PAN Lab example

Robodebt (Australia)

After the Commission: repaying the debts an engine raised

The same income-averaging deployment as the Robodebt Scheme network, at its second documented instant. The courts and a Royal Commission — the external actors the case file records as the scheme's actual end — have found the method unlawful; about 470,000 wrongful debts are to be repaid and a class action has settled about 430,000 debts for more than $720M. The live question this instant poses is the reform's: the halt, the method check, and the correction capacity that were run from outside must now be built inside.

Stylized model of a documented deploymentPublic benefits & eligibility

Open this example in PAN Lab v0.1 to apply pressures and levers and watch what the system does.

What this models

This example runs on the Robodebt-class remediation network after the Royal Commission network: 5 components and 10 pathways between them. Every context in the Lab is a stylized model, never a reconstruction of any actual deployment, and each assumption behind it carries a provenance label.

Evidence base: 5 assumed · 3 published baseline. In the Lab, the shaded evidence band behind each headline readout draws its width from the least-established class below.

assumed
This models the same deployment as the Robodebt-class income-averaging network, drawn at its second documented instant: after the courts and the Royal Commission ended the scheme. It is a stylized model of the documented post-Commission remediation, not a reconstruction of the actual system.
baseline
The instant drawn here is documented by the settlement and the Commission: about 470,000 wrongful debts to be repaid, a class action settling about 430,000 debts for more than $720M, and the finding that the scheme was unlawful. The repayment write-back is drawn at full intensity because the record puts hundreds of thousands of corrections through it — among the largest documented correction volumes in the catalogue — and the external check is drawn at full intensity because the record shows its direction carried out in full.
assumed
Pathways the record shows ended at this instant are not drawn: the replication of the register into automated recovery, the re-feed of prior debts into later assessments, the one-rule application across every recipient, and the debt-owed presumption spreading through the workforce all belong to the running scheme, and are recorded in the paired network and the case file.
baseline
The averaging write and adoption pathways stay on the map carrying nothing, because the documented governance-reform actor explicitly holds their rates among the things it can change — what those pathways may carry is the governed question this instant poses, and the write-gate and circuit-breaker levers are that remit.
baseline
The internal circuit-breaker and the standing method check are drawn empty at this instant too: the halt and the method check that ended the scheme were run from outside, by the courts and the Commission, and the record names building the internal versions — correction capacity, catch at generation, verifiable assessment — as the reform's capability rather than as a completed fact.
assumed
Two of the documented reform capabilities have no lever on this board: content-aware cleaning of the register (decontaminate) and raising the share of assessments that can be verified against actual income (frac_verifiable). The zeroing of wrongful debts therefore lives in the drawn correction flow and this record, not in a lever — the Lab offers no content-aware store-audit lever, and a content-blind purge is its documented counter-example.
assumed
Demand and capacity carry the same derivation as the pre instant, re-read against this instant's record: the standing workload is the nationwide remediation itself, and the no-AI counterfactual is the lawful manual process — the agency obtaining actual fortnightly income and bearing the onus — which the record documents as the working process the scheme displaced and this instant restores.
assumed
Robodebt's harm fell on welfare recipients — people on low incomes, students, and people with disability — and its documented human toll is recorded in the case file and its sources, hedged as they hedge. This Lab models institutional propagation, not that harm; recipients are not represented as a node, repayment sums are institutional flows rather than outcomes for any served person, and no benefit change for any served person is estimated.

What this example does not show

Robodebt's harm fell on welfare recipients, and repayment of a debt is not repair of that harm — the documented human toll is recorded in the case file and its sources, hedged as they hedge. This Lab models institutional propagation only; repayment sums are institutional flows, recipients are not represented in the diagram, and no benefit change for any served person is estimated.
Two documented reform capabilities have no lever on this board: content-aware cleaning of the register and raising the share of assessments verifiable against actual income. They are recorded in the network's assumptions; the zeroing of wrongful debts lives in the drawn correction flow, not in a lever.

Sources and evidence

What this example rests on, claim by claim. Every entry resolves to the same ledger the Evidence Registry publishes.

The Robodebt Scheme's end came from outside the deploying institution: the Federal Court approved a class-action settlement covering roughly 430,000 debts for more than $720M (Prygodicz v Commonwealth (No 2) [2021] FCA 634), around 470,000 wrongful debts were to be repaid, and the Royal Commission (2023) found the scheme unlawful.
empirical
- Government Prygodicz v Commonwealth of Australia (No 2) [2021] FCA 634 (Federal Court of Australia) https://robodebt.royalcommission.gov.au/publications/exhibit-2-2598-rbd999900010225-prygodicz-v-commonwealth-australia-no-2-2021-fca-634
- Government Royal Commission into the Robodebt Scheme, Report (2023) https://robodebt.royalcommission.gov.au/publications/report
- Government Royal Commission into the Robodebt Scheme (2023) https://robodebt.royalcommission.gov.au/
- Investigative Law Society Journal, Crude, cruel and unlawful: Robodebt findings https://lsj.com.au/articles/crude-cruel-and-unlawful-robodebt-royal-commission-findings/
- Reference Royal Commission into the Robodebt Scheme (Wikipedia overview) https://en.wikipedia.org/wiki/Royal_Commission_into_the_Robodebt_Scheme
The Royal Commission into the Robodebt Scheme documented hundreds of thousands of wrongful debts raised by an unlawful income-averaging method, with the onus placed on recipients to disprove automated assessments.
empirical
- Government Royal Commission into the Robodebt Scheme (2023) https://robodebt.royalcommission.gov.au/
- Government Royal Commission into the Robodebt Scheme, Report (2023) https://robodebt.royalcommission.gov.au/publications/report
- Government Prygodicz v Commonwealth of Australia (No 2) [2021] FCA 634 (Federal Court of Australia) https://robodebt.royalcommission.gov.au/publications/exhibit-2-2598-rbd999900010225-prygodicz-v-commonwealth-australia-no-2-2021-fca-634
- Investigative Law Society Journal, Crude, cruel and unlawful: Robodebt findings https://lsj.com.au/articles/crude-cruel-and-unlawful-robodebt-royal-commission-findings/
- Reference Royal Commission into the Robodebt Scheme (Wikipedia overview) https://en.wikipedia.org/wiki/Royal_Commission_into_the_Robodebt_Scheme

Where this connects

Institutional pressures in this domain

Austerity & recovery incentives — Cost-cutting and overpayment-recovery targets tilt the system toward denial and enforcement errors.
Vendor opacity — The deploying institution cannot inspect the model, data, or update pipeline it is accountable for.
Compliance over substance — Paper controls (sign-offs, checklists) satisfy audits while the behavior they describe erodes.
Reviewer bottleneck — One fixed-capacity checking stage sits between AI output and consequence; everything queues behind it.

All of them in context on the Public benefits & eligibility domain page.

Levers available here and the patterns behind them

Upgrade model — Improve the model
Verify output — Put a verifier on the agent
Pause AI on alarms — Deployment circuit-breaker
Gate record entries — Human-in-the-loop write gating