Intrusive snoop apps on the phone? Not if Ich Bin can help it.

Survey Finds Americans Skeptical of Contact Tracing Apps.

Confusion and skepticism may confound efforts to make use of digital contact tracing technologies during the COVID-19 pandemic. A recent survey found that just 42 percent of American respondents support using so-called contact tracing apps—an indication of a lack of confidence that could weaken or even derail effective deployment of such technologies.

Most contact tracing apps generally try to collect some form of information about a smartphone user’s encounters with other people and notify those users if they were potentially exposed to a confirmed COVID-19 case. But each app has its own approach to privacy and can differ in whether it collects more specific location data based on GPS or merely records close encounters with other smartphones based on Bluetooth radio-wave transmissions. Those differences, coupled with public misunderstanding of different apps, can make it tricky to assess public opinion of specific digital contact tracing technologies.

“We found that there is variation in terms of how willing people are to download the apps based on the features of the app,” says Baobao Zhang, a  Klarman postdoctoral fellow at Cornell University whose research focus is the governance of AI.  “There’s many different kinds of apps that are out there, so if you’re just going to ask about a contact tracing app, people might have very different views of what it does.”

In late April and late June, Zhang and her colleagues conducted two rounds of surveys focused on gauging American opinions of such apps. The results are described in a preprint paper first published on 5 May and later updated on 29 June; the update accounts for an initial problem with the survey software and includes a second round of survey findings.

The contact tracing apps in question build upon traditional contact tracing, but they are not the same. Traditional contact tracing is a tried-and-true public health measure that requires large numbers of human contact tracers to call and interview suspected or confirmed COVID-19 cases about their travel history for the purpose of warning family, friends, or strangers who may have been exposed.

But given how labor- and time-intensive manual contact tracing can be, some governments and companies have looked to digital contact tracing systems to help automate part of the process. These systems can include contact tracing apps, which in their most privacy-preserving form may be more accurately described as exposure notification apps. They function primarily to alert individual smartphone users rather than public health officials or human contact tracing teams.

Zhang and her colleagues used conjoint analysis to try gauging how Americans valued or viewed certain features of such apps. For example, app designs can can choose whether to collect GPS location data or to rely primarily upon Bluetooth key code exchanges. Whereas a GPS-based app might notify the user about their potential exposure to COVID-19 at a particular location—information that could also help jog fuzzy human memories—the Bluetooth-based app would typically only tell users that they had potentially been exposed to someone for perhaps a certain amount of time.

A GPS-based app is “probably is more effective from a public health standpoint,” says Sarah Kreps, professor of government and adjunct professor of law at Cornell University and coauthor of the survey research paper. But Kreps adds that the same GPS location data is “very intrusive from a privacy standpoint” because the information could reveal behavioral and lifestyle patterns about a person’s daily life.

Americans who took the survey did not seem to view apps very differently based on whether they incorporated GPS or Bluetooth. But respondents did change their minds when it came to whether an app featured a centralized vs. decentralized system of data storage. The centralized system shares much more information—such as a user’s anonymized ID and Bluetooth key codes—with a central server that might be overseen by a company or government agency. The decentralized system typically stores most of the collected data on individual phones in the interest of better protecting user privacy.

“We found that the decentralized data storage in the contact track tracing app increases people’s willingness to download it,” Zhang says.

But the survey research also suggests that it’s easy for people to get confused about which app does what despite the survey’s best attempt to also educate people about different app features. For example, the survey took time up front to explain that such apps would not identify anyone by name. Still, a later question showed that 30 percent of respondents believed the apps would identify infected people by name and share those names with smartphone users who might have been exposed to them.

“What’s interesting in our study is that even after informing respondents about how these apps work, we did a manipulation check to see if people understood and they don’t always get it right,” Zhang says. “So in terms of public education, I think there’s a lot more work to be done to correct some of the misinformation about these apps.”

There is certainly no shortage of confusion about digital contact tracing efforts. One prominent example is the Google and Apple Exposure Notification (GAEN) system. GAEN makes it easier for third-party developers to create apps that harness Bluetooth capabilities in both Android and iOS devices to exchange randomly-generated IDs whenever phone users are in relatively close proximity. Some countries have already built and deployed such apps based on the GAEN framework, but the handful of U.S. efforts attempting to do so have not yet been rolled out to the public.

The tech giants also worked to enable GAEN at an operating system level so that individuals can go into their smartphone system settings and choose to opt-in for receiving Bluetooth beacon notifications about having been in close proximity to a confirmed COVID-19 case who was using a GAEN-compatible app. If they hadn’t downloaded a GAEN-compatible app already, the notified users would then be prompted to download such an app to get more information.

But some smartphone users became alarmed when that option appeared in the system settings of their Android and iOS devices as part of routine software updates in June. Zhang recalled friends calling her and asking about whether Google or Apple had installed an app on their phones that was tracking them somehow. In reality, the GAEN system would not share an anonymized individual’s health status unless that person chose to opt-in via their phone’s system settings, downloaded a compatible app, and then manually entered the fact that they had tested positive for COVID-19 into the app.

“It was sort of this like shadowy feature of the [operating system], that I think because it didn’t the accompany the actual app, there was almost this suspicion that something was operating in the background without people knowing about it,” Kreps said.

The public reaction may have something to do with survey results showing that just 35 percent of Americans felt that Google and Apple should automatically install such an app on their phones through a software update. The GAEN system roll-out to Android and iOS devices did not automatically install such an app, but the distinction seems to have been lost on many people.

Furthermore, the survey research found no significant difference in people’s willingness to download an app based on whether it was developed by the Silicon Valley tech giants Apple and Google, by the U.S. Centers for Disease Control and Prevention (CDC), by a state government, or by university researchers. By comparison, an online survey commissioned by the software security company Avira found that American respondents tended to trust Apple and Google more than the government on contact tracing apps, even as overall support for contact tracing apps remained low.

Clumsy deployments of contact tracing apps in the United States may have hurt broader public trust in such technologies. North and South Dakota were among the first to make a GPS-based contact tracing app publicly available. But a privacy software manufacturer named Jumbo revealed that the app had violated its own privacy policy by sharing location data with the company Foursquare.

“That’s just a sort of public health disaster, because those kinds of episodes really undermine the trust that is necessary from the public for these kinds of apps to work,” Kreps says.

Data compiled by MIT Technology Review on national efforts to deploy contact tracing apps suggests that most have failed to gain traction among even a simple majority of their citizens. But Americans might feel more confident in exposure notification apps if the U.S. enacts more national and state laws that clearly protect individual privacy rights. In that spirit, U.S. lawmakers in Congress have introduced several bills that propose to regulate the health data collected by such apps and similar digital contact tracing technologies.

One of the more narrowly-focused examples is a bipartisan bill named the The Exposure Notification Privacy Act (PDF) that was introduced in the U.S. Senate. “It makes sure that the app is voluntary and it prohibits app developers from using the data collected by the app for commercial purposes, so I think that’s moving in the right direction,” Zhang explains. But it’s unclear if such proposals can gain traction while U.S. local and state governments grapple with fresh COVID-19 outbreaks in the wake of attempts to reopen businesses.

The survey suggests that American support for contact tracing apps, which sits at 42 percent, puts it squarely behind approval ratings for several other public health surveillance measures—notably traditional contact tracing and temperature checks. (Both of those measures received the backing of more than 50 percent of respondents.) But public opinion is not necessarily set in stone, if some of the evolving views reflected in the two survey rounds are any indication.

“Temperature checks weren’t ranked number one in terms of public support last time and now it is,” Zhang says. “Maybe because it’s become more common, people are more accepting of it.”

The survey also found that political affiliation plays a role in American views on some public health surveillance measures. Overall, more Democrats tended to favor many of those measures compared with their Republican counterparts. But the study found no significant partisan difference in support for contact tracing apps. That may hint at an opportunity for a bipartisan push to help Americans better understand and potentially try such apps, the researchers suggest.

”Who would have thought the masks would be politicized?” Kreps says. “But it suggests that probably everything eventually will be,” she predicts. “Given the partisan polarization in the political landscape, it suggests that if [digital contact tracing] is going to be successful, public health authorities might want to get out in front of it to depoliticize it to the extent possible.”