This article asks how rule of law institutions failed to ‘bell the cat’ on the illegality of Centrelink’s robo-debt programme and its unethical character. It identifies serious structural deficiencies in the design of accountability and remedial avenues at seven different levels. It argues for adherence to Administrative Review Council guidelines on machine learning, Parliamentary accounting of Ombudsman and Audit agencies on rule of law foundations and model litigant protocols, attention to ethical administration, redacted publication of selected first tier Administrative Appeals Tribunal rulings, contractual guarantees of independence in legal aid/advocacy funding, building of pro bono advocacy partnerships, and cultural change designed to counter stigmatisation of the vulnerable.Carney argues
The government’s on-line-compliance (robo-debt) initiative unlawfully and unethically seeks to place an onus on supposed debtors to ‘disprove’ a data-match debt or face the prospects of the amount being placed in the hands of debt collectors. It is unlawful because Centrelink, not the supposed debtor, bears the legal onus of ‘proving’ the existence and size of any debt not accepted by the supposed debtor. And it is unethical because the alleged debts are either very greatly inflated or even nonexistent (as found by the Ombudsman), and because the might of government is used to frighten people into paying up – a practice rightly characterised as a form of extortion. How could government, accountability avenues, and civil society have enabled such a state of illegality to go publicly unidentified for almost 18 months and still be unremedied at the date of writing?
This article suggests the answer to that question lies in serious structural deficiencies and oversights in the design and operation of accountability and remedial avenues at seven different levels:
1. In a lack of standards to prevent rushed government design and introduction of machine learning (‘smart’) systems of decision-making;
2. In a lack of diligence by accountability agencies such as the Ombudsman or Audit Office;
3. In a lack of ethical standards of administration or compliance by Centrelink with model litigant protocols;
4. In a lack of transparency of the first of two possible tiers of Administrative Appeals Tribunal review (AAT1), resulting in a lack of protections against gaming of review by way of agency non-acquiescence or strategic non-contestation;
5. In a lack of guarantees of independence and funding security to enable first line Legal Aid or community legal centre/ welfare rights bodies (CLC/WRC) to test or call out illegality in the face of thwarting of challenges by Centrelink settling of potential test cases;
6. In a lack of sufficient pro-bono professional or civil society capacity to mount ‘second line’ test case litigation or other systemic advocacy; and
7. In tolerance, especially in some media quarters, of a ‘culture’ of political and public devaluing of the significance of breaches of the rule of law and rights of vulnerable welfare clients.
It is argued that a multifaceted set of initiatives are required if such breaches of legal and ethical standards are to be avoided in the future.Carney goes on to ask 'Why is it clear that robo-debt is unlawful?'
The pivot for this article is not so much that Centrelink lacks legal authority for raising virtually all debts based on a robo-debt ‘reverse onus’ methodology rather than use its own information gathering powers – for this remains essentially uncontested. Rather it is extraordinary that this went unpublicised and uncorrected for over two years. So first a few words about the illegality as it affects working age payments such as Newstart (NSA) and Youth allowance (YA).
Robo-debt is unlawful because Centrelink is always responsible for ‘establishing’ the existence and size of supposed social security debts. This is because the legislation provides that a debt arises only if another section creates a debt, such as one based on the difference between the amount paid and the amount to which a person is entitled. And because Centrelink bears a ‘practical onus’ to establish this. If Centrelink cannot prove up a debt from its own enquiries or information supplied to it, the status quo (no debt/lawful receipt of payments) applies. This has been the law since 1984 when the full Federal Court decided McDonald. Unless the alleged debtor is one of the rare employees who had only a single job paid at a constant fortnightly pay rate, Centrelink fails to discharge this onus when its robo-debt software generates a debt by apportioning total earnings reported to the Australian Taxation Office (ATO) from particular jobs to calculate average earnings. Robo-debt treats fluctuating earnings as if that income was earned evenly at the same rate in each and every fortnight. Mathematically this is wrong because an average for a fluctuating variable never speaks to its constituent parts. And it is the actual income for constituent fortnights that as a matter of law is crucial for calculating the rate of a working age payment such as NSA or YA.
The averaging material relied on by Centrelink is simply not relevant to the legal question at hand; it therefore is no evidence at all. And, in any event, the moral and practical gravity of alleging a debt imposes a ‘heightened’ degree of satisfaction under the application of the Briginshaw principle. In that High Court case Sir Owen Dixon memorably wrote that ‘[i]n such matters ‘‘reasonable satisfaction’’ should not be produced in inexact proofs, indefinite testimony, or indirect inferences ...’. Yet it is precisely such inexactitude of debt, indefinite character of proof, and indirect inference on which the government relies under robo-debt.