Advertisement

Covid-19 Has Made Us All Dashboard Junkies

(Bloomberg Opinion) -- As Covid-19 picked up speed and ferocity, digital dashboards were everywhere.

Published by global health organizations, national and local governments, medical centers and media outlets, pandemic dashboards visualized an evolving narrative of life and death — from case numbers and mortality rates to testing capacity and ventilator access.

Very soon, dashboards were deployed to illustrate all manner of non-medical consequences — whether the collapse in air traffic, the rise in unemployment, the sources of response funding or the rules governing restaurants and bars.

In June, NASA co-created the “Earth Observation Dashboard” which allows “user-friendly tracking” of Covid’s planetary impact on “air and water quality, climate change, economic activity, and agriculture.”

Three interconnecting factors catalyzed this Covid dashboard dash:

the scale, speed and severity of the pandemic, which demanded the urgent assessment of complex streams of interconnected data; the widespread availability of dashboard generating software — including Tableau, Domo, Datawrapper and ArcGIS; the omnipresence of smartphones, tablets and computers on which dashboards could be (compulsively) viewed, compared and shared.

* * *

Before we sail any deeper into the ever-expanding dashboard universe, we need a definition.

At its simplest, a dashboard is any interface that visualizes one or more sources of data. That said, modern digital dashboards will share some or all of these characteristics:

Dashboards visualize[multiple] [modular] [real time] [critical] [customizable]data.

Furthermore, to be functional rather than just ornamental, dashboards require an interactive link to the real world, where data informs action, which in turn transforms data — and so on.

The most familiar model is a car, where the speedometer, tachometer and odometer are in a continuous and immediate feedback loop with the accelerator, clutch and brake(1). Similar loops exist in all mission-critical dashboards — from military drones to scuba gear:

If driving even a Ford Model T required a dashboard of dials, how much more vital are digital dashes to running not only modern corporations and manufacturing plants, but also their critical underpinnings: power grids, water supplies, global logistics, cloud computing, air traffic control and so on.

Of course, dashboards need not be instantaneous to be useful. Both the economic levers of government and one’s personal fitness goals must be assessed over weeks and months, and their visualization is no less valuable for this lag — especially when future scenarios can be explored using historical data and algorithmic analysis.

* * *

Although it took a pandemic to make dashboards a topic of daily conversation, “dashboard thinking” has been around for years — having escaped the confines of complex mechanical systems to penetrate corporate management and mainstream consumerism.

For business professionals, dashboards now visualize corporate “key performance indicators” of every conceivable hue:

And for consumers, dashboards track an ever-growing catalog of human endeavor — pregnancy, fitness, finance, medication, mental health, golf, rock climbing, home automation, meditation and prayer, to name but a few.

Apple’s latest iPhone OS transforms the home-screen from a static grid of apps to a dashboard of adaptive widgets — allowing users to gauge at a glance variables such as stock prices, steps walked and sleep patterns.

Indeed, so glued are we to these dashboard devices, we even have dashboards to track and regulate our screen time.

If you doubt the dashboard’s inexorable rise, Google Trends generates dashboards to plot the popularity of specific search terms — like “dashboard”:

And should you need further convincing, consider the dashboard of the Litter-Robot 3 Connect — “the highest rated, WiFi-enabled, automatic, self-cleaning litter box for cats”:

* * *

Like all data displays, dashboards are built on the “pillars” of data visualization:

comparison — how sources stack up composition — how parts relate to the whole distribution — how data is grouped and located interaction — how processes flow relationship — how data sources connect trend — how data change over time

By adding to these foundations immediacy, interactivity, flexibility and unified focus, dashboards offer a jailbreak from the static grids of Excel; a reprieve from the sequential death-marches of PowerPoint; and the freedom to take command and control of complex and critical systems.

Such tantalizing potential explains why internet-age politicians sporadically discover “dashboard government” — a hi-tech form of “joined up government” that channels the precocious optimism of Thomasina in Tom Stoppard’s “Arcadia”:

“If you could stop every atom in its position and direction, and if your mind could comprehend all the actions thus suspended, then if you were really, really good at algebra you could write the formula for all the future.”

In Britain, this optimism was epitomized by David Cameron’s much hyped “Number 10 Dashboard” — which proved shorter lived than his Angry Birds obsession. It’s unclear if the “NASA-style mission control” just launched by Boris Johnson will fare any better: the British Covid dashboard “missed” 16,000 cases, because of a spreadsheet file-size error.

In reality, the closest most politicians get to genuine “dashboard government” comes in times of crisis — when they decamp from traditional bureaucratic spaces (Oval Offices and Cabinet Rooms) to secure and smart “sit-rooms” where key figures interact with key data in real-time.

Although such locations existed before computerization — witness Churchill’s subterranean Cabinet War Rooms — as politics and technology become ever more entwined, “sit-room government” may become the norm, and running a country (or indeed a corporation) from a coffin-shaped(2) conference table will seem as anachronistic as navigating by the stars.

* * *

Our increasing reliance on dashboards is bound to stress-test the assumptions that underlie them. The most significant of these is whether we can truly avoid “Garbage In Garbage Out,” since nothing can lipstick a dashboard pig if its data are incomplete, incomparable, outdated or plain wrong.

During Covid, doubts have been raised both about the accuracy of specific online dashboards, and the neutrality of the numbers coming out of China, Russia, North Korea, Iran and even the United States.

But data does not have to be manipulated (or mishandled) to be misleading.

Donald Trump’s recent Axios interview with Jonathan Swan contained this revealing exchange about America’s Covid deaths:

Trump: Take a look at some of these charts.

Swan: I’d love to.

[ … ]

Trump: Here is one. Well, right here, United States is lowest in numerous categories. We’re lower than the world.

Swan: Lower than the world?

Trump: We’re lower than Europe.

Swan: What does that mean? In what? In what?

Trump: Look. Take a look. Right there. Here is case death.

Swan: Oh, you’re doing death as a proportion of cases. I’m talking about death as a proportion of population. That’s where the U.S. is really bad, much worse than South Korea, Germany, et cetera.

Trump: You can’t do that.

Swan: Why can’t I do that?

Trump: You have to go by where… look. Here is the United States. You have to go by the cases. The cases are there.

In addition to channeling “This is Spinal Tap,” Trump’s “you can’t do that” illustrates another key weakness of data viz in general and dashboards specifically: Selection is everything.

And, even without partisan meddling, selection is never neutral.

Dashboards privilege certain data types · Dashboards favor: immediate figures (daily Covid cases) over lagging indicators (monthly unemployment); “hard” quantitative numbers (Covid deaths) over “soft” qualitative words (lockdown’s impact on mental health); and open-source public data in universal formats over siloed private data in proprietary structures.

Dashboards reflect status quo thinking · Covid’s clarion call — “we’re all in this together” — quickly rang false when the virus began disproportionally killing the poor, the aged, the infirm, the incarcerated, essential workers and those from very specific racial and ethnic groups. Not only are these vital socio-economic and demographic figures still absent from most Covid dashboards, the New York Times had to sue the Centers for Disease Control to obtain the relevant U.S. figures on race and ethnicity.

Dashboards become walled gardens · The relative complexity of constructing a dashboard (compared to a slide deck) means that they tend to resist new data sources and interpretations; at the same time, users become accustomed to dashboard layouts and, in general, dislike redesigns.

Dashboards elide caution · Although designed to visualize complexity, for reasons of space and clarity dashboards often omit the explanatory and cautionary footnotes that bespeckle technical reports. One example here may be the wildly popular stock-trading app Robinhood which, by visually simplifying and even gamifying investing, may smooth over the traditional caveats of asset allocation and risk tolerance.

At this point in any debate on new technology, it’s obligatory to quote William Blake or Winston Churchill or Marshall McLuhan:

“We become what we behold. We shape our tools and then our tools shape us.”

If the origin of this line is disputed, the sentiment is not. Even if their data are unimpeachable, dashboards pose a range of epistemological risks:

The illusion of control · The average citizen poring over Covid dashboards in March and April had no idea if the visualizations were accurate, current or comparable; no power to influence domestic or international policy; and no ability to escape to a supposedly safer country. Alarmingly, many politicians were in the same boat. For all the Covid data available, the global response has been haphazard, inconsistent and in places disastrous, suggesting that the mass visualization of inputs does not necessarily lead to better outputs.

W.Y.S.I.W.Y.G. · The display of data inevitably influences its perception. With Covid, we talk of “waves,” “spikes,” and “flattening the curve” because of Hokusai-esque “fever charts”; and “hotspots” because of the ominous red circles engulfing badly hit areas. Presumably we would deploy different metaphors — and policies — if Covid dashboards showed, say, crime-scene photographs of the victims of domestic violence during quarantine.

Design · As dashboards become more pervasive and powerful, so every facet of their design becomes more critical — from the layout of touchscreens to avoid “fat finger” errors to the selection of colors (especially red and green) to assist those with color vision defects. The more subtle threat of user boredom and distraction explains why car steering wheels are now designed to vibrate when you stray out of lane, and why fake images of guns and knives are randomly projected onto airport X-ray screens.

Automation · The digital feeds and algorithmic analytics that enable systems to be “dashboarded” concurrently make those systems easier to automate. As human agency and oversight are eliminated, the (overt and subconscious) assumptions that guide automation become increasingly critical. If, as Cathy O’Neil cautioned in Weapons of Math Destruction, “models are opinions embedded in mathematics,” the more pervasive the models, the more consequential the opinions.

Panopticon · Consumer dashboard aficionados beware: What starts as a quest for “the quantified self” can slide into state-mandated surveillance. If this sounds like tin-foil territory, note that the government of Singapore has recently teamed up with Apple Watch to reward citizens for participating in “healthy activities” (walking, sleeping, meditating, immunization). And China — which already dabbles with “social credit” scores — is seeking to expand its Covid tracking apps into a permanent “intimate health guardian” that is “loved so much that you cannot bear to part with it.”

Notwithstanding these caveats, the bull case for dashboards is compelling, even if it risks straying into Arthur C. Clarke’s Third Law:

“Any sufficiently advanced technology is indistinguishable from magic.”

Without becoming too science fictional, exciting dashboard opportunities can be seen across the i/o spectrum:

Inputs · A tsunami of new data feeds (catalyzed by the Internet of Things), a revolution in data processing (machine learning, affective computing, sentiment analysis, etc.) and the inexorable expansion of computing power (Moore’s Law and beyond) will saturate the world with accurate, comparable and actionable data.

Analytics · Dashboards will become even more sophisticated and situationally aware — reordering inputs, reprioritizing outputs and reconfiguring layouts to reveal valuable new insights and opportunities. By using big data and deep learning to test hypotheses and game alternative scenarios, dashboards will evolve from illustrating current conditions and guiding short-term tactics to empowering future-shaping strategy.

Outputs · Dashboards will incorporate all the bells and whistles of “multimodal” interfaces — gesture interaction, voice recognition, vision tracking, haptic feedback, robotics, wearables — until they break through the OLED screen to inhabit “mixed-reality” “smart rooms” that make the 2002 movie “Minority Report” look and feel like a 2002 movie.

* * *

As the world braces for the most controversial and consequential U.S. presidential contest in memory, Covid dashboards are competing with electoral dashboards for our attention:

The highly unusual circumstances of this specific election spotlight another significant feature of dashboards — one that is simultaneously a strength and a threat: Dashboards keep you inside the machine.

In environments with taxing “task loads” where distraction can be fatal (airline cockpits, nuclear subs, intensive care), the narrow focus of a dashboard is essential. In more nuanced environments, however, thinking only “inside the box” can pose a range of unknowable risks.

Just as the smartest cars don’t tell you if it’s quicker to take a bus, and the most sophisticated “screen time” apps won’t urge you to ditch your phone, so electoral dashboards can’t visualize scenarios outside the written rules and accepted norms of the Constitution — even when those rules and norms are openly challenged.

Since coming to office, President Trump has repeatedly lied about electoral fraud, claimed he cannot lose a “legitimate” election and refused to commit to the peaceful transition of power. It remains to be seen if these threats to democratic order are real, or merely hyperbolic bloviation. But, in either case, none of them appears on any electoral dashboard — nor realistically could they.

To put it another way: If dashboards have been of limited use in controlling Covid, what use could they possibly be in protecting democracy against a demagogue?

(1) According to the Oxford English Dictionary, the first “dashboard” was “a board or leathern apron in the front of a vehicle, to prevent mud from being splashed by the heels of the horses upon the interior of the vehicle.” In 1904 the term was used to describe the control panel on cars; and in 1990, it was used in the context of computers. For those with an interest in sartorial etymology, dash:dashboard::spats:spatterdashers.

(2) In order to facilitate sight-lines, the British Cabinet table is “coffin shaped” — a metaphor not lost on the country that has suffered Europe’s highest rate of excess pandemic deaths.

This column does not necessarily reflect the opinion of the editorial board or Bloomberg LP and its owners.

Ben Schott is a Bloomberg Opinion visual columnist. He created the Schott’s Original Miscellany and Schott’s Almanac series, and writes for newspapers and magazines around the world.

For more articles like this, please visit us at bloomberg.com/opinion

Subscribe now to stay ahead with the most trusted business news source.

©2020 Bloomberg L.P.