The Software That’s Powering All the Coronavirus Dashboards

(Bloomberg Businessweek) -- With roughly a zillion different sources out there for coronavirus information, experts and authorities are rapidly iterating on ways to best present useful information to the public. Are deaths spiking, or is the curve flattening? How is the virus moving around, and how can it be stymied? To collate all of this information, many groups are leaning on geographic information system (GIS) software. It’s often used by governments and large businesses that need to account for changes in the physical world, usually meaning disaster preparedness and mitigation, as well as monitoring public infrastructure, such as plotting the location of 911 calls.

Perhaps the most famous example is the Johns Hopkins Covid-19 Dashboard, a website assembled by the university’s Center for Systems Science and Engineering. In the upper-left corner, in blaring red type, is a running tally of confirmed cases worldwide (3,628,824 as of the afternoon of May 5). Below that, a morbid leaderboard of confirmed cases divided by country (the U.S. has a depressingly untouchable lead). Front and center in the dashboard is a heat map of cumulative cases across the globe. You can see which countries managed to contain the virus by the fact that they aren’t covered in bright red spheres (the U.S. is covered in them). Other modular bits of data—tables, key figures, graphs—are peppered throughout the interface.

Various state and local governments have launched similar GIS implementations. There are dashboards from the University of Virginia; New Jersey; Florida; Los Angeles County; Travis County, Texas; San Diego County; Fairfax County, Va.; the Canadian province of British Columbia; and the United Kingdom. All of these are built using the same software, ArcGIS, developed by California-based software company Esri. The privately owned company was founded in 1969 and sits unassumingly in that tier of comfortable, decades-old tech companies built on products that have survived radical technological changes. No matter what year it is, people always need to crunch numbers and make maps. In 2014, Esri took in $1.1 billion in revenue, and as of 2016, it had cornered more than half the market for GIS software, counting more than 350,000 clients, according to Forbes.

Ryan Lanclos, who leads the disaster response program at Esri, says more than 3,700 organizations are using these dashboards for Covid-related information. Some of them were familiar with GIS software, but many are new to it. Esri assembled prepackaged software and GIS implementations that clients could slot their own specific regional data into, rather than having to build a system from scratch. The John Hopkins site is getting more than 3 million hits per hour, according to Lanclos.

By this point, it’s become pretty easy to identify the ArcGIS dashboard. They’ve all got the same user-interface iconography and Avenir typeface. But while the building blocks are the same, each dashboard is assembled in a different way. The interface is modular, depending on what you want to show. The Johns Hopkins one shows death tallies and heat maps, while UVA’s is a timeline that lets the user move backward and forward and watch the virus move around over the course of a few months. Sometimes the death count is in menacing red font, sometimes the color scheme is more measured. Some charts are bar graphs, some are pie charts, some are line graphs—on a few occasions, the charts are broken (e.g., a pie chart depicting a 92:8 ratio that splits 50-50).

This happens more often than not when viewing the mobile versions of these dashboards, a design constraint that poses additional challenges on top of designing for a broad audience. Balancing that load, and importing and organizing data, is a task that each dashboard handles a bit differently. Some pull data and recalculate on regular intervals, while others might grab new data each time someone reloads the page, sort of like how Facebook’s News Feed reranks itself each time a user loads it up. On some dashboards, the line graphs have each data point or segment labeled, but the graph is so small that the labels bleed together, forcing the user to mouse over each pinpoint dot on the graph to find precise info. On mobile, where there is no mouse, this maneuver might be impossible.

In addition to the mobile dilemma, Esri has had to consider balancing computing resources. “We certainly saw extremely large spikes in our online SaaS platform usage, and we did some reallocation of cloud resources to quickly handle this demand,” Lanclos said. Part of that involved helping clients take advantage of edge caching (computing that happens closer to the client side of a network, rather than the server side, like when Google Maps saves maps on your device to save bandwidth and work faster). The situation has been a learning experience for Esri. “We’ve seen an increase in traffic on our websites and a huge spike in requests to our disaster response program,” Lanclos added.

“You don’t want to overload any dashboard with too much information,” Lanclos said. “To reduce the complexity of the dashboard is really critical.” He specifically praised the work of San Bernardino County, Calif. That need for simplicity is even more apparent on smaller screens. These sorts of dashboards usually live on giant monitors; now they’re being pinged from mobile devices. According to Lanclos, 60 percent of Covid dashboard traffic has come from mobile, where responsive web design is often easier said than done. The dashboards were meant for “environments and operation centers with big screens, and everybody’s coming together to make a big decision, and monitoring and planning,” Lanclos said. Now anyone can throw the dashboard up on screen and leave it running all day, keeping tabs on the situation with the latest statistics, as if they work in a socially distanced, slow-motion version of a disaster response center for the Federal Emergency Management Agency, or the fictional Counter-Terrorist Unit from 24.

The coronavirus dashboards are all the products of good intentions, making public-health data readily available in detail, statistics presented at not just the state or county level, but down to your ZIP code, too. They make it easy to know if things are improving or degrading, based on whether numbers go up or down. Their initial inaccessibility and learning curve makes a layperson feel like a sleuth, and that they’re helping, even if the only way to help is stay home. The challenge is to make sure users have as much information as possible without getting overwhelmed immediately. A good dashboard is helpful at a glance, but also lets users take a deep dive into the minutiae, if they want to. But a good user remembers not to get lost in a well of depressing statistics.
 
Read next: A Look at the Most Promising Candidates for a Covid-19 Vaccine

©2020 Bloomberg L.P.