Introduction

‘There’s an app for that’ is a hallmark (and trademark) of a digital age that presents complex problems with technological solutions for near real time use. In 2019, the year leading up to the pandemic, Apple’s trademark ‘There’s an app for that’ signalled over 30 billion app downloads from their iOS App Store, which served over 1.5 billion active mobile and tablet devices. The Google Play app store expands the total to over 110 billion downloaded apps in 2019 (Statista, 2022). Instantly downloadable to a screen for a thumb, apps provide a specific type of ‘solution’ for digital interaction at scale, usually funded by venture capital (see Mann et al., 2022). The scale and ubiquity of these apps is not explained by filling a specific technological niche, but rather shows a configuration of technology, culture, market structures, industry and political networks that work together in what Geels (2014) calls a socio-technical regime. In this case, apps confer social expectations around connectivity, utility, and usability from handheld devices (Goggin, 2021). The crisis of the pandemic created a strong political impetus to offer a ubiquitous digital ‘solution’ via apps within the expectations of our appified lives. Yet, the complexity of integrating contact tracing app data and processes into extant (digital) public health infrastructures was fraught and remains under-researched. App-regime expectations tend to focus CTA research towards user-adoption studies (see Kondylakis et al., 2020; Cinque, 2022; Kostka and Habich-Sobiegalla, 2024) over considering why and how CTA exist, which would enable broader questions about pandemic politics to be considered. There is less research on how CTA come to be ‘implemented’ as public health—which we define through the construction and integration of contact tracing apps into and as digital infrastructures of public health systems. What an be learned about digital pandemic governance and the politics of public health into the future remains limited when framing mHealth from solely user adoption concerns. Our research question asks, what can we learn about mHealth and CTA in particular, if we treat them not as screen-based interfaces with public health, but as digital infrastructures of public health? In other words, what politics and practicalities are involved in implementing CTA as part of real-world infrastructure?

Our research question remains a vital one to answer as the apps on our phones are endemic to modern socio-technical relations. In the last 15 years, mobile phones have come to be sophisticated wearable networked sensor platforms, telemetrically relaying detailed information about their users’ physical life: from gait analysis through measuring minute rotations and acceleration, to GPS coordinates, to temperature, sleep patterns, proximity to other devices with Bluetooth, and so on. The efficacy of joining these mobile platforms into socially mobile and compute intensive sensing services has been considered for some time (Mizouni and El Barachi, 2013), but there was no need to do so for public health until recently.

It was in this context that the COVID-19 pandemic’s first wave swept in. In 2020, as the scale of the pandemic crisis was beginning to be appreciated in technology and policy circles, the need for rapid ‘solutions’ became apparent. Hopes of there being ‘an app for that’ reflected discourses of technological solutionism (Morozov, 2013) and refracted efficiencies of digital government that were often promised (Garson, 2004) but always on the horizon, especially in terms of mobile (Vincent and Harris, 2008) and health (Anthopoulos et al., 2016). Nevertheless, by October 2020 there were almost 500 apps on the iOS App Store related to COVID-19 across 98 countries (Albright, 2024). The raison d’etre of app capability aligned to exposure notification and digital contact tracing. Research on over 150 contact tracing apps showed the heterogeneity of CTA in design terms from ethical-privacy and technical-efficacy dimensions, and whether the private sector (n = 42) or government (n = 110) was building and deploying them (Levy and Stewart, 2021). Yet why and how such numbers of CTA come into being as public health implementations remains under-researched. It requires a perspective and a methodological approach that acknowledges the messy situation of diverse actors, the socio-technical context, and uniquely bounded jurisdictions coming together under the veneer of the ‘app’ solution.

We view this work as the study of information infrastructure (Star, 1999), at interfaces of public health. By interface here we reference the common boundary of two disparate bodies or systems. We consider how the place at which these independent and unrelated systems meet and act on each other are linked through infrastructures that are often understood through the appification of public health. One of our main contributions is to consider how public health platformisation—which from users’ perspectives might be easy to comprehend from primarily interacting with CTA interfaces—is much more complicated and ‘messy’ when mapped across social worlds involved in implementing public health with the infrastructure behind apps’ interfaces. These messy relations stand in contrast to effects-based or critical communication tropes for dyads of citizen-user or public-corporate control, and require a robust methodology to unpack. For these concerns, Situational Analysis is a useful methodology to make the intricate and ever shifting challenges of integration visible by asking how CTA inform, and at times complicate, the working of public health. After we outline relevant literature and our methods, the remainder of the article considers how CTA come to be in situations of politics, public discourse, technology, and organisations. We conclude by considering the relevance of these insights for future work. Overall, our work shows how underneath the veneer of the app interface, CTA incorporation into pre-existing public infrastructures offers a complex and often makeshift backend of social-technical architectures that can magnify the technical limits of the apps, jurisdictional politics, and social processes.

Literature for that

A critical literature of digital app interventions into COVID-19 quickly developed through the pandemic. Digital solutions, specifically CTA, were a common invention across much of the developed and developing world (Taylor, 2021; Mann et al., 2022). Comparative analysis on the ethical and privacy trade-offs (eg. Goggin, 2021, Levy and Stewart, 2021, Fahey and Hino, 2020; Connor and Doan, 2021) showed some of the debates within this solutionism. Critiques of consumer/citizen adoption from socio-technical and behavioural perspectives (see Amann et al., 2021, Klenk and Duijf, 2021, Jörling et al., 2022; Habich-Sobiegalla and Kostka, 2022; Geber and Ho, 2022) described how user-citizens (Nguyen, 2022) became the intersection of public health and corporate-government surveillance (Chen et al., 2022, Kim, 2022), even as early apps failed to interface with health systems or incorporate community participation in design (Idris et al., 2022). Work on CTA provenance opened data sovereignty debates (Musiani, 2022) in public health applications, where states consider how to articulate their jurisdictional authority over multinational digital ecosystems. However, later systematic reviews (Amiri et al., 2023) suggest apps’ interface with health systems or the process of interfacing remains unclear.

There are accounts of early app development for COIVD care by healthcare workers for healthcare workers (Schinköthe et al., 2020; Gios et al., 2021) but these are not positioned to consider integration with larger health systems or epidemiological tracking/tracing functionality. Ideas for new apps (Herrera Montano et al., 2022), the general concepts, applications, and advantages of mobile technology for mobile health in the pandemic (Sharma et al., 2022), and the cadence of app release aligned with government policy were considered (Abbas and Michael 2020, Lee et al., 2021). While accounts of how design and development processes of mobile health apps were undertaken (Ghozali, 2022), including how early app development was impacted by the COVID situation (Herendry, 2020), this work did not turn attention to CTA’s implementation as public health infrastructure.

Other studies considered epidemiological efficacy, ranging from congratulatory to abysmal (see respectively Vogt et al., 2022; Wymant et al., 2021), while systematic analysis of CTA availability and adoption remain mostly user focussed (see Cinque, 2022; Sujarwoto and Mharani, 2023) and ‘best practice’ meta-reviews of CTA apps return a preponderance of design concerns like privacy and user experience (O’Connell et al., 2021). Seemingly out of scope in CTA research has been organisational-infrastructural concerns. Scholarship on CTA that offers substantive focus on implementing CTA—meaning considering the situation of app development, deployment and use by health practitioners remains relatively underdeveloped.

There is a small but growing literature that considers why there are apps for that, and that rely on the study of science and technology to situate ‘the app’ amidst complex mappings of society. As an example, Geenen et al. (2023) show how the ‘digital notification cascade’ impeded the solutionism that CTA promised via technological limits (see also Masel et al., 2023) including process delay in what should have been ‘instantaneous’ notification. Compare that to work drawing from Jasanoff’s (2015) framework of a ‘cascade of deference’ to consider how public policy reacts to uncharted times and critiques techno-deterministic assumptions of what is and is not working by Yerramsetti and Manoharan (2023), who open the ‘epistemic and normative uncertainty’ decision makers faced when deciding on such technical solutions. In short, how policy makers consider their burden and imperative to call for and implement actions (with CTA) when faced with complex information offers new insights to the appified nature of COVID-19 response. Such work also considers how national culture impacts interface and functional design of contact tracing apps design (Oyibo and Morita 2023), or how technical prioritisation of privacy and data security limited the use of an app as an insightful tool for health care professionals (Pihlajamäki et al., 2023). These studies show how a Science and Technology Studies (STS) turn can make the importance of relations between human and non-human actors like institutions, hardware, software and tech corporations explicit when considering governance of CTA (Pizzul, 2023) and consider discourses that would otherwise be missed in the technical ‘solution’ of CTA as publicly presented (see Nguyen, 2022, Kim, 2022). Yet, this work still does not bring to light how public health practitioners came to work with and on CTA as infrastructures of public health. Considering the effects of the pandemic ranging from health to civil liberties, the unseen, non-ideal mess of implementation CTA, remains ripe for exploration.

Even if the design of digital interventions to COVID-19 had an ideal form of contact tracing via digital exposure notification, each app is ‘situated in their local experiences of the pandemic’ (Jang and Fischer, 2023, emphasis added) with both unique features in the app and constraints outside of it. For instance, the use of Open-Source Software in CTA solutions (see Louw, 2023) provides certain politics (i.e. live transparency) and constraints for policy makers, who might not only be listening to epidemiologists and developers, but civil society organisations that specify security concerns of unknown code or are circumspect of consultants working as pandemic profiteers.

It is therefore relevant to think of CTA—and their study- not just as apps for consumers or tools for medical practitioners, but as infrastructure that come to undergird policy decisions, machines of state, private firms and consumer/citizen opportunity. Ali and Khan (2023) offer an extensive survey that traces these infrastructures as a technology stack of mobile health interventions (from software, hardware, communication infrastructures and protocols), and with specificity to contact tracing, choices on centralised (government) and decentralised models (Apple-Google’s EN), including privacy trade-offs. However, the situation involved in implementing such technological infrastructures remains absent from studies in ways that calls for further consideration.

As such in this article we offer a map of the situational perspectives designing and implementing CTA in and for public health. Unlike a near instantaneous download into users’ phones, the integration of apps into systems of public health was not frictionless. Frames that view apps from user-centric narratives of the platformisation of public health (i.e. acceptance studies) can gloss over other perspectives of the situation that offer insights for future success. In our case, contact tracers, health region managers, bureaucrats and other professionals involved in the design and implementation of CTA attest to different facets of the situation, and paint a more complex and nuanced arena of digital epidemiology than current App narratives provide. Through Situational Analysis (SA) we ask how user-acceptance might be decentred to uncover alternative understandings of the situation (Clarke, 2005) of developing, deploying and integrating apps with public health.

Methodology and conceptual framing

While adoption studies have added significantly to CTA literatures, we are interested in the situation that surrounds, produces and must work with CTA in society. We employ Situational Analysis (SA) to uncover the complexity, fluidity and evolving multiplicity of social situations that would otherwise be overlooked when the subject of interest remains centred on the ‘app’. To do SA researchers are tasked with producing maps that help analyse and understand the full scope of a given situation including relationships, and power dynamics that can be unearthed through re-situating once’s analytical point of reference (Clarke, 2005). SA mapping techniques have been refined in the digital era to both incorporate digital methods (Markham and Lindgren, 2014), and further reflect how to continue to capture the constantly evolving elements (including pathogens!) that press situations onward (see Markham, 2022).

Previous work focussed on the COVID-19 Situation shows how ‘othering’ across social, political and media ‘worlds’ helped legitimise biopolitical actions even as these actions exacerbated social, political and cultural difference (Ristić and Marinković, 2022). Ways of adequately mapping the COVID-19 situation vary. Storeng and Puyvallée (2021) analyse public statements from health organisations or accounts framed in broadcast media to consider organisational perspectives vis-à-vis COVID-19 actions. Nguyen (2022) automated the scraping of government discourse in Vietnam to temporally map Milestones towards user-citizenship that Vietnam’s CTA engendered. Kim (2022) similarly showed the evolution of actors partnering in Korean CTA via curating an archive of media for discursive analysis. A thesis by Eveline (2021), interviews six relevant stakeholders creating the ‘Stopp Corona’ CTA, in Austria, along with analysis of primary documents. Notwithstanding contributions outlined here, a gap in the literature remains for critical work that describes and explains CTA by offering a situational analysis of their implementation and is informed by those actors across technology, policy, government and health that are creating, using and acting via app (data) within and to create the situation.

Such focus should be, we argue infrastructural, in that the outcome was not a shiny new app, but a socio-technical infrastructure to allow and analyse data flow distributed across mobile sensor suites that users interpret via apps. What sits beneath the shiny veneer of an app, across its discursive registers, technical parameters and organisational politics, will help us better understand CTA as a socio-technical phenomenon. Such app infrastructures might be harder to make visible for analysis but require careful attention (Star, 2010) as in this case they speak to what more broad pandemic politics were and can be. CTA are entities that users, governments, politicians, industry and other elements act towards (or with) in relation to their own communities of practice—from specific jurisdiction to power hierarchies. This means we are interested in how information infrastructures (Star, 1999; 2010) develop and what they promise amidst the crisis of the global pandemic. We are not aware of work that incorporates categories of (non)human actors and their organisations—namely public health units, government actors, etc—into situational analyses or empirical systematic reviews of CTA literature. We see a need to focus on CTA in ways that include the situational elements that were not highlighted in previous mapping, to focus on how tensions between actors’ political and technical constraints came to define and provide the digital ‘solutions’ to epidemiological problems

Method

To begin to fill this gap, we conducted interviews with 21 practitioners across 5 jurisdictions that worked on designing, deploying and using CTA in public health. The jurisdictions and their technical apps are summarised in Table 1, which includes (nested) changes to the apps and nested jurisdictions. The empirical data produced via our interviews was framed in relation to the aforementioned adoption and efficacy studies, which were emergent at time of interviews, as well as media studies scholarship on user-walkthroughs of CTA (see Yang et al., 2022). Our access to professionals and practitioners ebbed and flowed with successive waves of COVID in each jurisdiction, while corporate actors involved in app development were less inclined to speak of their work on the record, but individual developers’ Twitter posts (specifically @cha_myoung’s 24 Dec thread) offered data to help understand the situation of development where official government-corporate relations would not. For instance, this thread highlighted how while centralised data capture by governments defined traditional contact tracing, Apple and Google developers pushed back both on privacy grounds and design grounds for their Exposure Notification (EN) based apps; EN was not designed for contact tracing, but instead provided a less precise but hopefully more accurate picture of exposure risk in near-real-time; public health infrastructures were sometimes uninterested in that design, or demanded evidence based guidelines from sources like academic journals, even while EN technologies were in the process of being invented; politics in administrations and jurisdictions differentiated acceptance of the need for EN protocols. These elements that emerged in our primary and secondary data capture were layered into Clarke’s situational and arena mapping. And to the crux of it, in alignment with Clarke’s methodology to upend what is known and consider what is left out of maps, we focussed on how to make the infrastructural elements behind the app open for analysis. This led our research to query the politics of technical designs intended to serve as ‘solutions’ to health problems transmitted through social contact. At play is what Yang et al. (2022) have described as focusing on the ‘back end’ to understand platforms as socio-technical systems rather than centering focus on reading user experience. This meant that while rollout dates and public discourse might be necessary (i.e. the marketing of use case, privacy protection information, etc.) other socio-technical worlds not readily available to the public needed to be explored to sufficiently map the situation.

Table 1 Technical adoption aspects of COVID apps per jurisdiction sourced from Albright (2024).
Full size table

Off device, CTA relied on and were meant to direct activity in various systems that did not fit neatly into the app economy or public policy communication: public health policies and on-the-ground practitioners had different views of the COVID-19 CTA situation. This included systems integration with information communication technologies common in public health, like Salesforce or Excel, the institutions involved in policy decisions and practice work (i.e. public health units) and wider political interests are all part of the situation to consider. The visual result of our analysis is provided in Fig. 1, which shows the social worlds that emerged as relational to each other in pathways to implement CTA.

Fig. 1
figure 1

‘Covid Situational Mapping’ provides a multi-layered situational analysis of COVID-19 apps, that moves past the ‘front-end’ (user acceptance and experience) to engage alternative components and dimensions, including implementation and interface and political impetus for seductive solutions.

Full size image

Our primary method to gather data was interviewing of participants with relevant professional expertise in snowballed professional networks from civil society (law and policy, and information security) (n = 6), health (n = 8) (technology, policy, infectious diseases, epidemiology, contact tracing, management), private industry (n = 3), government (n = 3) (human rights, digital contact tracing) and epidemiology (n = 1). This sample reflects our opportunity to snowball a wide range of participants across five countries that had adopted different public health responses using CTA despite similar ‘ideal forms’ of digital tracing techniques and socio-political traditions of public health, which included Australia (n = 15), and then informants located in the United Kingdom (n = 2), New Zealand (n = 1), Canada (n = 2) and Singapore(n = 1). Our rationale was to saturate our understanding in the Australian context as it had both the longest running social distancing and lockdown measures in the western world and a strong federal government push for a CTA as a primary interface to the Nonpharmaceutical interventions expected to lead pandemic control; vaccines arrived late. In line with SA, we also sought to de-centre insights from the Australian experience via practitioners working in additional jurisdictions, which could unveil both local contexts and trans-jurisdictional relations of CTA situations. Our participants varied in role from executives in public agencies to contact tracers working ‘at the coal face’ of the public health system using CTA. This fieldwork occurred across August to October 2021 when successive waves of new COVID variants were encountered across sample countries. This presented a unique window into practitioner practice but slowed our work: we were mindful of adding to our participants’ workload, and only completed interviews post-wave, when our participants were more able, in line with Deakin University human research ethical approval HAE-21-07. Interviews were semi-structured, sharing core attributes as below, while framing questions and follow up prompts with regard to the specified field of expertise. Each interview started with background and experience questions; moved to questions about contact tracing with manual practice and digital data; enquired on specifics of country’s digital solution (e.g. whether Apple-Google EN or otherwise) and the implementation experience and lessons learned in incorporating CTA into their own workflow and the larger public health infrastructure.

Interview data was transcribed and analysed via iterative coding of themes in NVivo and to ensure inter-rater reliability two authors compared their analysis and discussed what the themes ‘meant’, or could be interpreted as meaning, in the context of the dataset and the academic literature. We used the interview themes to both identify and help situate various themes and expressions with adjacent media and policy discourses for each country (some of which were brought by interview subjects, some of which were considered in iterative and intermediate exercises to map pressures and relations), technical explainers on Apple-Google’s ‘Exposure Notification’ technology (Apple, 2020) as well as public pronouncements from actors that worked in private to create public CTA goods (see @cha_myoung 2021) via Apple-Google EN technology, and researcher-user experience of using CTA themselves. Visible from Table 1, and commented on by our interviewees as below, we note the purposes of each App, while ostensibly all CTA, were, in part, determining specific politics by relying on specific digital protocols that either work towards Contact Tracing with a centralised health Authority or Exposure Notification via more decentralised communications. This analysis happened along iterations of producing and repositioning map of elements, along with their categorical ordering, relation and resultant worlds (see Fig. 1 for the visual analysis), which informs the discussion below.

Discussion of emergent themes

Political impetus for the ‘seductive solution’ of CTA

Our first finding is that CTA offered a tangible social-technical interface for crisis politics to spread in public domains. The COVID-19 pandemic could be considered a ‘crisis’ that changed everything, and yet practitioner responses to the pandemic neatly followed trajectories evident in other times of ‘crisis’. Like crises before (for example, 9/11 and the ‘War on Terror’), the state needed to attest to its ability to respond to and control risk. This may help explain the ways in which CTA was developed and understood as ‘solutions’ to the challenges posed by the global pandemic. Our empirical data shows how political pressure to respond resulted in ‘solutions’ designed in ways that do not meet the needs of those working with them, and that the situation is messier than what ‘an app for that’ can offer. We turn to what our Interviewees said about the politicisation of COVID-19 and the need to (re)establish (bio)security. Interviewees commented on political manoeuvres by various governments were marketing solutions demonstrating to their public/s that they were working to resolve the crisis. Further, governments wanted to be seen to be committing all possible resources to identify and implement ‘solutions’, irrespective of the cost. Interviewee 20, an individual with expertise in epidemiology, spoke to the relaxation of spending controls due to what a ‘tough on covid’ response meant for the ballot box:

‘there was no way [the Elected Leader] wanted anything in the media to ever say that money was an issue … It was a really interesting experience for me to see the machinations of the politics…’

Interviewee 20, Health (Epidemiology)

The political pressures to do everything (or at least to be seen as such) is layered not only with roles of the media in democratic politics, but subsequent races to resourcing. Interviewees spoke to the types of political ‘turf wars’ between jurisdictions to resource and develop various (and duplicative) technologies to claim credit in the fight against COVID-19:

‘There’s a lot of taxpayer money that’s spent on duplication for egos and political jostling and the outcome is that we don’t [give or get] tools that work.’

Interviewee 13, Private Industry

As the Interviewee reflects, and we will explore further in the section that follows, despite the pandemic, world views affected implementation of solution strategy. Further, interviewees reflected that political pressure to respond quickly left governments with inadequate time to consult or to think through efficacy (from a perspective of CTA as public health infrastructure). This includes consideration of how COVID-19 CTA are used in practice within complex and bureaucratic organisational systems and at the ‘user-facing’ contact tracing activities in decentralised public health units. These dynamics are made clear when Interviewee 15 discusses the design and deployment of the CovidSafe app in Australia:

‘I begged [the Federal Government] to include civil society in that consultation period. They claimed publicly they didn’t have time, but that’s not true. It was at the ministerial level that that was rejected…there are a bunch of people like [deidentified experts] who literally had really useful input that would’ve helped with the design.’

Interviewee 15, Government (Human Rights)

The empirical data we have gathered from experts working in senior government positions and at the front line in public health units shows the pressures at the political and Ministerial level to respond by quickly designing and implementing ‘solutions’ without time (or perhaps the inclination) to consult with experts, or indeed those tasked with implementing and using the apps in public health responses. Governments would be loath for consultations to delay or undermine actions required in a crisis. This pattern of crisis-actioning is reminiscent of ‘The War on Terror’ where constraints to both budget and checks and balances of governance were cast aside (see Shane, 2011 for Yoo’s ‘torture memo’ in particular) in the need for action. The ‘action imperative’ in a COVID-19 situation is framed by Yerramsetti and Manoharan (2024) in epistemic terms where governance actors are forced to act without clarity and in complexity.

In the complex crisis of COVID-19 these inclinations were operationalised into a simple app. Both Yoo’s memo re Terror and the CTA offered limited understanding of the needs of those working at the front line but offered a ‘solution’ to the respective crises: Just torture them; just appify contact tracing. Writing at the time when COVID-19 first emerged as a global pandemic, surveillance scholars French and Monahan (2020, p. 8) commented that ‘while we do not know how this situation will unfold or resolve, we do have insight into how it fits within existing patterns and relations, particularly those pertaining to sociocultural constructions of (in) security, vulnerability and risk.’ There was indeed immense pressure on governments to act quickly to stop the spread of the virus and implement restrictive measures. Such measures were adjacent to the app, but reflective of its need and in some jurisdictions interoperable with it as a social interface, including, forced lockdowns in many parts of the world (some extending over multiple years), mandatory testing, contact tracing and isolating those with or that had been exposed to the virus (to manage risk), and a raft of measures to evidence vaccination status in order to work, travel or to be in public places such as restaurants where they remained open. This view of the crisis, like crises that have come before, offered a dedicated, coordinated and ongoing effort to surveil populations and use (or at least attempt to) data to drive responses. In short, crisis politics had themselves appified, with direct social-technical purchase to users’ most personal mobile devices, while direct lines of epidemiological efficacy via CTA remained unclear. We do not go so far to claim such politics were merely ‘app theatre’, mirroring a ‘security theatre’ (Schneier, 2009) where appearance of enhanced safety supersedes effectively reducing risk. However, we do note the nature of CTA crisis response could be nothing less than theatrical—promising a digital solution for a microbial and social problem set, all at the swipe of a thumb.

CTA implementation at the interface with public health design

Our second finding had to do with how CTA was not designed for public health, and instead reflected user-individual frames of design and jurisdictional politics and tech debt. Put another way, from one situational worldview, the end-user is not a member of the public, but public health practitioner. It seems that developers, policymakers and politicians did not share this world view. Political pressures to design ‘solutions’ did not match with how interviewees said CTA’s interfaced with socio-technical factors and actors of the public health systems; our data from practitioners reveals basic interoperability issues with extant information systems and human processes.

Apps are often envisioned through their front-end interface and the ideal user who interacts with it. However, this view runs the risk of providing a superficial analysis of a complex mediation of technology and social structures (Light et al., 2018). In the case of CTA, studying the ‘back-end’ of the app allows a much more nuanced look to how the shiny veneer of having an ‘app for that’ interfaces with the complex political, social and organisational realities of public pandemic control. Here we are interested in the complex material reality that sits beneath promises of an ‘app for that’. Specifically, the complexity of integrating contact tracing data from apps and other sources of information into ‘back-end’ systems to manage data and understand outbreaks; the institutional uptake and processes of these workflows in a time of ‘crisis’; and the mundane work of using a surprisingly wide array of ‘different things’ to piece together a system of contact tracing where novel technological solutions interfaced with the mundane, organisational and public. As summarised by one contact tracer: ‘We have different apps’ (21, Contact Tracer).

The practice of contact tracing shifted significantly as the envelope of risk and resources tied to COVID-19 expanded to pandemic proportions. One such change revolved around smartphone apps. While public discourse focussed on the seductive political simplicity of such apps (Storeng and de Bengy Puyvallée, 2021), the reality for making apps work in and with public health systems was much more complex. Our interviews with high level public servants, developers, public health units and contact tracers all showed that CTA were very rarely explained by them in the consumer-centric terms in policy and public discourse. Instead, the common concerns reflected by our sample related to an ecosystem of mundane technological practices that had to be strung together to leverage digital CTA data into human-machine workflows. CTA became the most visible part of socio-technical systems that employed multiple different approaches to capture, store and analyse data, to follow up public health actions, but CTA did not define these systems.

Integration of these systems and their component parts was not frictionless, rather far from it. As discussed above, political pressure reached crisis levels for action to increase resources and technology so to move contact tracing into a new paradigm. As the digital solutions moved downstream, contact tracers were ‘ordered to… not to go off system’ (18) even as the number of systems in use continued to expand and at some points worked at odds with each other. ‘The App’ in one Australian working context consisted of numerous consultant firms (publicly disclosed as Deloitte and Boston Consulting Group) running various phases of implementation and change management at a national level with rejected interventions by handset and operating system monopolies Apple and Google via their joint Exposure Notification framework; palpable tech-debt across jurisdictions at state level from utilisation of inhouse products like the Public Health Event Surveillance System, which was itself a custom version of the commercial product Maven Enhanced Disease Surveillance System, which was replaced by the Transmission Response Epidemiology Victoria Information System (TREVI) as configured by Salesforce and; fax machines, executive leadership (figuratively) parachuted in from the Army, with further troops brought in to support logistics. Within this set of actors, each wave of COVID created changing tracing expectations as operational directives continued to evolve. The UK likewise suffered from operational complexity that had to interface with app data: differences between England, Wales and Northern Ireland due to the ‘different constituent parts’ (7) that made up the political and health systems interfaced with Public Health England, which serves as the executive agency of the Department of Health and developed the CTA. Here Interviewees cited public distrust of government tracking and risk on the department for adding the Google-Apple EN framework into the mix, which was a departure from Australian policy.

In Canada, systemic rifts in techno-social world views were even more visible. The Canadian situation allows us to see not only COVID-19 epidemiology and technical best practices, but intra-country geographic, cultural and political schisms at work despite federal mandates and resources. Interviewees noted that federal-provisional cleavages meant another set of divides existed in implementing the technological plans and referenced privacy and jurisdiction issues at play (9;13). British Columbia was interested in location based QR codes to assist in contact tracing, but the federal government ‘didn’t have the capacity at that time to change the app’ (13). Some provinces and territories ended up involved in the federal government’s initial exposure notification service based on the Apple-Google framework (13). Other provinces opted for other systems, with Alberta’s ABTracetogether aptly named app employing contract tracing in a way that forwarded identities to the province’s central team like Australia’s Federal mandate. That Alberta’s more libertarian government was opting for a more privacy intrusive solution, albeit one not aligned to federal governments and/or big tech—was instructive of previous political schisms between Ottawa and Alberta more than specific technical-efficacy registers.

Even within the Canadian provinces that did agree to adopt the Federal solution, the implementers found that the provincial governments did not have sufficient insight or political sway in the powerful, and plentiful public health units within each province. As an example, Ontario has 34 public health units that one of our Interviewees identified as having ‘a lot of power…[but] nobody really went to the public health units and got public health units onboard’ (13). This was ironically confirmed via the research team, as our entrée into those public health units was unsuccessful.

Differences in CTA implementation existed at technical-process levels in Canada as well. While the federal app generated a ‘one time key’ via Apple-Google’s EN framework for positive testing users to alert proximity contacts anonymously, the process was implemented by public health workers. Our interviewees noted each province had a different system to distribute these keys and produced disparate efficacy for app functionality. For example, in Quebec there was a separate toll-free number that citizens were to phone once they were notified by their health authority as testing positive; Ontario provided an online portal. The app solution was navigating provincial-federal political relations, intraprovincial jurisdictional relations and all the technical resources—and impediments—that worked on and with the appified solution.

The diverse networks of jurisdictional knowledge, technical capacity and shifting policy priorities worked in complex ways. In one example, technical solutions for QR codes were eventually built into the Canadian federal app, but never implemented. Interviewees (13) suggested its (technical) building leveraged lessons learned from the UK and Australia, showing interjurisdictional knowledge transfer. Yet it never launched in Canada due to lack of interest from provincial partners in this ‘solution’ (13). The QR code was a technical solution with a situation that refracted political contexts and cadence of COVID-19 waves and vaccine rollouts. Other app features were not pushed past the design stage due to their political nature. For instance, one Interviewee (11) mentioned that risk calculators would in essence show the unequal distribution of risk in society for catching COVID-19, highlighting the fraught social justice consequences of technologies as ‘solutions’ to social and health problems. Those who could stay at home and then ‘go for a coffee’ from time to time would have a much lower risk score than the grocery store or health worker travelling on public transit (11); there were no design wins to implement this type of informational relation via government apps and exasperate public sentiment toward non-pharmaceutical measures to control the spread of the pandemic.

CTA implementation as a technical infrastructure

The above rifts in practice and technology open further discussion into how CTA are not an interface to public health and should instead be thought of as infrastructures of public health. mHealth cannot effectively be mobile devices supporting public health. It must instead be understood as information infrastructures of public health that touch practitioners and public alike. We state this after our situational analysis underscored that beneath the facade of ‘an app for that’ were depths of mundane organisational technique and technology that acted with and on the app itself—as the app acted on and with these public health infrastructures. Here we focus on how the humans interfacing with technical systems underneath the public ‘app’ crafted expectations of public health. On the other hand—or maybe because of limited functionality of specific systems, contact tracing teams used eclectic mixtures of technological practice. These included the bespoke Salesforce TREVI product for ‘all our notes and everything’ (21), Outlook for contacting peers, Genesys for phoning patients and potential contacts and as a workaround for other issues, Excel. It was not just the existence of so many systems, but the transitions from one to the next, with limited operability, that made contact tracing difficult. And then CTA were added into the mix. The solutionism (Morozov, 2013) of CTA was sometimes acutely felt by health practitioners. As one informant explained, ‘you can’t extract [CTA data to our system], so it’s pointless…they were just trying to make a technology work that just was totally not designed for what they were needing done’ (20). More specifically tracing the frictions outside the app, Excel was used due to Saleforce’s solution making it ‘impossible’ (18) to pass one set of data to the next system. Contract tracers were pensive to admit that the complexity of data integration pushed them to go ‘off system’ and onto paper or excel based record keeping (17, 18). From their perspective it was the software that they were ‘forced to work with’ (18) that drove the change in interface between CTA data feeds and human contact tracers to bespoke bridges of Excel. This human mitigation strategy worked until it didn’t: ‘we had to create our own spreadsheet, which very soon, at 300 cases a day, just became an absolute nightmare’ (17) at which time qualitative prioritisation strategies were employed to cut down on information overload.

The system behind the app (including organisational and tracing protocols) was designed for a threshold of positive cases. As the epidemiology exceeded these, the system was lost and qualitative calls on priority were made at local levels, irrespective of the app data feed. A public health region manager at this point did try to re-integrate the prescribed system but found that its ‘parameters [were] based on what we used to do, to clear cases back in wave two, and that’s all changed’ (17). At the same time, the ‘Salesforce thing … the complexity of it was so huge and it was so difficult to learn that we never used it to its fullest extent.’ (17). The app was the peak of a large and growing infrastructure of human and non-human actors that made up public health. Interoperability between each social understanding of the world (whether Salesforce’s position of technical compliance without epidemiological consideration, or Humans’ knowledge of how to share excel documents) did not help smooth implementation of a an appified ‘solution’.

The business of operationalising the app for public health shows the value of relational analysis, as the situation was far more complex than an automated digital practice of interfacing with the app. Social worldview—including the practical truths and knowledge of experience of practitioners, was on display. This had to do not just with obtaining data but also the partial, provisional and perspectival (Clarke, 2005; 8) work of reporting it within a historical and geographical situation. As one respondent noted: ‘we couldn’t do more than a few [traces] a day by following orders [to stay on the system], so everyone just went off the system and just didn’t tell the leaders that was happening. And then tried to use catchup methods to try and fill in the forms later after the event.’ (18). Teams reverted to or made-up practices that went against directives and systems not fit for purpose in ways they felt allowed them to do their jobs, rejecting the proposed set of system integrations with CTA and then covering their tracks as afterwork. This stemmed from being confronted with the limits of certain systems, trying to work around them and then abandoning them (20).

The technical situational-paucity beneath the app was, as one epidemiologist involved put it, ‘a fundamental lack of the right technology to do all the things that they needed to be done in the contact tracing process’, which created glaring inefficiencies in resource use (20). This view of inefficiency—and related consequences to effectiveness—were not lost to our practitioner interviewees. They expressed the limited agency they felt they had as part of the larger technical system they were made to interface with. One manager of a public health unit covering hundreds of thousands of people noted that they felt they were ‘just the end users of [the technical solutions] and had very little influence over it’ (18). Lack of agency in terms of design or implementation, with large contracts for consultants and technology companies being negotiated at scale and distance unfamiliar to them served to both make the ‘infrastructure’ of CTA invisible until they broke down and worked to ‘other’ human and non-human components of the solution.

Interviewees did speak to specific technical challenges of CTA themselves, including physical limitations to APIs and technologies like Bluetooth, decisions like the seemingly arbitrary cut-offs of ‘contact’ designations—such as 2 m for 15 min, privacy concerns aligned to deploying decentralised or centralised systems for public health purposes and problems of social uptake linked to these concerns. Many of these themes were documented early in the literature (Hogan et al., 2021; Abbas and Michael, 2020) while technocentric ‘solutions’ for such concerns evolve apace (see Shubina et al., 2021). While its infrastructure was invisible even to contact tracing practitioners sometimes, CTA design from the world view of phone-users was a very visible interface between the disease of COVID spreading and the work of contact tracing. This aligns with widespread cultural acceptance (Goggin, 2021) of there being an ‘app for that’ in ways that denote platformisation of information, a frictionless exchange of capital and individualisation of risk in society.

As an exemplar of the rift between expectation of frictionless integration of apps and their data, and the processes of contract tracing, developers were shocked to learn that the proximity-based contact data apps provided to human contract tracers were only looked at after extensive interviews and manual processes were engaged in by public health teams. The developers’ understanding was that the list of potential contacts from the app was to be used as the starting point, efficiently directing future endeavours on probabilities of connection. In the same jurisdiction, there was significant media attention for apps not finding new contacts; this was partly because their intended use was usurped by comprehensive—and arguably resource consuming—manual contract tracing procedures. Yet, the discourse, again, can only point to ‘the app’ as the focal point of the policy response failure, not the depth of digital and human infrastructures that sit beneath the phone-deployed interface. The app-centric world view misses the complex ways in which CTA interface with public health systems, procedures and priorities.

Apps were insufficient to define contact tracing or indeed re/solve the complex crisis that defined the COVID-19 contact tracing situation. The social side of the socio-technical assemblage of contact tracing is not just tracing potential vectors of infection. As one health professional put it, human contact tracers work also involves empathy that is important from a ‘mental health perspective. How [tracers] guide somebody through the process of a positive diagnosis’ (11). Other symptoms of this digital insufficiency were evident by there being too many apps. Governments released some that offered information on symptoms, but also asked or took geolocation data and were thus incompatible with the privacy policies that were promised with CTA, and so new apps had to be created. Canada also mandated an ‘ArriveCan’ app that was designed to expedite health and quarantine arrangements for travellers, and which was confused with the other two apps on offer (13) that provided symptom information and Exposure Notification. In Australia contact tracing was also implemented via QR codes in private unregulated ecosystems until the national public app replaced these Non Pharmaceutical Interventions as the pandemic and the crisis response to it progressed (Yang et al., 2022).

Finally, science itself was operationalised into the apps in ways that were not satisfactory but remained a necessary interface between the history of the science of epidemiology, the needs of public health messaging and the limitations of Bluetooth radio connections. Specifically, the standard of around 2 m and 15 min to determine exposure grew into acceptance across Bluetooth-based CTA. Our interviewees communicated these cut-offs were a probability-based solution based on old science and technical limitations. On the one hand the transmission rates seemed to change between waves/variants of the virus, while on the other ‘rules’ such as 2 m for 15 min seems to have been based on outdated guidance around droplet and aerosol distinctions rooted in 19th Century science (Randall et al., 2021; Tang et al., 2021). Some of our interviewees acknowledged these discrepancies: ‘it was difficult to explain that no matter what variant had come, we probably always would’ve had the 2 m, 15 min, because it wasn’t based on the transmission of the original virus. It’s based on [old studies and technical factors, and] we need a cut off. There is a lot behind that decision, but not necessarily what people want to hear is behind that decision.’ (13). In line with rather blunt pathology metrics, EN notifications employed by most CTA did not include a time of notification, just the day—even though government CTA designers and then users, requested more data. Practitioners saw this as a balance between maintaining privacy and efficacy of the system’s uptake and users’ own goals; ‘people wanted to find ways to think that they were safe’ (13) by providing folk theories towards discounting the ramifications of notification such as time of exposure or whether they were masked, among other strategies.

Such shifts in pathology and methodology are hard to communicate: public health officials in Canada did not want patients behaving differently depending on what variant they thought they caught per wave. On the other hand, a practitioner that helped make the UK app suggested that risk differentials per strain could be a feature not a bug of any app update; it would allow a better risk assessment of transmission and respond with greater accuracy on the probability of isolation being useful. Here the technical app was again interfacing with expression of folk theories and statistical probabilities—for meaning and effectivity, with practitioners having to balance a history of science, user preference and risk assessment and the above-mentioned political and jurisdictional issues—there is still not an app for that. The construction of CTA as the appification of digital solutions misses the informational infrastructural politics that plays out in the diverse practice of public health. Our interviewees, the shifting epidemiological and political realities of COVID-19 and discrete socio-technical capacities all point to imagining the digital necessity of CTA without configuring the social sufficiency of their implementation.

Conclusion

This paper presented a detailed situational analysis of COVID-19 CTA as infrastructure via the complex interplay between technology, politics and surveillance in pandemic management. Although there is extensive research on public adoption to the appification of health services during the pandemic, there has been little focus on the experiences and challenges faced by those directly involved in CTA’s implementation—from executives of public health to contact tracers. Our study addressed this gap by exploring the political and operational dimensions of designing and deploying CTA as technological solutions to a public health crisis. We showed that beneath the shiny veneer of an app interface is the messy certitude of Excel, tech-debt, mundane organisational techniques and jurisdictional politics that reshape public health infrastructures and their outcomes.

Our central claim acknowledges that while CTA adoption, app design and efficacy studies create necessary knowledge about mHealth, our situational analysis shows a need to consider wider sets of concerns of pandemic politics when we more fully understand CTA as infrastructures of public health. We view CTA as something that is acted on and with by practitioners, publics and politicians—they act on CTA infrastructures when they intervene, modify, or develop CTA and act with CTA infrastructures when they rely on their often invisible, back-end processes to carry out public health duties. Even as the CTA we looked at are currently retired, the lessons of our work include coming to see a situation of pandemic politics not only as solutionism, but the combined complexity of turf wars exasperated by interoperability problems; evolving epidemiological science not being matched by app design; and off-system workarounds that put human instincts back into complex integration plans. Apps also personalised and made tangible a politics of crisis. The next domain for app-politics might not be CTA. But, the frictions to surveillant norms and systems that have been experimented with in CTA have touched not just publics, but worked into public health infrastructure. Together these processes point to CTA as informational infrastructures that develop their own set of politics and possibilities. Future designs of mHealth infrastructure should carefully consider these lessons to ensure more thoughtful and effective implementation to use and maintain complex interplays beneath the app(s).

Our analysis is limited insofar it relies on the subjective experiences of our informants to open up the appification of public health and by necessity presents an eclectic mix of practitioners and geographies that built CTA: recruiting pandemic experts during a pandemic was hard, limiting our data via uneven stakeholder types across jurisdictions. But this was both a reflection of sampling limitations as well as the transnational linkages that tied CTA together across jurisdictions (e.g. core BlueTrace tech). While less then systematic, we view the knowledge reflected upon in-situ by participants as nonetheless useful to discern patterns of experience in creating and using CTA in public health. Acknowledging relations created through CTA development and deployment have helped decentre dominant discourses of mHealth appification (e.g. user acceptance) in ways that can better inform design and implementation of digital health in the future.