Ignoring harm, saving face: strategies of non-knowledge and knowledge avoidance, and what they mean for KM

In 2006, nearly 20 years ago, University of San Fransisco Professor Emeritus Steven Alter made the following observation in regard to knowledge management (KM) in the paper¹ “Goals and Tactics on the Dark Side of Knowledge Management”:

Although a great deal has been written about KM, the dark side of KM is largely ignored. Most discussions of KM take a utopian view that the goal is to capture essential knowledge and make it available wherever needed. Further, that knowledge will be collected and distributed accurately and with the best of intentions, leading to efficiency, better decisions, and protection of intellectual property. Not mentioned is the possibility that KM activities could be motivated by inappropriate goals or intentions to commit crimes.

A recent paper² by Adam Hannah and Linda Courtenay Botterill in the Australian Journal of Political Science introduces two dark side KM methods in the context of the Australian Government’s heinous and unlawful “Robodebt” scheme: non-knowledge and knowledge avoidance.

As I discuss in the RealKM Magazine article³ ” Two horror cases highlight the dangers of blind faith in what AI generates,” the Robodebt AI system was designed to catch people exploiting welfare. However, the AI system got many of the assessments seriously wrong to such an extent that it triggered numerous suicides⁴.

But Hannah and Botterill alert that what happened with Robodebt is much darker than just a problem of mere blind faith in flawed AI. They write that knowledge utilisation is often constrained and politicised in government policy, with ‘blindspots’ occurring for a range of reasons such as capacity limitations, ideological capture, and cognitive biases:

However, the repeated and strategic nature of knowledge avoidance makes Robodebt something more than a blindspot. Rather … it better resembles cases studied in the social science literature on strategic ignorance and non-knowledge, as well as recent literature on ‘malign’ policy-making.

Confirming this, warnings about the risks and likely illegality of the scheme were raised before it was implemented but not acted on, and then flaws were ignored, even when concerns were raised multiple times⁵.

Strategies of non-knowledge and knowledge avoidance go beyond an unintended lack of awareness. Rather, there is deliberate avoidance or concealment of knowledge. Instead of the traditional view of policymakers as well-intended but constrained, they have been described in research on this topic as acting with ‘malignity’, ‘misfeasance’, and even ‘administrative evil’.

Strategies of non-knowledge and knowledge avoidance

From their analysis, Hannah and Botterill highlight four specific strategies that key people in the Australian Government’s Departments of Human Services (DHS) and Social Services (DSS) used to avoid knowledge of the failings and harmful impacts of the Robodebt scheme:

1. Avoiding consultation.
2. Knowing what not to know and not to ask.
3. Non-inscription of inconvenient knowledge.
4. False narratives and focusing on ‘things that did not matter’.

1. Avoiding consultation

Robodebt was a major change to the practice and scale of welfare compliance in Australia, yet consultation was almost entirely absent from the policy process, while efforts to have input were ignored or treated with hostility. The lack of testing and consultation for such a significant new compliance measure has been described as ‘highly unusual’. Proposals by the Australian Council of Social Service (ACOSS) for stakeholder meetings and input were rejected. Front-line staff who raised concerns were ignored, had their views dismissed, or were treated in a hostile manner.

2. Knowing what not to know and not to ask

DSS did push back against the DHS Robodebt proposal at first, but its efforts to make the scheme’s flaws known were limited. Additionally, DSS staff who had engaged directly with the proposal showed a complete lack of curiosity when it came to its design and implementation. DSS staff saw Robodebt as a scheme that DHS had designed and pushed for, over the top of their objections. They felt that the institutional and cultural environment did not encourage speaking out. Instead, it was more convenient to not inquire too much as to what another agency was up to. But this was not something trivial to overlook.

3. Non-inscription of inconvenient knowledge

Throughout the design and implementation of Robodebt, there was a repeated unwillingness to have information that cast doubt on the scheme communicated in written form, whether internally or externally. For example, DSS staff members ‘softened’ language around their concerns with the initial proposal when they were writing to DHS staff. Staff were also explicitly told to not put things in writing.

This approach extended to attempts to actively mislead external investigations. DHS and DSS successfully misled a Commonwealth Ombudsman inquiry by responding to requests for data with delays, presenting selective evidence, or claiming to not have data they did in fact have. Additionally, when DHS asked PricewaterhouseCoopers (PwC) for advice on the delivery of the Robodebt, PwC quickly came to the view that the scheme was not ‘fit for purpose’ and would have to be substantially redesigned. However, while PwC’s findings were verbally communicated and they were paid in full, DHS did not want delivery of the final report, which PwC consultants began referring to as the ‘non-report’.

Pulling back from written acknowledgement or verification of the flaws in the scheme can also be seen in relation to legal advice.

4. False narratives and focusing on ‘things that did not matter’

The primary misleading narrative relied on by DHS and relevant Ministers stated that ‘there would be no change to how income is assessed or overpayments calculated as part of this proposal’. Similar phrasing was used in response to whistleblowers and critics. But the use of income averaging meant that there was in fact such a change, and a significant one.

A related and even more perplexing claim was that because welfare recipients had an opportunity to submit information about their income, then therefore their debts could not be said to be inaccurate. In reality, while the calculation would be ‘accurate’ relative to the input of yearly data, it may bear little relationship to income that was actually earned on fortnightly basis. This conforms to a pattern of focusing on the mechanics of the online system, rather than the fundamental issues with the scheme.

What does this mean for knowledge management?

Strategies of non-knowledge and knowledge avoidance by Australian Government agencies have allowed the unlawful Robodebt scheme to be implemented, causing serious harm. Yet, dark side KM methods remain an inappropriately neglected issue in KM, despite the alarm having first been sounded by Professor Steven Alter nearly 20 years ago⁶.

Unfortunately, just as troubling as what has happened with knowledge and Robodebt is that the KM community’s neglect of dark side KM means that there are actually disturbing examples of non-knowledge and knowledge avoidance even within the KM community. For example, some continue to promote Toyota as an example of good KM, while apparently not wanting to know about the company having been found responsible for serious cases of knowledge hiding and manipulation⁷ in the United States and Australia.

As well as a need to start taking dark side KM much more seriously, two specific approaches can assist with challenging the strategies of non-knowledge and knowledge avoidance:

1. Knowledge risk management (KRM).
2. Codes of ethics and conduct.

1. Knowledge risk management (KRM)

The knowledge risk management (KRM)⁸ approach has been proposed by Professors Susanne Durst and Malgorzata Zieba. This approach recognises that knowledge may bring not only positive outcomes, but may also be related to certain organizational threats. Strategies of non-knowledge and knowledge avoidance are related to, but more sinister than, the increasingly researched knowledge risks of knowledge withholding, hiding, and hoarding.

2. Codes of ethics and conduct

Codes of ethics and conduct serve to constantly remind staff and the members of communities and networks that their actions must be in the best interests of everyone, and not cause people harm. The Global Think-Tank of Organizational Tacit Knowledge Management (GO-TKM) has become the first KM body to progress the development of a Code of Ethics & Conducts⁹. All major KM bodies and networks should do similar, and knowledge managers should also encourage their organisations or clients to develop codes of ethics and conducts to help ensure the integrity and decency of the ways in which they work with knowledge.

Header image source: Created by Bruce Boyes with Microsoft Designer Image Creator.

References: