Tasting Notes for Corona Virus Maps
I’ll be updating this from time to time as more maps filter in. Watch this space!
The COVID-19 virus is a big deal. As of this writing, over *20,000 people on the planet have died because of it. Thousands more will almost certainly die.
Yes, it is true that your individual chance of contracting and succumbing to this illness is very low. As has often been pointed out, 18,000 Americans die every year from the flu that humanity has gotten very accustomed to living with and treating with nothing more than soup and daytime television. Does that mean it’s not an issue? Does the fact that the flu kills fewer than half as many people per year in the United States as car accidents make you more comforted about the flu or more wary of cars? Assessing risk and figuring out how freaked out one can appropriately be is a thing that humans are notoriously bad at. Yes, the likelihood that you will die from the Corona virus is vanishingly small but the likelihood that you will know someone by the time this whole thing ends who has been killed seems almost inevitable. Obviously public panic is never an adaptive response to disaster but we do have a responsibility to inform the public about a real, new and very deadly threat without scaring the bejeezus out of people and contributing to stress which is responsible for at least 120,000 American deaths per year on its own.
So how do we get people to take this seriously without inciting a panic? Well, let’s see how cartographers have tried.
Alright, well, we gotta start somewhere and where better than Youtube? This map is put together with current-ish data from the CDC by Wawamu Stats which dedicates its channel mostly to animated graphs. Hopefully this isn’t their last cartographic project because I really think that viral content that is informative and digestible has great public education potential. Like it or not, more people get their news from Facebook than from more reputable forms of legacy print media. How is it doing?
Good: The map is pretty visceral. It’s definitely setting a mood and commands severity. Red and black do that. But just…
Bad: Ugh, the dreadful music. The non-linear scale. The little grammatical errors in the labeling. How the countries are sometimes labelled with non standard abbreviations and sometimes spelled out in inconsistently sized text. The crowded national flags tucked into the spaces between land forms. Even the red-to-black color ramp reeks of bad default settings. The multiple statistic tickers. I can’t. If I were a professor, I wouldn’t dignify this entry with a grade. We’re moving on.
British tabloids are well-known for level-headed analysis and awful graphics that can be found later on r/mapgore.
It looks even worse now (March 15).
Good: It gets your attention, I guess?
Bad: It generates panic which is counterproductive and deeply irresponsible. The label for Rhode Island says “DC,” Nebraska isn’t highlighted whereas Maryland is. If you’re going to cover up Hawaii, don’t bother giving it its own detail. This was clearly done in a hurry to maximize clicks and toilet paper hoarding. Lastly, really dig the “3 FROM CHINA” note on the Diamond Princess cruise ship label. I’ll grant them plausible deniability but it’s not the most self aware commentary if you don’t know how the virus got on board the ship and there’s already so much anti-Asian racist sentiment out there. In such circumstances, the media have an even greater responsibility to promote civil order. This map is awful.
Then there are the professionals. This is a Mapbox world map created by Health Map, a group of doctors and developers at the Boston Children’s Hospital who have been visualizing global health issues since 2006. One of the nice things about their format is that you can compare pandemics very easily so you can tab between Zika and Ebola and you can see their relative spread. It’s very trustworthy, reliable and easy to use for the public and professionals alike. It really is the best of the best and the animation shows the growth of the pandemic pretty well even if at the end it looks less like a Jackson Pollack and more like someone just poured all the paint out.
Good: There’s impressive detail here and a great summary table in the upper right that shows the total numbers of confirmed cases as well as the all-important time of last update which sometimes gets left out of current events maps like these. The neutral black background and yellow/red markers show a dramatic bad thing happening and it’s a bit ominous without being over-the-top. Newer map edits have a neon color scheme that feels a little more value-neutral.
Bad: Clicking each circle is pretty hard when they overlap and your cursor turns from an arrow into a blunt grabby hand which makes it worse. Because the circles stack on top of each other, a small recorded incident could easily be covered by a larger incident that came afterwards nearby so this might be one of those rare situations where a heat map might make sense, or even have each circle fade away after a couple of seconds to show better what comes afterwards. It’s possible there was a particularly bad week that happened that wouldn’t be seen because it was covered up so it’s also not great at showing what is happening at a particular time in the way a simple line graph might show a sudden uptick in the slope. Come to think of it, a line graph would be better at showing cumulative increases too. What is this map really trying to communicate?
I didn’t realize people were still using Google My Maps (formerly Fusion Tables) but Dutch news service BNO is. My Maps reminds us that it’s still a handy little web map function that may not be as pretty as Carto, Mapbox or any number of other similar plug and play services but it’s getting the job done.
Good: My Maps as a platform is cheap and simple and the map kind of shows it. But the sidebar has a ton of information about every single case. It’s mesmerizing.
Bad: Stacked icons means you’ll never find the incident you’re looking for or accurately discern how many cases were identified in a particular place. Even the marker symbols between victims who are currently sick vs the ones who succumbed are far too easy to mix up.
The news services are still garbage.
Good: OK, I’m actually on board with fuzzy heat maps because they show intensity as well as extent which is important. Density matters in this instance. I’d like to see the author trim the excess away from the coasts, perhaps by adding a screen or stencil that clips out the water areas so there isn’t a little toxic cloud floating over the water. There are no cases in the ocean to my knowledge.
Bad: Red-Green colorblindness affects the ability to perceive red and green so a red and green color scale is basically the worst scale you can use especially when it doesn’t also change brightness or some other quality to the symbol. The studies aren’t great for the rest of the world but of those with Northern European heritage, 8% of those with Y chromosomes and .5% of those with XX chromosomes won’t be able to read your map and see what an awful shade of red and green that were chosen here. Also, make a key you animals. Is green the absence of positive tests? Because in a place with insufficient testing like the US and UK, an absence of positives is very different than the confirmation of a number of negatives. I could see this scale being used to compare test results as a ratio (twice as many positive tests than negative produces red, for example) as long as the color scheme were adjusted.
Whatever I say about My Maps I temper my aesthetic preferences with optimism because this free tool lets anyone share important information, like how Universities across the US have been affected.
Good: The people with the data can make the maps. This is the promise that webmaps made to us.
Bad: Look, it ain’t pretty but I’m just glad that someone was able to make this. Your software or cartographic expertise could be the finest in the land but if the people with the ideas and the data can’t access it, it’s all for naught.
I found this map on the ABC Orlando news site with a caption that says ‘courtesy of ESRI’ which leads me to believe ESRI built this map and the accompanying dashboard and then gave everyone the html code to embed in their news pages. If true, this is an amazing strategy to spread the brand and the strategy definitely tracks with a growing desire for brand recognition. I feel like more and more ESRI is actually being used as an identifiable brand on the maps themselves so all those t-shirts at ESRIUC must be paying dividends.
Good: The popups aren’t super pretty but they have total numbers of confirmed cases, deceased, recovered and existing cases. This is useful to label as the pandemic rages on and the likelihood increases that we forget what metrics we’re actually keeping track of. There’s also an effective mobile version of this map that works great on a smaller screen.
Bad: who labels continents? Really. I’m also mildly annoyed that national and state data values are placed on centroids which works great for regular shaped regions but tends to look unbalanced for those that are not contiguous like New York and Michigan.
Lest you all think I’m an ESRI shill, here’s one from the R Stats nerds.
Good: There’s a slider bar which is kind of slow but it works and it’s cool. The markers are translucent and the color and size indicate the same thing so it’s accessible for people with color perception issues.
Bad: It’s kind of bland. To be honest, the rest of the data visualizations in the dashboard are far more impressive than the map.
NBC New York
This map uses off-the-shelf data visualizer program Datawrapper and dang is it pretty! It probably didn’t take too much effort to make this which is even more impressive (not as a cartographer but as a developer).
Good: For reasons that I’m iffy on the previous map, I’m a real big fan of this one which has smooth, effortless looking animation that shows not individual incidents but total counts by nation or or state/province. Watching the marker grow over Hubei Province in China makes it very easy to quickly compare the number of cases as well as the dramatic growth rate in February.
Impressively, this map works great as a standalone interactive as well as an animation. It’s great design without getting bogged down with too much information. I think I’d like to see a component of this graphic compare cases to fatalities but since the lethality of the virus seems fairly consistent, there may not be as dramatic a story there to be told.
Bad: I don’t know, are we still super into making everything “dark mode”? Yes? Ok, make this one dark mode and it’ll be perfect.
San Francisco Chronicle
I’m a biased Californian and they had such a good PG&E outage map last fall so I thought I’d check in to the Chronicle again. And I’m glad I did. Now that the Covid-19 cases are stacking up in the Bay Area and testing is finally ramping up, there’s a lot more actionable and vivid data right here in my neighborhood.
It is behind a paywall but if you have a subscription, go ahead and take a look here.
Good: Calming but authoritative color palette. A big, readable dashboard with the number of cases. Smooth animation showing the number of cases in the Bay Area counties back to February. The circles get bigger. It’s a simple and easy to understand concept. The accompanying graphs are probably even more effective at showing the tipping point when exponential growth is really visible.
Bad: While each county has a label with the population, each circle on the map still represents only the raw total number of patients. Marin County has a third the population of San Francisco and it should make the audience raise an eyebrow if the two counties have the same number of patients. They don’t of course, but I think it would make an interesting story to see if recent Shelter-In-Place policies, which were not uniformly applied across the region until the statewide order, have created different case densities in different regions. We’re getting dangerously close to population map here.
The New York Times
Say what you will about the failing New York Times, their infographics department is always on point.
Good: The main reason I like this map is that it has at its core a basic call to action. The major afflicted countries are labelled clearly. There is an approachable, readability here that speaks to a lay audience. There is a gradient of risk with a red palette that commands attention but there is no attempt to visualize the total number of cases in each country, only the CDC risk level associated with each area and an actionable recommendation for that risk level. Maybe the number of patients is less important than projections of severity informed by available hospital infrastructure and population density. For a purely hypothetical and extreme example, perhaps we imagined that there were 1,000 cases in Mongolia and I was very wary of going to visit but if I later learned that they were all in the capital Ulan Bator and my travel plans took me to a city a 1,500 miles away, I might recalibrate my perception of risk. This NYT map is an excellent case of making the map to answer the important question, not merely the obvious question: should I panic?
Bad: I can’t quite discern the distinction between the color for “confirmed cases” and “travel, but be careful” and I don’t think both of those colors are on this map so I can’t compare them.
NYT Part II
The Times is also keeping track of cases on a national scale, using county level detail to show new cases as they develop.
Good: The map isn’t overwhelming to the eye. There is one type of scaled symbol with a transparency so you can see density of cases unlike some we’ve looked at above. This map effectively answers the question: where are the most cases right now? and doesn’t really elaborate on that. I’m very comfortable with this and honestly, I wish more cartographers would set more reasonable goals for themselves with regard to messaging.
Bad: I prefer consistent state abbreviations but I’m really reaching here. This is a great map.
University of Washington
This another dashboard that has popped up and one of the few using Carto to its full potential beyond the simple tickers.
Good: The real innovation here is the side bar that shows the rates of growth for selected regions. I live in California and I’m still eager to see the growth rate taper. This feels like more actionable data because it seeks to confirm whether regional or local measures like Shelter In Place orders are effective at bending the growth curve down. The graph I see today means that we’re still in for a rough ride. Regardless, the color scheme is softer and less dire even while it conveys tons of great data in an authoritative and serious manner.
Bad: I’m not a huge fan of the basemap selector radio buttons in the upper left. Changing the background doesn’t really change much or add to the map and no one is going to use the satellite imagery. There’s this weird thing that the authors did where they chopped the US up into these arbitrary looking sub regions to aggregate local and county data. There’s no reason that anyone needs to zoom in all the way in to see where precisely the individual positive cases were identified so I see why they thought it’d be a good idea to make mid-scale visualization to visualize problem regions. This may have been done to break down the election map problem we’re seeing now where the map is mostly empty save for major population centers. Are we looking at a population map or a map of positive cases? Ultimately this data chunking solution (which uses a weird marker symbol) is like the electoral college in that it makes rural areas more noticeable but at the cost of seeing where the biggest, most acute problems are.
Center for Systems Science and Engineering, Johns Hopkins
This is the map that haunts my dreams. It’s a full dashboard of terror. It has everything you need right down to the county level. There are graphs, numbers, rankings and it’s all updated on a near real-time basis. It’s fantastic and addictive and awful.
Good: There’s a ton of data. It’s aggregated down to county (this seems to be a fairly recent innovation) and it has alot of big descriptive statistics up top to distill the big picture. It’s clear, effective and emphasizes severity.
Bad: Of course, emphasis on severity seems to be bacfiring right now. There’s a not insignificant part of the population that thinks Covid-19 is a conspiracy, a media hoax or just meme hype and they’re probably not allayed by a dashboard that drips with so much morbid intensity that the text might as well be in Chiller. I have minor squabbles with the leaderboard that lists national cumulative counts but disaggregates the US into individual states and even counties makes the total US count look shockingly low while the map itself is splattered in red. Is it really an effective comparison to point out that New York City has as many cases as South Korea? Surly that is as misleading as comparing the entire US to South Korea as well?
Finally, information addiction is a phenomenon in 2020 and I certainly have it. This dashboard has lots of great data that is truly clickable, explorable and anxiety inducing but what does this map actually want you to do? I refresh this map several times a day now and I can’t tell you if it’s actually making me change my behavior for the better or just making me freak out. Between the ominous red/black color scheme and the austere bold fonts, this kind of engagement is ominous and paralyzing rather than motivating. What do I do with this information except refresh and refresh and refresh and do math in my head and try to draw conclusions about my safety and how long I’ll be staying at home and what will happen to my job… when I’m not an epidemiologist. I rely on concise answers from experts so I don’t start wildly speculating about a subject I have no expertise in (a data scientist has been roundly criticized for his amateurish ‘debunking’ of what he calls corona hysteria and far too many think his opinion matters at all). How do we amplify the voices of scientists and disempower amateurs with inflated opinions of their wikipedia degrees? Maybe providing this amount of information aimed at a general public audience isn’t the best thing to do right now.
One thing I noticed about all of these maps is that they utilized recognizable global projections that distort the area around the poles and make land masses at high latitudes look bigger than they really are *coughMercatorcough*. That’s a weird design choice and it was probably not a choice that the cartographer consciously made because these projections are often the defaults of these kinds of programs and there’s a strong desire to show people a map that speaks their visual language. We all think we know what the earth looks like and sometimes it takes a moment or two to calibrate when we encounter a world map that doesn’t quite fit our expectations but what we’re missing here is a sense of density. Some places have way more people and maybe these maps should reflect the fact that China is not a petri dish of disease so much as a place with 1/6 of the human population. This is a variation of the old cartographic trope “land doesn’t vote” which often gets bandied about in discussions of polling and election maps. Should these numbers show relative infection rates instead of total numbers? Maybe. That might be a better indication of transmissability risk and provide more meaningful information to audiences who are trying to figure out what they should do.
A good pandemic map should set an appropriate tone. It should be factually accurate but it also needs to portray a sense of severity without inducing panic. Displaying land area and population centers on an equal area global projection might take that into account better.