A Retrospect on Remote Proctoring
Distance education over the years
Distance education, distance learning, correspondence education, and most recently, remote learning all describe a method of instruction where the teacher and the student need not be in the same room, or even be communicating with each other at the same time. Although remote learning shot to prominence when schools across the globe closed in response to COVID-19, it had already existed for decades, albeit in different forms. In fact, print-based correspondence education was used in Canada as early as 1889 to “provide degree opportunities for rural teachers who were unable to attend McGill University full time.” Today, hundreds of thousands of Canadians study through distance education, ranging from elementary school students to working adults. Distance education is no longer seen as the last choice for individuals who are physically unable to attend in-person classes, but rather a preferred option for many seeking greater autonomy, flexibility, and a more comfortable learning environment.
An unprecedented opportunity
Naturally, the rise of remote learning called for the advent of remote proctoring. Heralded as a tool that could revolutionize educational assessment, remote proctoring was gradually adopted by universities and other educational venues, although it had yet to enter mainstream usage until the onset of COVID-19. As schools grappled with the trade-off between providing a quality education and the risks and dangers of in-person learning, remote learning and proctoring offered a way out. This presented a golden ticket to companies offering remote proctoring services. Proctorio founder Mike Olsen commented, “in 2020, we were like a train going 100 miles an hour, and we couldn’t stop it.” In the wake of a global shift to online learning, these companies found that they could suddenly cater to a more traditional audience. Prior to mass adoption, students being taught through remote learning largely opted in due to an understanding that the instruction method was better suited for their needs than traditional in-person teaching. This may have presented an image of success. However, as students from all walks of life were abruptly thrust into their first encounters with this technology, previously unrecognized issues came to light, suggesting that that initial image may have been a mirage.
The digital divide
The transition to remote proctoring was especially difficult for many students who lacked a fast and reliable internet connection, or simply looked or behaved differently from a proctoring software’s idea of a “normal” student. One dimension along which cracks immediately appeared has been termed the ‘digital divide.’ Remote learning and proctoring require access to internet, and while the UN General Assembly recognizes that as a human right, it is one not yet shared by all. There are differences between households based on income, race, and population density. Although the Government of Canada estimates that 89.7% of Canadian households have access to high-speed internet, this average is comprised of 99.2% of urban households that enjoy such access. Only 54.4% of rural households can make the same claim. University of Minnesota student, Katrina Martin, recalled how she “had to take her first chemistry test in a Starbucks parking lot because she didn’t have the bandwidth at home to accommodate Proctorio.” Variations in internet quality can not only lead to a much more stressful and frustrating experience for exam takers, but it can also lead to students finding themselves unable to submit their exam altogether, or being charged with academic misconduct.
Indifference towards learning differences
Moreover, proctoring software is imperfect, and the imperfections penalize some groups more than others by design. Disabled or neurodivergent students in particular may naturally and non-maliciously behave in ways which are penalized. As lawyer Lydia X. Z. Brown writes for the Center for Democracy & Technology, because atypical eye movements, motor tics, and other physical responses are “naturally occurring characteristics of many types of disabilities, there is no way for algorithmic virtual proctoring software to accommodate disabled students. The point is to identify and flag atypical movement, behaviour, or communication; disabled people are by definition going to move, behave, and communicate in atypical ways.” These movements, which are often already worsened due to exam-related anxiety, can be further exacerbated by uncompromising systems and remote proctors unfamiliar with a student’s learning accommodations. Similarly, neurodivergent students are prone to being unfairly punished for processing information in ways that may be interpreted as suspicious when compared against the average neurotypical student. These differences in learning are more common than one might realize, with over 15% of Canadians over the age of 15 possessing a disability and over 5% of Canadians possessing both a disability and a diagnosed developmental disorder. These proctoring services cause a disproportionate level of harm to such students. Privacy advocates Evan Enzer and Sarah Roth note, “for nearly every one of us, neurodivergence is nothing to be concerned about, but school surveillance technology treats our differences as a threat. Much like the shame we felt when teachers singled us out, it hurts students when surveillance tech targets neurodivergence.”
Failures in facial recognition and data protection
This is before even broaching the myriad of issues with facial recognition software more broadly (i.e., the consequences of training models using data sets underrepresenting women, BIPOC, and other marginalized individuals), as well as the lack of safeguards for student data. In 2020, six Democratic Senators wrote an open letter to ExamSoft CEO Sebastian Vos regarding the former, stating, “as we have seen far too often, students have run head-on into the shortcomings of these technologies—shortcomings that fall heavily on vulnerable communities and perpetuate discriminatory biases. Students of colour, and students wearing religious dress, like headscarves, have reported issues with the software’s inability to recognize their facial features, temporarily barring them from accessing the software.”
Remote proctoring software also tends to be highly invasive, collecting a large amount of data and making covert changes to a student’s settings—changes which are, at times, not reverted properly even after the exam has been submitted and the software has been uninstalled. Widely used proctoring services ProctorU, ProctorTrack, and Proctorio all experienced data breaches or vulnerabilities. Most notably, in 2020, there was the alleged leak of 440,000 ProctorU user records, “including email addresses, full names, addresses, phone numbers, hashed passwords, the affiliated organization, and other information,” as well as an incident where the Proctorio CEO posted the chat logs of a student at the University of British Columbia on Reddit himself. Yet because schools are the actual customer of the product, students have little to no ability to opt out of an exam that is proctored remotely, and are required to provide their consent in order to complete their courses. Remote proctoring services demand sweeping permissions to student data yet have not proven themselves capable of implementing the appropriate safeguards to protect this data. Overall, remote proctoring has also been undeniably hit-or-miss with quality, likely due to the accelerated rollout pace. Glitches in the software, unresponsive support staff, and a lack of clarity with regards to why a student’s behaviour might be flagged all create an inordinately and unnecessarily stressful environment that is not conducive to its objective: acting as an objective assessment of how well students have understood academic material.
Examining our options
Despite these imperfections, it is undeniable that remote learning makes education more accessible—in fact, many students find it more conducive to an effective learning environment than in-person classes. Both forms of instruction have their own distinct characteristics, and students, with their own complex and diverse sets of needs, may differ on whether one serves them better than the other. During the onset of the pandemic, remote proctoring acted as a functional solution to an urgent problem. However, the systems as they are leave much to be desired, and there is more work to be done before they can act as reasonable long-term substitutes to in-person testing.
Additionally, when alternatives to in-person examinations are called for, university instructors should redesign assessments rather than rely on remote proctoring services to prevent violations of academic integrity. University of Colorado Denver researcher Shea Swauger remarks, “There’s a much bigger narrative about why we think surveillance is an answer to questions that are fundamentally pedagogical.” Proctored exams, whether in person or otherwise, are not always the most appropriate assessment format – while they have a time and a place, they also promote behaviours contrary to the spirit of learning, such as cramming and rote memorization, and result in little to no retention of the actual material. The pandemic can and should serve as more than a crisis for institutions to turn the page on; it can become an accelerator for instructors to re-examine their own assessment formats. Understandably, academic institutions turned to remote proctoring in haste out of a dire need of a solution that would, at least conceivably, preserve academic integrity. Now, they owe it to their students to build out other viable alternatives.