On October 1, New York State released a mobile app to help citizens anonymously share positive Covid-19 diagnoses and warn others of possible exposure to Covid-19. Announcing its release, New York Governor Andrew Cuomo proclaimed that this app will “not only bring contact tracing to a new level,” but also “give people comfort.”1 Others had a less rosy outlook on the technology. Less than two months later, amid a resurgent holiday wave, the State of California released its Covid-19 app. Governor Gavin Newsom cautioned during the launch’s news conference that he “didn’t want to ‘overstate’ the utility of the app.” Newsom keenly observed that “the more people that participate in it…the more effective [the] program can be.”2 Why aren’t more people using Covid-19 apps, and why does it matter?
The success of software is limited by the actions its users take. Most software seeks to solve real world problems, and they must weigh competing values to determine what costs to impose on users. In this case, Covid-19 apps aim to contain the pandemic more effectively and save lives. As a technical solution, the apps must strike a compromise between privacy, free movement, and low viral transmission. However, these underlying value trade-offs are not always made explicit to users. As a result, misunderstanding and mistrust have blinded the public from recognizing the incompatibility of their priorities and discouraged them from adopting Covid-19 apps. Ultimately, Covid-19 apps remind us that in order for technology to be successful, acceptance that competing values are unavoidable is necessary.
Let’s first look at why, and how Covid-19 apps can help fight the pandemic. To be clear, Covid-19 apps are a supplement to, rather than replacement for, other disease outbreak interventions. Epidemics only end when chains of transmission are cut. To this end, the standard approach of infectious disease experts is “test-trace-isolate”: After a diagnostic test discovers an infected person, public health officials identify that person’s close contacts and instruct the infected to self-isolate. Then, the infected’s close contacts are asked to self-isolate and get tested, repeating the cycle.3 During a pandemic, contact tracing can be particularly challenging: patients may misremember their contacts, contacts don’t consistently answer phone calls, and public health staff are often overwhelmed by case numbers.4 While the exact features vary between states, Covid-19 apps focus on automating contact tracing. Central to the New York and California apps, along with 18 other states, is the Exposure Notifications feature developed in a collaboration by Apple and Google. Exposure Notifications allow users to anonymously share Covid-19 diagnoses with other smartphone owners who came into close proximity with the user in the previous two weeks.5 Notified users can also receive recommendations from public health authorities on testing and self-isolation. This automation sidesteps imperfect human memory and limited contact tracing staff, reaching more of the exposed and notifying them faster than conventional contact tracing. Crucially, Covid-19 apps are not a solution to viral transmission in-and-of-themselves. They exist in a symbiotic relationship with mass testing and self-isolation of the infectious as necessary but individually insufficient pandemic interventions.
The value offered by Exposure Notifications is information—anonymous alerts of exposure to Covid-19. Here is an analogy: whenever Alice gets physically close to someone, she gives them a random sketch. Then Alice finds out she is sick and wants to tell everyone what was in contact with, so she puts a copy of her last two weeks of sketches in public. Because the sketches are random, there is no way for anyone to know Alice in particular is sick. Bob keeps a copy of every sketch someone gives him and periodically checks against the public sketches. After meeting up with Alice he finds a match, but since Bob doesn’t track when he got each sketch, all he knows is that he came into contact with someone sick, not that Alice is sick. Alice can share her diagnosis with Bob without revealing anything about where she has been, who she is, or who she has been in contact with. Exposure Notifications work by passing anonymous messages between smartphones and would provide Alice and Bob with the same privacy protections.
Self-isolation is necessary to curb the spread of Covid-19, but Exposure Notifications provide information to make quarantines targeted. The pandemic will only end when either SARS-CoV-2 runs out of possible hosts (through herd immunity or a vaccine) or transmission reduces into exponential decay. Unfortunately, the high transmissibility of SARS-CoV-2 poses challenges to achieving the latter. SARS-CoV-2 is an airborne virus that can survive on surfaces for days, and up to 40% of Covid-19 cases may be asymptomatic. 6, 7 The CDC estimates that asymptomatic or presymptomatic individuals cause up to 50% of SARS-CoV-2 transmission. As a result, self-isolation is necessary to stop chains of community spread. Unable to identify most non-symptomatic individuals, many American states and counties have resorted to blanket shelter-in-place orders.8 By automatically notifying users, Exposure Notifications inform users to self-isolate before the onset of symptoms or a positive diagnosis. Researchers first estimated that 60% of the adult population would need to opt-in to contact tracing apps to have the same impact as a shelter-in-place order. Widespread usage of Exposure Notifications presents a clear advantage over stay-at-home orders: targeted quarantine would allow more businesses to remain open, lightening the economic impact of a protracted pandemic.
Contact tracing apps, like all public health interventions during a pandemic, would necessarily strike a compromise between privacy, free movement, and low viral transmission. These can be three mutually exclusive goals: the end of the Covid-19 pandemic, protections of individual privacy, and the reopening of businesses and communities. The ease of SARS-CoV-2 transmission presents an inescapable tradeoff between preventing viral transmission and freedom of movement. A shelter-in-place order would be a more extreme restriction on communities, but the same end can be achieved by target quarantine of exposed individuals and contact tracing. But this presents a second tradeoff: effective contact tracing violates individual privacy. To identify those who must isolate, public health authorities must intrude into individuals personal lives. Surveillance technology that is private-by-design, like Exposure Notifications, is still surveillance technology. Covid-19 contact tracing apps present a compromise, attempting to target movement restrictions, protect user identities, while recommending to users the actions needed to reduce community spread. In doing so, Exposure Notifications make a value judgement on the relative importance of the three priorities. Technology can’t offer a shortcut through value tradeoffs. It merely offers means to achieve the compromise decided by human stakeholders.
User-centric Exposure Notifications take a privacy-first approach to contact tracing and rely on individual responsibility for success. Exposure Notifications are based on a protocol developed by an international community of cryptographers and legal experts with the express goal of “ensuring that any proximity tracing technology doesn’t result in governments obtaining surveillance capabilities which will endanger civil society.”9 Users must opt-in to Exposure Notifications, which doesn’t track user locations and keeps user identities anonymous to Google, Apple, government authorities, and other users.10 Rather than using WiFi or GPS, which locate a smartphone, Exposure Notifications use Bluetooth to communicate with nearby smartphones without location data. Possibly most convincing, Exposure Notifications is open source, meaning all of the above claims are verifiable by third parties.11, 12 By respecting user privacy, Exposure Notifications place responsibility for preventing community spread on users. Like other public health measures in the U.S., compliance is voluntary.13 Because Exposure Notifications protect user anonymity, governments couldn’t compel compliance even if desired. Users must choose to take action, whether it be sharing a positive diagnosis or following health guidelines after receiving a notification.
Public mistrust and misunderstanding of Exposure Notifications has led to low user adoption, reducing their usefulness. As of 2019, 81% of American adults own smartphones; however, by late October last year, less than 10% of adults in nine of the ten states using Exposure Notifications had downloaded a Covid-19 app.14, 15 Modeling suggests that even “15% adoption could reduce Covid-19 infections by 15% and deaths by 11%”—not enough to end the pandemic, but significant as the U.S. crosses a million cases and tens of thousands dead per week.16, 17, 18 More recently three states reached 20% downloads.19 These states’ relative success owes to reminders from iOS and Android, while states using conventional marketing have maintained lower adoption rates. Rapid app adoption relies on either mass marketing or word-of-mouth virality, neither of which state health officials may have access to. Marketing funding is scarce, and the anonymity of Covid-19 contact tracing apps impede the sharing necessary for an app to go viral.20, 21 Public opinion may be mixed at best; polling suggests that just shy of half of Americans support mass adoption of smartphone contact tracing.22 After years of illegal surveillance and data abuse by the government and technology companies, public skepticism is justified.23, 24 Social media misinformation, skepticism of contact tracing, rising distrust of experts, and politicization of the pandemic likely contribute to further hesitancy.25, 26 Beyond mistrust, Americans misunderstand Exposure Notifications and their privacy protections. Despite explaining otherwise, one poll found that “30 percent of respondents believed the apps would identify infected people by name and share those names.”27
Whether resulting from mistrust or misunderstanding, the public doesn’t have confidence in Exposure Notifications: six-in-ten Americans believe that location tracking through cellphones wouldn’t make much of a difference in limiting the spread of Covid-19.28, 29 Public skepticism is a self-fulfilling prophecy. The lower the adoption in a given community, the fewer exposures to Covid-19 the apps can detect and the more confidence is eroded.
Ultimately, the voluntary compliance necessary for Exposure Notification’s success likely doesn’t exist. To reduce viral transmission, Covid-19 must promote behavior that follows public health recommendations. Local health department contact tracers have reported that persuading close contacts of the infected to self-isolate has been particularly difficult. If the public disregards instructions from local health officials, it is hard to believe that they would follow the recommendations of an app. Covid-19 apps can tell you when to self-isolate, but don’t make that act any easier or more appealing. Even encouraging users to share positive diagnoses proved challenging in a pilot program conducted at the University of Arizona. Despite a tech-savvy population and university endorsement, “fewer than half of people who tested positive for the virus were using the app,” and among app users only half opted to share their diagnoses.30 Supporters of Exposure Notifications argue that any level of adoption “[provides] some level of protection” against viral spread.31 However, the technology itself doesn’t provide protection; it provides information. The direct value of Exposure Notifications is information about possible exposure to Covid-19. Protection only comes from individuals taking action.
Lack of user trust and engagement pose significant barriers towards the effectiveness of Exposure Notifications in the U.S. While downloads and compliance may increase over time, clearly Exposure Notifications are no panacea. Covid-19 apps are not necessary to end the pandemic, but they can be an effective tool. Technology that contributes to solving “big problems,” like Exposure Notifications with the pandemic, require uncomfortable value judgements. The Covid-19 pandemic presents fundamental trade-offs between privacy, movement, and disease transmission. Americans can dislike the compromise Exposure Notifications present, but the trade-offs that Covid-19 apps struggle with are unavoidable. For technological solutions to be effective, engineers, users, and policymakers must reflect on their competing values and ensure that technology is designed in dialogue with its stakeholders. There are no free answers to the pandemic. Just as technology can’t bridge tradeoffs between priorities, it can’t solve a problem where individuals are not willing to act.