Only Surveillance Can Save Us From Coronavirus
Big data offers tools to stop the pandemic right now—if we change our definition of privacy.
In those parts of the world under lockdown, people have discovered a renewed appreciation for the internet and the resources it offers—books, movies, music, endless news content, and the tools to communicate with friends and family. Imagine the lockdown without the internet; it is not a comforting thought.
But there is a second network closely rivaling the internet. It, too, is present everywhere, connecting people and countries with unmatched speed, a global organism that constantly replicates information. This second network is the coronavirus.
The two have remained largely separate, hampering our efforts to manage and fight the pandemic. Advertising campaigns have produced posters for city streets covered with giant viruses in an effort to keep us at home. But the virus itself has remained invisible; it has even largely escaped the magnifying glass of the internet.
It could be otherwise. Consider Google co-founder Sergey Brin’s immediate response to the 9/11 attacks, as described in the memoir of a former employee. He rushed into his company’s headquarters possessed by the thought that the company’s search logs could reveal the identity of the terrorists. He understood that by then the internet was no longer simply a tool. It had become an artificial world where a large part of our lives took place. In this artificial world, troves of data were readily available, and Google could work as a sensor of potential threats or anomalies.
[Mapping the Coronavirus Outbreak: Get daily updates on the pandemic and learn how it’s affecting countries around the world.]
Since then, the internet’s system of surveillance has been perfected beyond our imaginations (and usually beyond our acknowledgment). It now extends over email, social media, credit card payments, cell-phone use, and travel records. Every flight and hotel booking, every payment and bank transfer, every call, every picture can be brought together in a single platform and mined for interesting or unusual patterns.
Yet, while extending in every direction, this system was completely powerless to warn us about the coronavirus or to help us fight it. That is about to change—and our concept of personal privacy will have to change with it.
So far the Western counterpandemic effort has been stunningly primitive. Social distancing policies have been imposed in a haphazard way, along a predetermined spectrum and in reaction to outdated data. In many places, governments have little recourse but to send bureaucrats driving around town to monitor behavior.
If only governments used all the technology already available to them. Each case could then be treated according to individual parameters. Once a person has been confirmed to be infected, his or her close contacts could automatically be traced and instructed to get tested. Meanwhile, the infected person’s compliance with lockdown instructions could be tracked using digital tools that monitor individual travel and behavior patterns. Of course, this would require governments accessing cell-phone users’ individual data—and eliminating the legal hurdles currently preventing them from doing so. (In Germany, Deutsche Telekom has shared aggregated data with health authorities, in compliance with existing privacy laws, to help measure overall social distancing. But Ulrich Kelber, Germany’s federal data protection commissioner, quickly added that tracking individual smartphones to monitor quarantine would be a “totally inappropriate and encroaching measure.”)
The United States is already making initial efforts in this direction. On Tuesday, Politico reported that White House senior advisor Jared Kushner’s task force had reached out to technology companies with the goal of creating a national coronavirus surveillance system. According to those reports, the government would not be responsible for directly collecting health data but would rely on states, hospitals, and health data companies. In some cases, the goal would be to rationalize the use of public health care resources, but in time the potential to monitor the health data of individual citizens would be practically unlimited. Some have compared it to a Patriot Act for biosecurity. (U.S. President Donald Trump said in a recent press briefing that he liked the idea because it sounded “scientific.”)
The counterterrorism analogy is useful because it shows the direction of travel of pandemic policy. One obvious possibility as we slowly lift the physical restrictions now in place is to replace those blunt instruments with something more advanced and intelligent. Imagine a new coronavirus patient is detected. Once he or she tests positive, the government could use cell-phone data to trace everyone he or she has been in close proximity to, perhaps focusing on those people who were in contact for more than a few minutes. Everyone in that list would receive a message ordering them to immediately go into isolation. They would themselves be tested, and the process would start again.
Your cell-phone signal could then be used to enforce quarantine decisions. Leave your apartment and the authorities will know. Leave your phone behind and they will call you. Run the battery down and a police car will be at your door in a manner of minutes, as happened recently in Taiwan, where the government has instituted arguably the most advanced coronavirus surveillance system anywhere in the world.
In the United Kingdom, the National Health Service is developing a contact-tracing app that would keep tabs on close-proximity contacts through Bluetooth signals and then automatically notify those people if a user tests positive. According to internal documents, that basic service might be upgraded with a number of extra functions, such as notifying users if they spend more than one hour out of their houses during a lockdown by nudging them to go back home.
Would it work? The first answer is that some kind of contact-tracing app has become inevitable. The current restrictions are not sustainable over the long term; it’s just that no one feels brave enough to lift them without introducing some replacement capable of delivering the same security with fewer social and economic costs. The proper framing for the question is as a choice between the current restrictions and a new surveillance system. Some have argued that traditional tracing methods will be sufficient, but their inadequacy in dealing with an outbreak of the sort we now face has been thoroughly discussed in several recent scientific studies. That should have been obvious; just imagine a couple of health officials trying to contain an outbreak by driving around town to interview potential spreaders, asking them to write down the names of everyone they talked to since being infected.
Privacy advocates will find it difficult to convince the public that traditional definitions of privacy are of higher value than the freedom to leave your house. No one should want to abandon privacy as a value, of course, or even to downgrade its importance. But it’s entirely possible to develop a concept of privacy that is less metaphorical, with new boundaries consistent with the practical challenges society now faces.
There are legitimate causes for concern. It is one thing for authorities to be given enough power to perform their role. Too much power might be abused. We know that vast surveillance powers conferred to fight terrorism can and in some cases have been used, for example, to combat illegal immigration. Similarly, health data gathered to fight a pandemic could end up being used, for example, to determine health insurance premiums.
Secondly, there is something too arbitrary about the system as it is currently being developed: a text message out of nowhere and suddenly you might be locked at home. If not designed carefully, this might not end up being a tool to help us guide our conduct but rather to punish us for faults we have no idea how to avoid.
But while these are valid concerns, they should not be decisive. The analogy to the counterterrorism system developed after the 9/11 attacks shows that there are ideally two different phases in the operation of a surveillance system. In the initial stage, it is still performing under stress as the threats it was built to prevent are already present in our midst. As the system turns increasingly toward prediction and prevention, individual behavior is significantly less affected by it. In time, it is possible to approach something of a steady state where surveillance is directed toward external threats rather than the routine or normal operations of a given society.
There are always points where a free society is in direct contact with external threats. The important thing is to limit the operations of every surveillance system to this external perimeter. There should always be a private area to which citizens can retreat; that area need not, and must not, be brought under the system’s observation and control.
Critics will have to confront the question of what alternative they propose. Surveillance is meant to stop or prevent intrusions before they penetrate a society’s defenses. For those who insist on lower levels of surveillance, the only alternative is to build a more resilient society, one that can deal with those intrusions after they happen. That may sound much more attractive, but in reality the social and individual costs can be forbiddingly high. Take the case of future pandemics. A resilient society would have to invest massively in health care, creating large redundancies to be used in an emergency. It would have to train its citizens in a range of appropriate behaviors. It might even have to redesign its cities and transportation networks and change its labor laws and economic structures so that a potential outbreak would remain contained. At this point—to return to our analogy—some might be tempted to say the virus had won.
Privacy hawks may disagree, but a surveillance system can be a way to leave people alone. Properly designed, it consists of a kind of security perimeter—mostly virtual—within which life can go on undisturbed. The brunt of power is exercised at the perimeter, with as little as possible seeping inside.
In any event, we are only at the beginning. The real game is not to create the technology to monitor restrictive measures during an outbreak but to prevent it from happening in the first place. In two or three years, after the coronavirus threat has receded, the challenge will be to build the kind of system that can prevent future pandemics before they occur, just as the goal of an effective counterterrorism policy must be to stop potential attacks during the planning stage rather than to catch the culprits after they succeed.
Remember again Google’s Brin and his realization that his company gave him an Eye of Sauron into the past. Using search logs, he could travel back in time and catch those responsible for the attacks. In principle, a similar method could be used to detect a viral outbreak in real time. If a number of people in the same geographical cluster search for the same symptoms on the internet, a powerful algorithm will know what is happening before any scientist can even suspect it. As the New York Times has pointed out, in the week ending on April 4, searches for “I can’t smell”—an acknowledged symptom of coronavirus infection—were highest in New York, New Jersey, Louisiana, and Michigan, four of the states with the highest prevalence of the novel coronavirus. In fact, searches related to “loss of smell” during this period almost perfectly matched state-level disease prevalence rates. And searches for “non sento odori” (“I cannot smell”) were spiking in Italy as the coronavirus outbreak spread, well before a report was released identifying the possible symptom.
Authorities are likely to want to probe even further. One method that has garnered attention is to use internet-connected thermometers to upload temperature readings to a centralized platform. One U.S. company, Kinsa, created a national map of fever levels in late March and has been faster and more accurate than anyone else at anticipating coronavirus infection trends. Kinsa has more than 1 million thermometers in circulation and has been getting up to 162,000 daily temperature readings since the virus began spreading in the United States. The company recently adapted its software to detect spikes of fever that do not correlate with historical flu patterns and are likely attributable to the coronavirus.
In this case, readings are voluntary, and they should remain that way. But if the mandatory use of network-connected temperature bracelets will remain a concept for science fiction stories, one should be less affirmative about the use of temperature sensors in public taxis, aimed at detecting the atypical arrival of infection carriers at a given city or country.
The dirty secret is that the use of big data and predictive algorithms has become a routine occurrence in counterterrorism and law enforcement. The chances that the system will not be gradually expanded to include biosecurity look very slim, not least because the coronavirus pandemic has already become immeasurably more devastating than any terrorist attack could hope to be.
Ultimately, we need to bring the internet down to the physical world to enhance our ability to understand physical processes in real time. The alternative is to retreat entirely to the digital internet—to work remotely and talk remotely and live remotely—while abandoning the natural world to our enemy, the virus.