Proctoring and Contract Cheating

'Good Proctor or “Big Brother”? AI Ethics and Online Exam Supervision Technologies' by Simon Coghlan, Tim Miller and Jeannie Paterson comments 

This article philosophically analyzes online exam supervision technologies, which have been thrust into the public spotlight due to campus lockdowns during the COVID-19 pandemic and the growing demand for online courses. Online exam proctoring technologies purport to provide effective oversight of students sitting online exams, using artificial intelligence (AI) systems and human invigilators to supplement and review those systems. Such technologies have alarmed some students who see them as ‘Big Brother-like’, yet some universities defend their judicious use. Critical ethical appraisal of online proctoring technologies is overdue. This article philosophically analyzes these technologies, focusing on the ethical concepts of academic integrity, fairness, non-maleficence, transparency, privacy, respect for autonomy, liberty, and trust. Most of these concepts are prominent in the new field of AI ethics and all are relevant to the education context. The essay provides ethical considerations that educational institutions will need to carefully review before electing to deploy and govern specific online proctoring technologies.

The authors state 

Recently, online exam supervision technologies have been thrust into the public spotlight due to the growing demand for online courses [Ginder et al., 2019] and lockdowns during the COVID-19 pandemic [Flaherty, 2020]. While educational institutions can supervise remote exam-takers simply by watching live online video (e.g. via Zoom), an evolving range of online proctoring (OP) software programs offer more sophisticated, scalable, and extensive monitoring functions, including both human-led and automated remote exam supervision. Such technologies have generated confusion and controversy, including vigorous student protests [White, 2020]. Some universities have dug in against criticism, while others have outright rejected the technologies or have retreated from their initial intentions to use them [White, 2020]. At the root of disagreement and debate between concerned students and universities are questions about the ethics of OP technologies. This essay explores these ethical questions. By doing so, it should assist students and educators in making informed judgements about the appropriateness of OP systems, as well as shining a light on an increasingly popular digital technology application. 

OP software platforms, which first emerged in 2008 [ProctorU, 2020b], are now booming. A 2020 poll found that 54% of educational institutions now use them [Grajek, 2020]. Increasingly, OP software contains artificial intelligence (AI) and machine learning (ML) components that analyse exam recordings to identify suspicious examinee behaviours or suspicious items in their immediate environment. OP companies, which can make good profits from their products [Chin, 2020], claim that automating proctoring increases the scalability, efficiency, and accuracy of exam supervision and the detection of cheating. These features have an obvious attraction for universities, some of which believe the benefits of OP technologies outweigh any drawbacks. However, the complexity and opacity of OP technologies, especially their automated AI functions [Hagendorff, 2020], can be confusing. Furthermore, some (though not all) students complain of a “creepy” Big Brother sense of being invaded and surveiled Hubler [2020]. Predictably, some bloggers are instructing students how to bluff proctoring platforms [Binstein, 2015]. 

Scholars have just begun exploring remote and automated proctoring from a range of perspectives, including pedagogical, behavioral, psychological, and technical perspectives [Asep and Bandung, 2019, Cramp et al., 2019, González-González et al., 2020]. Nonetheless, and despite vigorous ethical discussion in regular media [Zhou, 2020], blog posts [Torino, 2020], and on social media, the ethics of emerging OP technologies has received limited scholarly analysis (cf. Swauger [2020]). Although moral assessments can be informed by empirical data about online and in-person proctoring — such as data about test-taker behavior [Rios and Liu, 2017] and grade comparisons [Goedl and Malla, 2020] — moral assessments depend crucially on philosophical analysis. In the following ethical analysis, we identify and critically explore the key moral values of academic integrity, fairness, non-maleficence, transparency, privacy, autonomy, liberty, and trust as they apply to OP technologies. 

Some of these concepts are prominent in the new field of AI ethics [Jobin et al., 2019], which is burgeoning as AI moves increasingly into many facets of our lives, including in education. In this paper, we suggest that OP platforms are neither a silver bullet for remote invigilation nor, as some would have it, a completely “evil” technology [Grajek, 2020]. This ethical analysis will help to inform concerned individuals while setting out important ethical considerations for educational institutions who are considering OP platforms, including how they might devise appropriate governance frameworks for their use and remain accountable for their decisions. It will also provide a context for various future empirical investigations of OP technologies. 

The essay is structured as follows. The Philosophical Approach section briefly explains the relevance of the central moral values to the OP debate. The Background section provides relevant context concerning exam invigilation and outlines central technological capabilities of popular OP programs. The Discussion section examines important ethical issues raised by the emergence of OP software. Finally, the Conclusion summarizes the ethical lessons for educational institutions and others.

'Assignment outsourcing: moving beyond contract cheating' by Rebecca Awdry in (2020) Assessment and Evaluation in Higher Education states 

The extent and reach of commercial cheating opportunities is ever present; thousands of websites promote differing business models offering assignments in multiple languages and currencies. In addition to commercial companies, students are known to outsource their assignments from friends and family. Assignment outsourcing and contract cheating are not a new problem, yet research to-date has been conducted utilising different survey tools and in different locations, thereby alluding accurate comparisons. This paper reports on a large international research project which utilised the same survey tool across multiple countries. Respondents most commonly reported outsourcing assignments from friends and family, and peer-sharing sites, as compared to essay mills. Large differences were found in the self-reported outsourcing behaviours between countries. Due to the differing outsourcing methods used, a new definition is offered for these behaviours: assignment outsourcing.

Awdry comments 

The problem of students cheating at university by getting others to complete their assignments for them is something long researched in international literature (Bowers 1964; Stavisky 1973; Haines et al. 1986; McCabe 2005; Hughes and McCabe 2006a, 2006b). In recent years this has been referred to as contract cheating (Clarke and Lancaster 2006). Studies have looked at the motivating factors for cheating (Underwood and Szabo 2003; Devlin and Gray 2007; Beasley 2014; Rigby et al. 2015; Brimble 2016), the prevalence and extent of the problem (Clarke and Lancaster 2006; Curtis and Clare 2017; Bretag et al. 2018; Newton 2018; Curtis and Tremayne 2019) as well as demographic and predictor variables (McCabe and Trevino 1997; Stone, Jawahar, and Kisamore 2010; Ives et al. 2017). Contract cheating is not a new phenomenon, although only within the last decade or so has research been done which focusses on this type of cheating and the businesses offering it (for example, Harris and Srinivasan 2012; Lancaster and Clarke 2016; Bretag et al. 2018; Ellis, Zucker, and Randall 2018; Harper et al. 2018; Lancaster 2019; Medway, Roper, and Gillooly 2018; Rowland et al. 2018; Amigud 2019). Prior to this, research considered the wider terms cheating or plagiarism, and surveys may have included a question or two on outsourcing (for examples see Bowers 1964; McCabe 2005; Hughes and McCabe 2006a). 

Students can outsource their assignments in a variety of ways, including essay mills, bespoke assignment services, essay bidding services, peer-to-peer file sharing sites (peer-sharing sites), and obtaining work from other students, colleagues, friends and family members. Bespoke, or ‘contract’, sites allow users to request a piece of work written to their specifications and to their timeframes; essay mills more commonly have pre-written assignments and users can search for their topic and purchase/download instantly. Whilst these sites usually provide services for a monetary fee, users may also gain ‘loyalty credits’ and can obtain assignments for free, or at reduced rates. Advertisements for these types of sites regularly appear in students’ social media feeds, Google searches, and on campus through unauthorised campaigns. The marketing and promotional methods used by the sites are advanced and sometimes predatory (see Medway, Roper, and Gillooly 2018; Rowland et al. 2018). Indeed, so advanced is the extent of these sites that there are review websites and blogs advising students which service would best meet their needs, for example, http://topaperwritingservices.com. 

Peer-sharing sites are increasingly popular with student communities, often offering services for low or no cost. Sites allow users to upload their own documents in exchange for other materials which gets them credit towards a download. Available documents range from lecture notes and presentations to completed assignments (Rogerson 2014; Rogerson and Basanta 2016). Bidding sites, although offering a complete spectrum of tasks and skills for sale (from cleaning to construction), have been used by students to request others complete their work. The business model allows users to upload the requirements of what they are looking for, and respondents bid to complete the work, offering different prices and completion times; users then select the bid most preferential to what they require. Some sites are solely for academic work, for example, http://bid4papers.com (Lancaster and Clarke 2007; 2016). 

Aside from the organised methods and companies, a common method through which students cheat at university is by getting a friend, family member or other student to complete their work for them (Đogaš et al. 2014; Bretag et al. 2018). Advice by the UK Quality Assurance Agency (QAA) states that ‘third-parties’ offering assignments to students may include websites, peer-sharing sites or individuals such as a friend, relative or colleague; obtaining an assignment might not always involve a monetary exchange, and may be free or ‘for favours’ (Quality Assurance Agency 2017). 

Whilst the term contract cheating may be suitable when describing some behaviours (such as bespoke essay production or other forms of contracting out work), it does not necessarily accurately describe other methods of obtaining work, for example from friends or family. Anecdotally, getting ‘help’ on assignments from a family member or other student is not something which students deem to be contract cheating, as no contract is involved. The original definition of contract cheating posed by Clarke and Lancaster (2006), was based upon students’ use of assignment bidding sites. The definition has been altered in response to increased awareness in these types of behaviours from students globally. However, although Lancaster’s current definition is broader, it appears counter to the term itself:

Contract cheating describes the process through which students can have original work produced for them, which they can then submit as if this were their own work. Often this involves the payment of a fee and this can be facilitated using online auction sites. (http://thomaslancaster.co.uk/contract-cheating, October 2019)

Specifically these behaviours may not be based upon a contract and the use of the word ‘can’ in the second line demonstrates that the student may not submit the work for credit, which would therefore not denote cheating. Additionally, it appears counter intuitive to alter the meaning behind the term, and rather retain it for those behaviours which can be considered contract cheating (essay mills, bespoke and bidding sites). Student cheating behaviours are varied and broad, hence a more encompassing term to represent these complexities is proposed (and used throughout this paper): assignment outsourcing. 

Most data on university cheating are limited to a disciplinary, institutional or country focus and do not present a global context. Whilst a local context may be useful for understanding some student behaviours, due to the international nature of websites (where domain names, IP addresses, local contact numbers, and writers may all be situated in different countries [Sivasubramaniam, Kostelidou, and Ramachandran 2016; Ellis, Zucker, and Randall 2018]), an international survey of students may provide global insights to understand cheating from the users’ perspectives. xx Further, as Ives et al. (2017) noted, much less work has been done in regard to surveying students on attitudes about academic dishonesty and reasons for engaging in these behaviours in South America, Eastern Europe and Africa than in Australia, the UK and North America; and specifically on outsourcing behaviours. When considering student understandings and engagements with cheating and outsourcing, and policy approaches across Europe, large differences were seen between countries in their attitudes towards cheating (Glendinning 2014, 2016; Council of Europe 2017). Moreover, as students can undertake part of their course in another country or language, contextualisation between different educational settings and common local outsourcing methods could provide useful insights for study tours and collaborations between international institutions (Glendinning 2016). 

Data currently available on contract cheating, from different survey tools, definitions and behaviours, have found variance in reported rates of outsourcing. For example, Newton (2018) reported an average self-reported contract cheating rate of 3.52% based on historical analyses of studies undertaken internationally; Bretag et al. (2018) reported 5.78% for a range of contract cheating behaviours in Australia (2.2% obtained assignments); 30.6% was reported in Romania (Ives et al. 2017) and 8% in the Czech Republic (Foltýnek and Králíková 2018). Surveys used to explore the topic differ between studies and countries; surveys/research question are not consistent or similar; studies are located in one discipline, university or country. This prevents any direct or longitudinal data comparisons, and evaluations between research studies remain a challenge. 

Finally, most research has considered outsourcing as one category, having combined the different methods or modes through which students obtain the work. By exploring assignment outsourcing within only one paradigm, the methods, modes and purposes are not clearly differentiated, nor compared, which is a significant gap in existing research. Without having a far deeper understanding of the ways in which students outsource their assignments, pro- and re-active strategies may be ineffective. A strategy to reduce students’ temptations to buy an assignment from a contract cheating service may be entirely ineffective for promoting honesty and encouraging students to complete their own work, and not provide assignments to peers.

Workforce and Robotics

The new MIT The Work of the Future: Building Better Jobs in an Age of Intelligent Machines report by David Autor, David Mindell and Elisabeth Reynolds comments 

Three years ago, robots, artificial intelligence (AI), and self-driving cars seemed to be coming fast. A widely cited study projected nearly half of all jobs in industrialized countries could soon be performed by robots or AI. A New Yorker cover published in late 2017 showed robots striding to work on a sidewalk where a disheveled human panhandler begged for coins. During the 2019 Super Bowl, six TV commercials featured robots or AI-enabled assistants. One beer advertisement showed robots gleefully surpassing humans in running, bicycling, and golfing, but ended with a robot gazing wistfully through a window at people socializing in a bar. Humans would soon be outcompeted in every arena except social drinking, this ad seemed to say. 

In this context, MIT President L. Rafael Reif commissioned the MIT Task Force on the Work of the Future in the spring of 2018. He tasked us with understanding the relationships between emerging technologies and work, to help shape public discourse around realistic expectations of technology, and to explore strategies to enable a future of shared prosperity. The Task Force is co-chaired by this report’s authors: Professors David Autor and David Mindell and executive director Dr. Elisabeth Reynolds. Its members include more than 20 faculty members drawn from 12 departments at MIT, as well as over 20 graduate students. 

In the two-and-a-half years since the Task Force set to work, autonomous vehicles, robotics, and AI have advanced remarkably. But the world has not been turned on its head by automation, nor has the labor market. Despite massive private investment, technology deadlines have been pushed back, part of a normal evolution as breathless promises turn into pilot trials, business plans, and early deployments — the diligent, if prosaic, work of making real technologies work in real settings to meet the demands of hard-nosed customers and managers. 

Yet, if our research did not confirm the dystopian vision of robots ushering workers off of factory floors or artificial intelligence rendering superfluous human expertise and judgment, it did uncover something equally pernicious: Amidst a technological ecosystem delivering rising productivity, and an economy generating plenty of jobs (at least until the COVID-19 crisis), we found a labor market in which the fruits are so unequally distributed, so skewed towards the top, that the majority of workers have tasted only a tiny morsel of a vast harvest. 

Four decades ago, for most U.S. workers, the trajectory of productivity growth diverged from the trajectory of wage growth. This decoupling had baleful economic and social consequences: low-paid, insecure jobs held by non-college workers; low participation rates in the labor force; weak upward mobility across generations; and festering earnings and employment disparities among races that have not substantially improved in decades. While new technologies have contributed to these poor results, these outcomes were not an inevitable consequence of technological change, nor of globalization, nor of market forces. Similar pressures from digitalization and globalization affected most industrialized countries, and yet their labor markets fared better. 

History and economics show no intrinsic conflict among technological change, full employment, and rising earnings. The dynamic interplay among task automation, innovation, and new work creation, while always disruptive, is a primary wellspring of rising productivity. Innovation improves the quantity, quality, and variety of work that a worker can accomplish in a given time. This rising productivity, in turn, enables improving living standards and the flourishing of human endeavors. Indeed, in what should be a virtuous cycle, rising productivity provides society with the resources to invest in those whose livelihoods are disrupted by the changing structure of work. 

Where innovation fails to drive opportunity, however, it generates a palpable fear of the future: the suspicion that technological progress will make the country wealthier while threatening livelihoods of many. This fear exacts a high price: political and regional divisions, distrust of institutions, and mistrust of innovation itself. 

The last four decades of economic history give credence to that fear. The central challenge ahead, indeed the work of the future, is to advance labor market opportunity to meet, complement, and shape technological innovations. This drive will require innovating in our labor market institutions by modernizing the laws, policies, norms, organizations, and enterprises that set the “rules of the game.” 

As this report documents, the labor market impacts of technologies like AI and robotics are taking years to unfold. But we have no time to spare in preparing for them. If those technologies deploy into the labor institutions of today, which were designed for the last century, we will see similar effects to recent decades: downward pressure on wages, skills, and benefits, and an increasingly bifurcated labor market. This report, and the MIT Work of the Future Task Force, suggest a better alternative: building a future for work that harvests the dividends of rapidly advancing automation and ever-more powerful computers to deliver opportunity and economic security for workers. To channel the rising productivity stemming from technological innovations into broadly shared gains, we must foster institutional innovations that complement technological change. 

We are living in a period of significant disruption, but not of the kind envisioned in 2018. The final phases of researching and writing this document occurred during the 2020 months of COVID-19. Our technologies have been instrumental in enabling us to adapt via telepresence, online services, remote schooling, and telemedicine. While they don’t look anything like robots, these remote work tools too are forms of automation, displacing vulnerable workers from low-pay service jobs in industries like food service, cleaning, and hospitality. 

We face a labor market crisis stemming from the COVID-19 pandemic. Millions are unemployed. But technological advances had little to do with this crisis. Long before this disruption, our research on the work of the future made it clear how many in our country are failing to thrive in a labor market that generates plenty of jobs but little economic security. The effects of the pandemic have made it even more viscerally and publicly clear: Despite their official designation as “essential,” most low-paid workers cannot effectively do their jobs through computing platforms. Most must be physically present to earn their livings. Some see robots taking over those roles (though few have yet). Others see the indispensable role of human flexibility as people have been essential to transforming supply chains. Still others see COVID-19 as an automation-forcing event. However it plays out, we will be living with the effects of COVID-19 on technology and work for a long time, though those effects will look different from what anyone had anticipated in 2018. 

Other forces have also roiled the 2018 visions of the future, including the rupture between the world’s two largest economies and a surge of political turmoil and economic populism. These pressures are reshaping alliances, breaking apart and reorganizing global business relationships, and even altering patterns in human migration. The United States and China had friction before, but nothing like the fracture that is now occurring. What began as a trade war has morphed into a technology war. 

This clash is filtering out through the economy and threatens to hinder innovation, which increasingly emerges from countries around the world, often by researchers who are collaborating across borders and time zones. How can we make sure that technological advances, whenever they come, yield prosperity that is widely shared? How can the U.S. and its workers continue to play a leading role in inventing and shaping the technologies and reaping the benefits? 

Following two years of study, data collection, and analysis, the Task Force draws the following conclusions: 

1. Technological change is simultaneously replacing existing work and creating new work. It is not eliminating work altogether. 

No compelling historical or contemporary evidence suggests that technological advances are driving us toward a jobless future. On the contrary, we anticipate that in the next two decades, industrialized countries will have more job openings than workers to fill them, and that robotics and automation will play an increasingly crucial role in closing these gaps. Nevertheless, the impact of robotics and automation on workers will not be benign. These technologies, in concert with economic incentives, policy choices, and institutional forces, will alter the set of jobs available and the skills they demand. This process is both challenging and indispensable. Inventing new ways of accomplishing existing work, new business models, and entirely new industries drives rising productivity and new jobs. Such innovations bring new occupations to life, generate demands for new forms of expertise, and create opportunities for rewarding work. Most of today’s jobs hadn’t even been invented in 1940. The United States needs not less, but more technological innovation to meet humanity’s most pressing problems, including climate change, disease, poverty, malnutrition, and inadequate education. Meeting these challenges through investment and innovation will create opportunity and improve well-being. 

2. Momentous impacts of technological change are unfolding gradually. 

Spectacular advances in computing and communications, robotics, AI, and manufacturing processes are reshaping industries as diverse as insurance, retail, healthcare, manufacturing, and logistics and transportation. But we observe substantial time lags, often on the scale of decades, from the birth of an invention to its broad commercialization, assimilation into business processes, widespread adoption, and impacts on the workforce. We find examples of this incremental pace of change in the adoption of novel industrial robots in small and medium-sized firms, and in the still-imminent large-scale deployments of autonomous vehicles. Indeed, the most profound labor market effects of new technology that we found were less due to robotics and AI than to the continuing diffusion of decades-old (though much improved) technologies of the internet, mobile and cloud computing, and mobile phones. This time scale of change provides the opportunity to craft policies, develop skills, and foment investments to constructively shape the trajectory of change toward the greatest social and economic benefit. 

3. Rising labor productivity has not translated into broad increases in incomes because labor market institutions and policies have fallen into disrepair. 

Peer nations from Sweden to Germany to Canada have faced the same economic, technological, and global forces as the United States, and have enjoyed equally strong economic growth, but have delivered better results for their workers. What sets the United States apart are U.S.-specific institutional changes and policy choices that failed to blunt, and in some cases magnified, the consequences of these pressures on the U.S. labor market. 

The U.S. has allowed traditional channels of worker voice to atrophy without fostering new institutions or buttressing existing ones. It has permitted the federal minimum wage to recede to near-irrelevance, lowering the floor under the labor market for low-paid workers. It has embraced a policy-driven expansion of free trade with the developing world, Mexico and China in particular, yet failed to direct the gains towards redressing the employment losses and retraining needs of workers. 

No evidence suggests that this strategy has paid off for the United States. U.S. leadership in growth and innovation is longstanding: It led the world throughout the 20th century, and led even more definitively in the several decades immediately after World War II. Conversely, the labor market maladies documented here are recent. Nothing suggests that these failures inevitably follow from innovation or constitute costs worth paying to gain the other economic benefits that they ostensibly deliver. We can do better. 

4. Improving the quality of jobs requires innovation in labor market institutions. 

In the absence of deliberate policy, good jobs are under-supplied by markets and yet have broad social and political benefits, especially in a democracy. Work is a crucial human good. “Not simply a source of income,” Task Force Research Advisory Board member Josh Cohen writes in a MIT Work of the Future research brief, “work is a way that we can learn, exercise our powers of perception, imagination, and judgement, collaborate socially, and make constructive social contributions.” Even when work is solely a means of acquiring an income, it should offer a sense of purpose and not require submission to demeaning or arbitrary authority, unhealthy or unsafe conditions, or physical or mental degradation. The U.S. must innovate to rebalance the desire of employers for low-cost, minimal commitment, and maximal flexibility, with the necessity that workers receive fair treatment, reasonable compensation, and a measure of economic security. The U.S. must craft and enforce fair labor standards, ensure effective collective bargaining, set a well-calibrated federal minimum wage, extend the scope and flexibility of its unemployment insurance system, and modernize its dysfunctional system of employer-based health insurance. 

To channel the rising productivity stemming from technological innovations into broadly shared gains, we must foster institutional innovations that complement technological change. 

5. Fostering opportunity and economic mobility necessitates cultivating and refreshing worker skills. Enabling workers to remain productive in a continuously evolving workplace requires empowering them with excellent skills programs at all stages of life: in primary and secondary schools, in vocational and college programs, and in ongoing adult training programs. The distinctive U.S. system for worker training has many shortcomings, but it also has unique virtues, for example, offering numerous points of entry for workers who may want to reshape their career paths or need to find new work after a layoff. The U.S. must invest in existing educational and training institutions and innovate to create new training modes to make ongoing skills development accessible, engaging, and cost-effective. 

6. Investing in innovation will drive new job creation, speed growth, and meet rising competitive challenges. 

Investments in innovation grow the economic pie, which is crucial to meeting challenges posed by a globalized and fiercely technologically competitive world economy. Throughout our studies, we found technologies that were direct results of U.S. federal investment in research and development over the past century and longer: the internet, advanced semiconductors, artificial intelligence, robotics, and autonomous vehicles, to name but a few. These new goods and services generate new industries and occupations that demand new skills and offer new earnings opportunities. The U.S. has a stellar record of supporting innovations that inventors, entrepreneurs, and creative capital deploy to support and create new businesses. We must foster and grow the U.S. innovation system to ensure that when workers are displaced by technological change, they can move to new jobs in new industries. Simultaneously, we can shape the direction of innovation through public investment and policy to maximize these benefits. 

Adopting new technology creates winners and losers and will continue to do so. Involvement of all stakeholders — including workers, businesses, investors, educational and social organizations, and government — can minimize the damage done to individuals and communities and help ensure that the jobs of the future offer benefits that are shared by all. We explore this inclusive approach by examining the institutional frameworks around work, including how education and training programs can be made more effective and inclusive, as well as new ways of empowering workers who may never have the protections afforded by traditional union structures. 

This report begins by documenting and diagnosing the challenges facing the U.S. labor market. Next, we survey the technological frontier to draw lessons about the pace and direction of change and its likely impacts on employment, skill demands, and opportunity. Finally, we synthesize insights from work and technology to consider how our policies and institutions should innovate to leverage technological and economic opportunities while surmounting the substantial challenges that lie ahead.

Afterlives

'The Posthumous Privacy Paradox: Privacy preferences and Behavior Regarding Digital Remains' by Tal Morse and Michael Birnhack in New Media & Society (forthcoming 2021) comments 

Scholars have observed a gap between users’ stated preferences to protect their privacy and their actual behavior. This is the privacy paradox. This article queries the persistence of the privacy paradox after death. A survey of a representative sample of Israeli Internet users inquired of perceptions, preferences, and actions taken by users regarding their digital remains. The analysis yielded three distinct groups: (1) users interested in preserving privacy posthumously, but do not act accordingly; for these users, the privacy paradox persists posthumously; (2) users who match their behavior to their preferences; for these users, the privacy paradox is resolved; (3) users interested in sharing their personal data posthumously, but do not make the appropriate provisions. This scenario is the inverted privacy paradox. This new category has yet to be addressed in the literature. We present some explanations for the persistence of the posthumous privacy paradox and for the inverted privacy paradox.

'What If Your Parrot Outlives You? Preparing for Your Bird’s Future' by Gerry W Beyer, a useful US resource for Australianstudents undertaking comparative study, comments 

 Dogs, cats, parrots, and other pet animals play significant roles in the lives of many individuals. The bond between a pet owner and his or her companion is strong. It is of vital importance to include pets when a pet owner makes plans for disability and death. This article provides an overview of the techniques a pet owner should consider when planning his or her estate with emphasis on parrots.

Indigenous Data Sovereignty

'Indigenous Data Sovereignty in the era of Big Data and Open Data' by Maggie Walter, Raymond Lovett, Bobby Maher, Bhiamie Williamson, Jacob Prehn, Gawaian Bodkin‐Andrews and Vanessa Lee in (2020) Australian Journal of Social Issues comments 

Indigenous Data Sovereignty, in its proclamation of the right of Indigenous peoples to govern the collection, ownership, and application of data, recognises data as a cultural and economic asset. The impact of data are magnified by the emergence of Big Data and the associated impetus to open publicly held data (Open Data). Aboriginal and Torres Strait Islander peoples, families and communities, heavily overrepresented in social disadvantage related data will also be overrepresented in the application of these new technologies. But, in a data landscape Indigenous peoples remain largely alienated from the use of data and its utilization within the channels of policy power. Existing data infrastructure, and the emerging Open Data infrastructure, neither recognise Indigenous agency, worldviews nor consider Indigenous data needs. This is demonstrated in the absence of any consideration of Indigenous data issues Open Data discussions and publication. So, while the potential benefits of this data revolution are trumpeted, our marginalised social, cultural and political location suggest we will not share equally in these benefits. This paper discusses the unforeseen (and likely unseen) consequences of the influence Open data and Big Data and discusses how Indigenous Data Sovereignty can mediate risks while providing pathways to collective benefits.

Internet of Bodies

The Internet of Bodies: Opportunities, Risks, and Governance (RAND, 2020) by Mary Lee, Benjamin Boudreaux, Ritika Chaturvedi, Sasha Romanosky, Bryce Downing comments 

A wide variety of internet-connected “smart” devices now promise consumers and businesses improved performance, convenience, efficiency, and fun. Within this broader Internet of Things (IoT) lies a growing industry of devices that monitor the human body, collect health and other personal information, and transmit that data over the internet. We refer to these emerging technologies and the data they collect as the Internet of Bodies (IoB) (see, for example, Neal, 2014; Lee, 2018), a term first applied to law and policy in 2016 by law and engineering professor Andrea M. Matwyshyn (Atlantic Council, 2017; Matwyshyn, 2016; Matwyshyn, 2018; Matawyshyn, 2019).  

IoB devices come in many forms. Some are already in wide use, such as wristwatch fitness monitors or pacemakers that transmit data about a patient’s heart directly to a cardiologist. Other products that are under development or newly on the market may be less familiar, such as ingestible products that collect and send information on a person’s gut, microchip implants, brain stimulation devices, and internet-connected toilets. 

These devices have intimate access to the body and collect vast quantities of personal biometric data. IoB device makers promise to deliver substantial health and other benefits but also pose serious risks, including risks of hacking, privacy infringements, or malfunction. Some devices, such as a reliable artificial pancreas for diabetics, could revolutionize the treatment of disease, while others could merely inflate health-care costs with little positive effect on outcomes. Access to huge torrents of live-streaming biometric data might trigger breakthroughs in medical knowledge or behavioral understanding. It might increase health outcome disparities, where only people with financial means have access to any of these benefits. Or it might enable a surveillance state of unprecedented intrusion and consequence. There is no universally accepted definition of the IoB. For the purposes of this report, we refer to the IoB, or the IoB ecosystem, as IoB devices (defined next, with further explanation in the passages that follow) together with the software they contain and the data they collect. 

An IoB device is defined as a device that

• contains software or computing capabilities 

• can communicate with an internet-connected device or network and satisfies one or both of the following: 

• collects person-generated health or biometric data 

• can alter the human body’s function. 

The software or computing capabilities in an IoB device may be as simple as a few lines of code used to configure a radio frequency identification (RFID) microchip implant, or as complex as a computer that processes artificial intelligence (AI) and machine learning algorithms. A connection to the internet through cellular or Wi-Fi networks is required but need not be a direct connection. For example, a device may be connected via Bluetooth to a smartphone or USB device that communicates with an internet-connected computer. Person-generated health data (PGHD) refers to health, clinical, or wellness data collected by technologies to be recorded or analyzed by the user or another person. Biometric or behavioral data refers to measurements of unique physical or behavioral properties about a person. Finally, an alteration to the body’s function refers to an augmentation or modification of how the user’s body performs, such as a change in cognitive enhancement and memory improvement provided by a brain-computer interface, or the ability to record whatever the user sees through an intraocular lens with a camera. 

IoB devices generally, but not always, require a physical connection to the body (e.g., they are worn, ingested, implanted, or otherwise attached to or embedded in the body, temporarily or permanently). Many IoB devices are medical devices regulated by the U.S. Food and Drug Administration (FDA). Figure 1 depicts examples of technologies in the IoB ecosystem that are either already available on the U.S. market or are under development. 

Devices that are not connected to the internet, such as ordinary heart monitors or medical ID brace- lets, are not included in the definition of IoB. Nor are implanted magnets (a niche consumer product used by those in the so-called bodyhacker community, described in the next section) that are not connected to smartphone applications (apps), because although they change the body’s functionality by allowing the user to sense electromagnetic vibrations, the devices do not contain software. Trends in IoB technologies and additional examples are further discussed in the next section. 

Some IoB devices may fall in and out of our definition at different times. For example, a Wi-Fi-connected smartphone on its own would not be part of the IoB; however, once a health app is installed that requires connection to the body to track user information, such as heart rate or number of steps taken, the phone would be considered IoB. Our definition is meant to capture rapidly evolving technologies that have the potential to bring about the various risks and benefits that are discussed in this report. We focused on analyzing existing and emerging IoB technologies that appear to have the potential to improve health and medical outcomes, efficiency, and human function or performance, but that could also endanger users’ legal, ethical, and privacy rights or present personal or national security risks. 

For this research, we conducted an extensive literature review and interviewed security experts, technology developers, and IoB advocates to under- stand anticipated risks and benefits. We had valuable discussions with experts at BDYHAX 2019, an annual convention for bodyhackers, in February 2019, and DEFCON 27, one of the world’s largest hacker conferences, in August 2019. In this report, we discuss trends in the technology landscape and outline the benefits and risks to the user and other stakeholders. We present the current state of gover- nance that applies to IoB devices and the data they collect and conclude by offering recommendations for improved regulation to best balance those risks and rewards.

Contact Tracing and Pandemics

The expert National Contact Tracing Review: A report for Australia’s National Cabinet November 2020 comments 

 We were tasked by National Cabinet with reviewing COVID-19 contact tracing and outbreak management systems in each state and territory to determine their ability to support an active economy by Christmas 2020. This includes systems for testing and tracing, quarantine and isolation, outbreak management, data exchange, and surge capacity. Although our remit was the current COVID-19 pandemic, we note that most of our recommendations may be relevant to managing future pandemics caused by other infectious diseases. As we visited each jurisdiction in October 2020, it became clear to us that internal borders will only reopen and remain open if state and territory leaders have confidence in how their interstate counterparts are managing the pandemic. By the same token, the economy will only bounce back if Australians feel confident they can participate and travel safely. Many of our recommendations are aimed at building this confidence and ensuring it is well founded. 

The states and territories have decision making authority for public health and will remain responsible for their own contact tracing and outbreak management systems. Our review acknowledges this autonomy while identifying areas where changes to processes, information sharing and technology will improve national capability. 

Our remit was contact tracing and outbreak management. These systems must perform extremely well if we are to successfully live with COVID-19 until a vaccine or an effective therapeutic arrives, and perhaps longer. However, contact tracing and outbreak management are necessary but not sufficient components of an overall response and they are measures we would prefer never to have to activate. Crucially important in the first line of defence are measures relating to physical distancing, personal hygiene, staying away from work and gatherings if unwell, testing if symptomatic, mask wearing where required, limiting access to vulnerable communities where appropriate, COVID Safety Plans, attendance limits at public events, and quarantine for international travellers and others at risk of having been exposed. 

Overall, we found very strong commitment to prevention and control measures across the country. All jurisdictions are committed to implementing effective COVID-19 contact tracing and outbreak management systems, have increased their investment and are training and preparing constantly. Across all states and territories the information technology systems used for contact tracing have improved significantly over recent months. 

However, we found processes that can be improved. In some jurisdictions, interviews with contacts are recorded on paper before being entered into a database, causing delays and the potential for error. Contact information is inconsistently collected when people visit venues. Text messages to people with COVID-19 and contacts are not always in the preferred language of the person. Domestic airline passenger lists and contact details are not always accurate. Real time performance metrics are not sufficiently ambitious. 

Our report sets out the characteristics of an optimal contact tracing and outbreak management system, and invites every jurisdiction to evaluate its performance against this blueprint. 

We also recommend clear, measurable and transparent metrics that should be published by each state and territory to allow the public to track performance. 

The two key performance metrics we recommend relate to fast testing and fast notifications to contacts. The currently agreed national target of 48 hours from reporting a positive test result to directing close contacts to quarantine is inadequate from the point of view of suppressing community transmission. 

We recommend that test results should be available within 24 hours of a sample being taken, maximising the likelihood that people will isolate themselves while awaiting test results. We recognise this may be difficult in remote parts of Australia, but it is an important stretch goal and confirmation of our national capability. 

Further, we recommend no more than 48 hours in total from the time a test sample is first taken to the point at which close contacts of a confirmed case are notified to quarantine. Advice to us is that if this turnaround time is achieved, we can substantially reduce community transmission. 

Across the jurisdictions we discovered quite different digital solutions for case management and contact tracing, developed in isolation. In some instances, the digital systems are built on similar underlying platforms, but they are heavily configured and require different training for users. 

However, the panel does not recommend the creation of a single integrated national contact tracing system. The important thing is that information is shared efficiently, where necessary. States and territories must be able to access and transfer information about cases and contacts where people have crossed borders. Currently, such information is conveyed through phone calls or emails, a practice that would not withstand high case numbers. 

For this reason, we recommend the development of a digital data exchange mechanism. Building this capability now would prepare the states and territories for coordinated contact tracing to more effectively manage future outbreaks. 

The mechanism we suggest would allow the states and territories to share contact tracing data, and incorporate contact tracing data from sources such as airline and shipping passenger manifests, registries of test results and relevant government agency data stores. Only data relevant to contact tracing would be transferred, such as phone numbers, addresses, case interviews and diagnostic test results. No data would be held or stored in the data exchange. As such, we are confident the data exchange can be consistent with privacy requirements and community expectations. We make a number of recommendations to improve the use of technology. 

In that context, we recommend that the states and territories share information about new and emerging technologies, such as electronic venue and workplace attendance registration systems, smartphone apps to monitor self-quarantine, new diagnostic tests and wastewater surveillance. For example, the venue attendance app used in the ACT is as simple as “click and enter”, the only information shared is an email or phone number, no information is used for marketing and data are purged every 28 days. 

Patient testing, contact tracing and case management should be fully digital end to end, starting at the point of testing. This includes collection of information, reporting of results, contact tracing, case management and outbreak management. 

However, while a fully digital system dramatically improves the efficiency of contact tracing, it will never replace the need for well trained contact tracers and expert public health oversight, especially for difficult interviews, cluster analysis and outbreak responses. All states and territories should employ a permanent workforce for tracing and outbreak management, with senior public health leadership, and should have an additional surge workforce trained and at the ready. Digital case management and contact tracing systems should allow easy and secure onboarding of contact tracers from other states and territories and from the Commonwealth. 

In the event of an outbreak, every effort should be made to go hard and go early. The driving principle for contact tracing must be to never fall behind, which means operating procedures should allow a risk based prioritisation of contact tracing practices that if the surge workforce becomes overwhelmed. These would include, for example, initial notification of close contacts by text instead of by phone. 

Desktop exercises and field rehearsals should be run regularly to ensure the system can deal with a sustained surge of around four new cases per day per million population and be able to rapidly scale up should there be a further escalation. 

As Australia takes steps to reopen, we emphasise that a national testing and contact tracing system is only as good as its weakest link. No jurisdiction can afford to let down its guard. Each must have a strong focus on continuous improvement, including regular stress testing, a highly trained workforce, high functioning technology, and a commitment to transparency on performance metrics. We must keep awareness high and the safety message front and centre if we are to avoid the complacency that can be a dangerous companion to low case numbers. 

COVID-19 remains a complex and highly communicable disease. Even with the best systems in place, outbreaks are likely to be unavoidable. We are acutely aware of the lockdowns being imposed once again in many countries as the world struggles to find a way to live with the pandemic. 

However, we believe that Australia’s internal borders and economy can safely, confidently and successfully reopen, and the nation can manage an early cluster or outbreak and a moderate number of confirmed cases in the community without resorting to wide area lockdowns. To ensure this, each state and territory needs to be well aligned to the characteristics of an optimal contact tracing and outbreak management system as outlined in this report, alongside important measures to prevent transmission.

Recommendations in the report are - 

 The bigger picture 

 

1. Continuous improvement 

 

1.1 All jurisdictions should aspire to continuous improvement and reflect upon, evaluate and externally communicate their performance against the list of ‘Characteristics of an Optimal Contact Tracing and Outbreak Management System’. 

 

2. Preventative public health measures 

 

2.1 Maintain the focus on preventative public health measures, including those that were agreed by National Cabinet early in the course of the pandemic.   

 

Constant preparation 

 

3. Workforce and training 

 

3.1 Ensure ongoing investment in the medium to long term in accredited training programs for applied epidemiology and applied public health training. 

 

3.2 The Commonwealth, states and territories should consider increasing the number of public health training positions in all jurisdictions. 

 

3.3 All states and territories should continually invest in training surge workforces to be employed in a reserve capacity. 

 

3.4 Ensure there is capacity for the Commonwealth to mobilise a trained contact tracing surge workforce through the Australian Public Service to assist states and territories with contact tracing should the need arise. 

 

3.5 Continue funding rapid deployment capability to coordinate a standby pool of equipment (including personal protective equipment and transportable laboratory equipment) and senior clinical and public health experts for extreme situations requiring surge capacity anywhere in Australia. 

 

3.6 Undertake forward planning for the pathology laboratory workforce, given the ongoing requirement for high volume testing in the near and medium term. 

 

4. Stress tests 

 

4.1 States and territories should undertake desktop and functional simulation exercises to verify the performance of their contact tracing and outbreak management systems. 

 

4.2 Desktop and functional simulation exercises should be based on four new confirmed cases (not in quarantine) per day per million population (but no fewer than four per day per jurisdiction) for a week or more. This daily case number is consistent with the Framework for National Reopening adopted by National Cabinet. 

 

4.3 Extreme stress testing should be based on up to ten times the standard stress testing numbers. 

 

End to end contact tracing 

 

5. Never fall behind 

 

5.1 An effective contact tracing and case management system will cope with high case numbers. In extreme conditions, the jurisdiction should in the first instance recruit workforce assistance from other jurisdictions and the Commonwealth. If this proves insufficient, it is nevertheless essential to keep up with managing new cases. In order to never fall behind, the extent of contact tracing measures should be reduced on a risk minimisation basis. 

 

5.2 COVID-19 testing resources and strategies 

 

5.3 Continue to fund COVID-19 pathology tests through the Medicare Benefits Schedule and other funding arrangements. 

 

5.4 Continue to ensure adequate supplies of testing reagents and build stockpiles during quiet times. 

 

5.5 Ensure that pop-up test sites can be rapidly deployed, in under six hours in metropolitan locations, and in under 24 hours in regional locations. 

 

5.6 Pathology laboratories should use diagnostic instruments from multiple vendors to ensure resilience during times of global shortages of reagents. 

 

6. Support for maintaining national standards 

 

6.1 Ensure Commonwealth epidemiological and public health expert support is provided to the Communicable Diseases Network Australia for ongoing work for COVID-19 and other notifiable diseases, including development and maintenance of the Series of National Guidelines. 

 

Data Exchange 

 

7. Technical capability 

 

7.1 Develop a ‘Data Exchange’ capability to facilitate contact tracing, through the exchange of data between states and territories, and access to contact tracing data from relevant government agencies.  

 

7.2 Data should not be stored in the Data Exchange itself, thereby allowing a simplified, decentralised design with high levels of privacy and security. 

 

7.3 The exchange of data would ideally be as near to real time as is practical and consistent with Commonwealth, state and territory security and privacy requirements. 

 

8. Data sources 

 

8.1 The Data Exchange would access a variety of data sources such as appropriate administrative databases, airline and shipping passenger contact tracing information, other relevant government agency databases, contact tracing databases from other states and territories, and COVID-19 diagnostic test result repositories. 

 

8.2 Evaluate the possibility of the Data Exchange in the medium term accessing the Australian Immunisation Register for relevant vaccine status. 

 

8.3 In all instances data requests must be restricted to data that are relevant to a public health response, such as phone numbers, addresses, case interviews and diagnostic test results. 

 

8.4 Domestic airlines should supply accurate passenger contact tracing information on request, accessible by the Data Exchange. Accuracy would be improved by requiring photo ID checks for all domestic passengers. 

 

8.5 Australian Border Force should work with international airlines and shipping companies, supported by bilateral travel agreements, to provide accurate passenger contact tracing information on request, accessible by the Data Exchange. 

 

9. Implementation 

 

9.1 For efficiency in implementing the Data Exchange: – Limit the initial implementation to a pilot involving Victoria, NSW, ACT and the Commonwealth. – Development of the pilot should be based on an indicative scope of technical work developed by these jurisdictions and others that wish to contribute. – Deployment in other jurisdictions would proceed if an evaluation of this pilot implementation concludes that it is successful. 

 

9.2 Implementation of the Data Exchange should not delay any existing plans for sharing of contact tracing data, such as may be provided by airlines and Australian Border Force. 

 

Outbreak management 

 

10. Identify sources 

 

10.1 At low case numbers, epidemiologists and other public health experts should strive to identify the source of infection for all confirmed cases. Where the source of infection is unknown, detailed upstream mapping of contacts to identify the source of infection should be undertaken. 

 

11. Predictive analytics 

 

11.1 Develop, evaluate and share advanced analytics software for outbreak analysis and predicting risks, to support existing expertise. 

 

Technology 

 

12. Pathology test technologies 

 

12.1 Researchers and public pathology laboratories should continue to invest in developing and validating new COVID-19 specimen collection and diagnostic methods. 

 

12.2 A framework should be developed on the role and use of rapid antigen tests, to support the public health response to COVID-19 and enable tracking of all positive and negative test results by public health authorities. 

 

13. Automation and digital support 

 

13.1 Fully digital and partially automated end to end systems should be implemented within each state and territory to support collection of case information, reporting of COVID-19 test results to the health department, allocation of confirmed cases, contact tracing, digitally issued quarantine directions, case management and outbreak management. 

 

14. Attendance registration 

 

14.1 Recording of contact tracing information of attendees should be a condition of entry to restaurants and other public venues, institutions and workplaces. Electronic data collection should be strongly encouraged, with pen and paper only being used if the former is unavailable. 

 

14.2 Where attendance data are recorded for contact tracing, only the minimum information required for that purpose should be collected. Data collected for contact tracing should only be used for contact tracing purposes, kept securely and permanently deleted after 28 days. 

 

14.3 Contact tracing information must be made available to health authorities in a timely manner, at most within 24 hours of request, to assist contact tracing. 

 

14.4 Where smartphone apps are used, they should have simple “click and enter” functionality to encourage compliance. 

 

14.5 To maximise participation, ensure effective communication of the benefits of attendance registration. 

 

14.6 States and territories should consider using a single smartphone app within their jurisdiction, or require that all smartphone apps adhere to the above requirements. 

 

15. Other technology solutions 

 

15.1 Evaluate consent based systems that can download contact tracing information from smartphones. 

 

15.2 The Commonwealth should lead the development of arrangements between states and territories and payment card providers so that contact tracers from the states and territories will be able to request contact details of persons who have made a transaction at a hotspot venue, noting that privacy rules will apply and in some jurisdictions legislative change may be required. 

 

15.3 Develop, use and share proven web portals and smartphone apps for quarantine monitoring and tracking entry into high risk settings, such as residential aged care homes. 

 

16. COVIDSafe app 

 

16.1 The Commonwealth should continue to enhance the functionality of the COVIDSafe proximity app, particularly with respect to the duration for identifying contacts and enhancing notifications to users on the status and operation of the app. 

 

16.2 The Commonwealth should consult with the states and territories on ways to optimise incorporation of COVIDSafe contact information early in the contact tracing process. 

 

16.3 The Commonwealth should consult with the states and territories on the best means to report usage of the app in contact tracing. 

 

17. Wastewater testing 

 

17.1 The public health, clinical and wastewater sectors should build on existing research and field testing of wastewater detection to validate its role as an early signal of potential outbreaks. 

 

17.2 Determine whether a goal of 50% coverage of the Australian population is practical and useful, with appropriate coverage of urban and rural areas. If so, aim to achieve this level of coverage in the medium term. 

 

17.3 States and territories should publish results regularly. 

 

A conversation with communities 

 

18. Involve communication experts early and throughout 

 

18.1 Integrate and embed communications and media experts in health, emergency, police, customer service and other relevant government departments to ensure that public health messages are pitched appropriately for state wide and local audiences, and vulnerable communities. 

 

18.2 Work with community leaders to ensure that public health messages are culturally and linguistically tailored to each community, and understood and amplified through existing formal and informal networks. 

 

19. Avoidance of confusion 

 

19.1 All messages to affected communities, families and individuals should be evaluated to minimise any risk they could be misinterpreted. 

 

19.2 Consistent messages should be given to all individuals in affected families, and consistent guidance provided to leaders and staff in affected settings, such as workplaces, schools, and places of worship. 

 

19.3 Automated text and web messages provided to people in isolation and quarantine should be offered in their preferred language. 

 

Earning community confidence 

 

20. Reporting confirmed cases 

 

20.1 Confirmed cases identified in quarantine are a sign of a well functioning system that is able to mitigate community exposure and transmission. Confirmed cases identified in the community are cases that are more complex and have to be actively traced and managed. States and territories should publicly report daily on: – New confirmed cases identified in the community. If zero cases, the number of days since the last confirmed community case. – New confirmed cases identified in quarantine. If zero cases, the number of days since the last confirmed case. 

 

21. Performance metrics reporting 

 

21.1 The Commonwealth, states and territories should agree and publicly report weekly national performance metrics, including: – The number of hours from collecting the COVID-19 specimen to notifying all people of their results, with the target being fewer than 24 hours at the 90th percentile. – The number of hours from collecting the patient’s COVID-19 specimen to notifying their close contacts that they must quarantine, with the target being fewer than 48 hours at the 90th percentile.

In discussing community communication the report states

Efficient contact tracing and outbreak management are necessary but not sufficient to successfully live with COVID-19. Preventative measures such as attention to personal hygiene, social distancing, early testing at the first sign of symptoms, and voluntary quarantine when symptomatic will continue to be essential components of the first line of defence against COVID-19. 

People have a right to know what is expected of them and how the pandemic and the response is unfolding. Therefore, a substantial commitment to a broad spectrum of public communication activities is required across government. This messaging may need to be strengthened in times of adjustment, particularly when restrictions are tightened. It is important to have communications and media experts integrated and embedded in health, emergency, police, customer service (NSW) and other relevant government agencies to ensure public health messages are consistent and pitched appropriately for Australia wide, state wide and local audiences. 

Ongoing strong, consistent and culturally accessible and appropriate messaging through community engagement is vital to building and maintaining public awareness, trust, acceptance and confidence. Regular and proactive communication and engagement with the public, specifically with at risk populations, can also help alleviate confusion and avoid misunderstandings. 

Education is needed to improve community understanding and health literacy, particularly with regards to infection prevention and control. The community must be encouraged to take personal responsibility and understand the impacts of their behaviour. 

Consistency 

It is important that as we move towards a COVID normal society the public remains vigilant. This is most likely if messages from the Commonwealth, state and territory public health and political leadership are consistent. Currently, there are inconsistencies in the messaging around getting tested between the jurisdictions. For example, the Commonwealth advice is “if you have cold or flu like symptoms you should seek medical advice about having a test for COVID-19”. However, in other jurisdictions the advice is that “anyone with mild COVID-19 symptoms should get tested”. 

This inconsistency could create a barrier to some individuals getting tested, while also delaying the time it takes for an individual to get tested from symptom onset. Similarly, there is inconsistent advice about what people should do while awaiting test results and when they should resume usual activities. 

There are also inconsistencies in key terms used by government and the media, including the use of the terms ‘community transmission’, ‘mystery cases’, ‘physical distancing’ and ‘social distancing’. 

Where possible, states and territories should review their use of messaging around new cases, community transmission and mystery cases. There is currently not an agreed definition and these terms are being used differently in each state and territory. The review has adopted usage of terms to describe confirmed cases as ‘those identified in the community’ and ‘those identified in quarantine’. Further details can be found in Chapter 7 – Earning community confidence. 

The term social distancing has become a commonly used term by governments and media and is used interchangeably with the term physical distancing. However, these terms can have different interpretations. Consistent use of ‘physical distancing of at least 1.5 meters’ would promote community understanding and practice. 

When a person is notified of their COVID-19 test result, usually via a text message, it is important to clearly articulate the result to avoid confusion. Explicitly reporting detection of the virus, or no detection of the virus, in the test sample is preferred. The use of terms such as ‘positive’ and ‘negative’ can be misunderstood as a ‘good result’ or a ‘bad result’. 

As part of contact tracing and case management, it is important consistent information and directions are given to all individuals in isolation and quarantine, including their immediate family and household. As part of an optimal end to end contact tracing and case management system this is optimised by allocating a single case manager to each household. 

Consistency in messaging is vital across all community settings. During an outbreak clear, concise and consistent messages to affected settings such as workplaces, school and places of worship is important. This is optimised through ensuring communications and media experts are integrated and embedded in all government agencies, including health, emergency, police and, customer service (NSW), and ensuring all agencies are collaborating to present one voice. 

In addition to consistency, it is important states and territories are checking to ensure public health messages are understood and not being misinterpreted. For example, South Australia has adapted their weekly state wide population health survey to include questions on understanding and adherence to COVID Safe messages. This includes questions on actions people are taking to protect against COVID-19 and reasons for not getting tested if symptomatic. 

Working with community leaders 

Australia has a diverse population, thus it is important that messaging is tailored for our various community groups, including people from culturally and linguistically diverse backgrounds.  

Inclusion of community leaders in supporting and implementing public health measures is key to an effective response. 

Working with community leaders has proved very important in states and territories with remote Aboriginal and Torres Strait Islander communities and also metropolitan areas with culturally and linguistically diverse communities. During the first wave, Aboriginal and Torres Strait Islander leaders called on governments to provide additional protection to remote communities, which was provided at a Commonwealth level through the Biosecurity Determination limiting travel to remote areas. States and territories with remote communities have invested extensive resources to assist communities and build trust and rapport to support them to protect and then reopen communities safely. 

As an example, Queensland has committed to a co-ownership approach with local Aboriginal and Torres Strait Islander communities, drawing expertise from Aboriginal and Torres Strait Islander Community Controlled Health Organisations, the Queensland Department of Aboriginal and Torres Strait Islander Partnerships and the National Indigenous Australians Agency as well as working directly with mayors and CEOs of discrete and remote Aboriginal and Torres Strait Islander Communities. Through this approach, Queensland have provided support to communities, including targeted testing and targeted scenario planning for the unique circumstances of remote communities.  

It is also important to engage with cultural and religious leaders in metropolitan areas, especially areas with large culturally and linguistically diverse populations. Community leaders in diverse cultural settings can help ensure key messages around physical distancing, hygiene, tightening restrictions and the importance of getting tested. In some communities there is stigma attached to contracting COVID-19 and getting tested. Cultural leaders can assist to break down these barriers. For example, in Melbourne cultural leaders helped to reduce stigma and bolster testing in underrepresented cultural groups. 

Translation of key messages and resources 

Ensuring all Australians can understand key messages around COVID-19 is vitally important in keeping the community safe by ensuring people adhere to public directions. With more than 300 different languages spoken in Australian homes, it is important that key messages are translated into appropriate languages and tailored to communities. 

There are a number of translated materials on Commonwealth, state and territory designated COVID-19 webpages. The number and quality of these translated materials has improved greatly since March 2020, however it is an area that should undergo continued review and refinement. 

It is imperative that translated messages are updated regularly to reflect any changes in the original English messages. 

Further, the accuracy of the translations should be verified by reverse translation back into English to ensure the messages are concise and comprehensible. 

As part of ongoing case management, most states are utilising automated daily text messages to monitor people in isolation and quarantine to remind about their obligations to remain in isolation or quarantine. Messages are generally sent in English, but contact tracing systems built on modern software applications can easily send these messages in preferred languages. 

In addition, collecting information on language spoken at home alongside other contact details and symptom information at the point of COVID-19 specimen collection would enable the automated result notification to be sent in a person’s preferred language. This would help ensure people understand their test result, especially when COVID-19 virus is detected, to ensure the patient understands the need to isolate and await a case interview.

Validity

'Transparency and reproducibility in artificial intelligence' by Benjamin Haibe-Kains, George Alexandru Adam, Ahmed Hosny, Farnoosh Khodakarami, Massive Analysis Quality Control (MAQC) Society Board of Directors, Levi Waldron, Bo Wang, Chris McIntosh, Anna Goldenberg, Anshul Kundaje, Casey S. Greene, Tamara Broderick, Michael M. Hoffman, Jeffrey T. Leek, Keegan Korthauer, Wolfgang Huber, Alvis Brazma, Joelle Pineau, Robert Tibshirani, Trevor Hastie, John P. A. Ioannidis, John Quackenbush & Hugo J. W. L. Aerts in (2020) 586 Nature E14–E16 comments 

 Breakthroughs in artificial intelligence (AI) hold enormous potential as it can automate complex tasks and go even beyond human performance. In their study, McKinney et al. showed the high potential of AI for breast cancer screening. However, the lack of details of the methods and algorithm code undermines its scientific value. Here, we identify obstacles that hinder transparent and reproducible AI research as faced by McKinney et al., and provide solutions to these obstacles with implications for the broader field.

Haibe-Kains et al argue that the work both demonstrates the potential of AI and the challenges of making such work reproducible: 'the absence of sufficiently documented methods and computer code underlying the study effectively undermines its scientific value' and 'limits the evidence required for others to prospectively validate and clinically implement such technologies'. 

 Scientific progress depends on the ability of independent researchers to scrutinize the results of a research study, to reproduce the study’s main results using its materials, and to build on them in future studies (https://www.nature.com/nature-research/editorial-policies/reporting-standards). Publication of insufficiently documented research does not meet the core requirements underlying scientific discovery. Merely textual descriptions of deep-learning models can hide their high level of complexity. Nuances in the computer code may have marked effects on the training and evaluation of results, potentially leading to unintended consequences. Therefore, transparency in the form of the actual computer code used to train a model and arrive at its final set of parameters is essential for research reproducibility. McKinney et al. stated that the code used for training the models has “a large number of dependencies on internal tooling, infrastructure and hardware”, and claimed that the release of the code was therefore not possible. Computational reproducibility is indispensable for high-quality AI applications; more complex methods demand greater transparency. In the absence of code, reproducibility falls back on replicating methods from textual description. Although, McKinney and colleagues claim that all experiments and implementation details were described in sufficient detail in the supplementary methods section of their Article to “support replication with non-proprietary libraries”, key details about their analysis are lacking. Even with extensive description, reproducing complex computational pipelines based purely on text is a subjective and challenging task. 

In addition to the reproducibility challenges inherent to purely textual descriptions of methods, the description by McKinney et al. of the model development as well as data processing and training pipelines lacks crucial details. The definitions of several hyperparameters for the model’s architecture (composed of three networks referred to as the breast, lesion and case models) are missing (Table 1). In their publication, McKinney et al.1 did not disclose the settings for the augmentation pipeline; the transformations used are stochastic and can considerably affect model performance. Details of the training pipeline were also missing. Without this key information, independent reproduction of the training pipeline is not possible. 

Numerous frameworks and platforms exist to make artificial intelligence research more and reproducible (Table 2). For the sharing of code, these include Bitbucket, GitHub and GitLab, among others. The many software dependencies of large-scale machine learning applications require appropriate control of the software environment, which can be achieved through package managers including Conda, as well as container and virtualization systems, including Code Ocean, Gigantum, Colaboratory and Docker. If virtualization of the McKinney et al. internal tooling proved to be difficult, they could have released the computer code and documentation. The authors could have also created small artificial examples or used small public datasets to show how new data must be processed to train the model and generate predictions. Sharing the fitted model (architecture along with learned parameters) should be simple aside from privacy concerns that the model may reveal sensitive information about the set of patients used to train it.

Haibe-Kains notes techniques for achieving differential privacy: many platforms allow sharing of deep learning models that in addition to improving accessibility and transparency can accelerate models  into production and clinical implementation. 

Another crucial aspect of ensuring reproducibility lies in access to the data the models were derived from. In their study, McKinney et al. used two large datasets under license, properly disclosing this limitation in their publication. The sharing of patient health information is highly regulated owing to privacy concerns. Despite these challenges, the sharing of raw data has become more common in biomedical literature, increasing from under 1% in the early 2000s to 20% today. However, if the data cannot be shared, the model predictions and data labels themselves should be released, allowing further statistical analyses. Above all, concerns about data privacy should not be used as a way to distract from the requirement to release code.