Coxcombs

Comparisons are also made with WEB du Bois’ spirals (as shown in my previous post here) but in fact his spirals do not represent any measurement over time, rather they are just impactful ways of highlighting large numbers in smaller spaces. And finally, some have compared to Ed Hawkins’ well known climate spiral. This does do a great job of showing seasonal data year on year, but this animated graphic does not “spiral” with measurements shown further from the centre as time increases, so it’s by no means an exact comparison.

A more relevant comparison would be my Donald Trump spiralling tweets visualisation which does use a genuine spiral for visualising a timeline (an Archimedes one at that, for the spiral purists!). Of course, my levels of notoriety mean that this has (rightly) not been brought up at any point in discussions. But it is at least a more comparable example of a spiral being used as a timeline to draw attention using unconventional chart methods.

It’s only fair to mention that on all sides of the debate there have been suggestions for improving the chart. First of all, the spiral has been found to be somewhat irregular, adding to feelings of uncertainty/unease about the spiral being off-centre (thank you to Robert Kosara’s eagereyes.com blog for detail here). And the below version from Amelia Wattenberger makes a number of cosmetic improvements (making use of the side by side image created by Amanda Makulec in her own excellent critique)

Original NYT spiral chart (left) and Amelia Wattenberger’s remake (right)

Moving away from the pale red that reminded more sensitive viewers of worms or intestines, the intensities of blue and green do a great job of adding colour encoding case numbers, allowing us to more easily spot seasonal differences. In addition, the spiral is more compact, even and centralised, and the mark overlaps are a great way of highlighting outlying large values (you will know that I am a fan of that effect from my previous post on overlapping marks)

So what’s my opinion? It’s probably no surprise that I love it. And I probably love Amelia Wattenberger’s remake a little bit more. But that’s not important Overall, here’s why I think the NYT Opinion spiral graph can be wholly justified. The chart has usually been lifted and placed into a tweet or blog post (I’ve done it myself in this very blog post!), but it’s important to give the full context which is to remind readers that the chart forms part of an opinion piece, a full article which in its entirety gives the full context and explanation behind it. Furthermore, the article includes other graphics such as the more conventional line chart of cases over time which is exactly the chart that many of the spiral’s detractors are clamouring for. 

USA COVID Cases in line chart form

There’s little doubt in my mind that the attention garnered from the chart and the conversation and discussion generated around it mean that it’s had far more impact than the line charts on their own would have had. We’ve seen many versions and variants of line charts, but nothing quite like this until now. As a result, many folks like me have paid more attention to US COVID figures than ever before, and have read the full accompanying article and opinion piece. Job done! 

Throughout Questions in Dataviz I encourage you, the reader, to understand conventional opinions and “rules” in data visualisation, to challenge them, and to consider alternatives, taking inspiration from others or devising your own versions. Are there alternative ways of showing timelines, for example? Consider your audience, and whether the positives from alternative or unconventional chart types will counterbalance any perceived negatives. In this case, the audience is the general public and not policy makers or decision makers who need instant precision to make a data driven decision (and, perhaps equally relevantly, not data visualisation purists or academics who will be quick to criticise!)

In examples like this I’m always OK if you take a position on either side of a particular debate – it’s absolutely OK to dislike the spiral for any number of reasons. And of course, the third “it depends” option is usually very valid too for most data visualisation arguments! But Questions in Dataviz will also encourage you to see both sides of a debate and decide for yourself which angle you prefer to take given the design choices you want to make, the visual impact you are trying to achieve and the needs and expectations of your audience. In my case, this usually leads to a fun, impactful if experimental original piece that purists may baulk at but that I as a designer am usually proud of. This is what I encourage, and this is exactly what I think the designers of the NYT spiral have achieved brilliantly.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s