John Pane, chair of Electronic Frontiers Australia, says it is like the panopticon. “You never know whether the boss is watching you, which is why it’s called bossware. It works in the same way.”
Pane has a long history in senior privacy roles in the big banks, Australia Post and Deloitte. Over this time he has witnessed the evolution of the internet and with it workplace surveillance that has evolved into “bossware”.
Pane emphasises that surveillance is nothing new. It began as soon as the internet did, logging web activity to prevent employees “doing naughty things”. Different professions have different levels of oversight. Call centre workers and those in logistics networks are subject to more intense scrutiny than most office workers.
Working from home during the pandemic opened the door to broader surveillance – and some companies took up the opportunity to monitor their workers remotely. The feature set for such software was initially limited, logging internet history and keystrokes. As bandwidth and processor speed grew, however, and sensor capabilities increased, screenshotting, location tracking, webcam operation and even biometric data capture were added. Part of this is facilitated by the fact employee data is exempt from protections afforded to other data under the Privacy Act. Legally, this is a live issue. A 2022 review has challenged the workplace exemption, with its proposals to be debated in federal parliament this year.
Data capture is only half the conversation, however. The ascent of artificial intelligence has bloated the claims of productivity monitoring, rating and ranking software.
“The dirty secret about AI is that it’s totally driven by surveillance,” explains Peter Lewis, executive director of Essential Media and founder of The Australia Institute’s Centre for Responsible Technology.
Monitoring software interprets and categorises captured data and asserts whether the worker is productive or not. Based on this, platforms can pit employees against each other on leaderboards, potentially influencing promotions or firings. Take one example: Suzie Cheikho was fired from her job at IAG Australia in part because keystroke logging showed she typed on average less than one character a minute. While the Fair Work Commission upheld IAG’s decision to dismiss Cheikho, it reveals how metrics are used in company decision-making.
Many platforms on the market have been developed in the United States, where labour laws are more lax. A post by the deputy US chief technology officer for policy, timed for International Workers’ Day last year, detailed the dangers of excessive monitoring. Eight of the 10 largest US companies capture and assess individual workers’ productivity. The most well-known are conditions across the Amazon delivery network, where almost all warehouse worker activity is captured on high-precision sensors, and some drivers report urinating into bottles to avoid repercussions from long toilet breaks.
Amazon also uses harvested data to target unionising. Pane is quick to cite the Amazon example, saying: “It’s treating people like livestock in mass production lots. It’s an invasion of our autonomy and human rights and people should not be exploited in that way.”
Using quantification and measurement to drive decision-making is becoming ubiquitous throughout US businesses. “We’ve been obsessed by that as a society in the US for quite some time,” says Lisa Kresge, a policy research associate in the Technology and Work Program at the UC Berkeley Labor Center. Kresge has written extensively on workplace monitoring and says software is “mistaking productivity for busy activity”.
Most workers spend time involved in non-digital labour, such as speaking to co-workers. Each job is also different, yet most software is one size fits all, deployed across a range of industries. Configuration has limits, with biases that emerge in one context invisible in another. “We have no idea whether or not these systems are actually working,” Kresge says, adding there are “different layers of problems that can be introduced with zero oversight”.
In Australia, Kresge’s reservations are shared by local experts, in particular the lack of transparency around bossware. “What are the parameters that determine whether someone’s productive or not?” asks Pane.
Lewis is more equivocal, saying it might be possible to develop good software if employees are involved but “if you don’t have human oversight, you’re totally exposing yourself to the architecture of the original algorithm”. He cites the robodebt scandal as an example of where automation without oversight can end up.
Peter Leonard, a professor at the School of Information Systems and Technology Management at UNSW Sydney, says his greatest concern is “there is no legal requirement for consultation or workplace bargaining over these kinds of employer intervention outside NSW and ACT”.
It is nearly impossible to get a complete view of how pervasive monitoring software is within Australian businesses. Reports are mostly dated or based on small samples. A representative from Florida-based company Veriato told The Saturday Paper its employee monitoring software had “hundreds of Australian customers” across companies of all sizes.
Leonard said he would be surprised if anything less than 75 per cent of the larger companies in Australia were practising monitoring. He explains there are reasonable uses, such as ensuring sensitive data and systems remain secure, but includes the caveat that security data should be insulated from managers deciding career prospects.
A vanguard for surveillance is the financial sector, which legally requires surveillance to prevent the leaking of privileged market information. Julia Angrisano, national secretary of the Finance Sector Union, says her members are being tracked through “surveillance by stealth”. She speaks to Leonard’s point about the creep of purpose, when data captured can be reused elsewhere: “Data sets and surveillance are becoming increasingly used by finance companies to inform managerial decisions; workers’ data and privacy protection hardly exist.”
While writing this piece, I trialled two monitoring platforms. The first was WorkTime by Canadian company NesterSoft Inc, whose marketing material made claims regarding Australian privacy law. After installing a widget I was free to observe myself as a manager might. With a deadline for this piece looming, I felt that I was working quite hard, but WorkTime begged to differ.
It categorised much of my toil under the heading “Undefined”, labelling most of the rest as “Unproductive”. Surely much of the “Undefined” category was a configuration issue, yet it was definite that my time on Gmail was unproductive. Once installed, the application was invisible and alerted me only to ask admin permissions to access information on applications. I soon forgot it was there. At the end of one day, it determined I was productive for only 16 minutes.
More visible but also more invasive was Insightful by San Francisco-based Insightful.io, which installed a visible tracker for me to “clock-in” and “clock-out”. It also screenshotted my desktop, initially sending through empty backgrounds as it did not have permission to view app contents. This gave the impression I was doing nothing. Once permission was granted, it was free to send through contents of WhatsApp chats, Zoom meetings, whatever was on my screen.
On both platforms it was clear to me the data coming through was not accurate, but as an employee I would not have a manager’s view, so would be unable to defend my activity.
Both Pane and Kresge suggested this level of monitoring was a violation of something inherently human. For Kresge, it was that humans exist in relation to other humans and individual data was always only part of the story. For Pane, it was the expectation of work. “It does seem almost colonial …” he said. “Humans cannot work like an automaton. We can only concentrate, or do a task with concentration, for a fixed period of time before we stop and do something else.”
This article was first published in the print edition of The Saturday Paper on January 27, 2024 as "Worse for bossware".
This month marks 10 years since the first edition of The Saturday Paper. The paper is as audacious now as it was then: a rejection of conventional wisdom about what makes the news and who will read it.
To celebrate those 10 years - and the issue-defining journalism produced in them - we are offering all new subscribers a two-year digital subscription for the price of one. That's $298 worth of journalism for $109.
Get more of the best journalism in the country - and celebrate the success of a newspaper built on optimism.
Select your digital subscription