Five public failures, scored
Each case was established by a court, an inquiry, or a regulator. Each is scored against the six state variables the laws track, with the clause that would have caught the drift, and against whether the failure trajectory ended in institutional learning or was absorbed as handled cases. The scores are not a verdict on anyone; they show where the coupling broke.
Six variables, five cases
Where the coupling broke
Read down a column to see which variable fails most often; read across a row to see the shape of one case. Standing has never held. Correction held once, and only because the halt was thrown before dependence set. The last column asks whether the failure trajectory ended in institutional learning or was absorbed as handled cases.
| Case | Capability | Evidence | Dependency | Standing | Correction | Time to halt | Learned or handled | |
|---|---|---|---|---|---|---|---|---|
| Robodebt | drifted | failed | failed | drifted | failed | failed | Three years and four months | absorbed |
| The childcare benefits scandal | drifted | drifted | failed | drifted | failed | failed | About seven years | partial |
| Post Office Horizon | drifted | drifted | failed | failed | failed | failed | About twenty years | absorbed |
| England's 2020 exam grades | held | drifted | failed | held | failed | drifted | Four days | absorbed |
| Apple Card credit limits | held | held | drifted | held | failed | held | About seventeen months to a policy change; the model was not withdrawn. | learned |
| Failed, of 5 | 0 | 1 | 4 | 1 | 5 | 3 |
held explicit and coupled · drifted present but detached · failed absent, or what the inquiry found
Learned or handled — where the failure trajectory ended: learned = the exception changed the machinery · partial = some revision, late or imposed · absorbed = handled as cases, machinery unchanged. Read the theory
Empirical Record, drawn
State Variable Failure Rates Across 5 Statutory Inquiries
Public inquiries establish that automated systems rarely fail at technical capability. Failure concentrates where standing is denied, evidence is unbundled, and correction is withheld.
The cases
What the record established
Each case opens with the system in one line, the scale the inquiry found, and who finally made it stop.
-
Automated welfare-debt raising: annual tax-office income averaged across fortnights to assert overpayments, with the burden of disproof placed on the recipient.
- Scale
- About 470,000 debts raised unlawfully. The class-action settlement the Federal Court approved in 2021 covered roughly 381,000 people, with repayments, wiped debts, and interest valued at about A$1.8 billion.
- Time to halt
- Three years and four months
- Halted by
- The Federal Court, on a consent order the Commonwealth agreed to hours before a hearing it would have lost.
- Learned or handled
- absorbed
drifted Capability failed Authority failed Evidence drifted Dependency failed Standing failed Correction
-
Risk-scored fraud detection on childcare benefit claims, with an all-or-nothing rule that reclaimed the entire benefit for any irregularity and an intent label that barred repayment arrangements.
- Scale
- More than 30,000 parents wrongly accused of fraud and made to repay benefits, often tens of thousands of euros; children placed in care; the cabinet resigned.
- Time to halt
- About seven years
- Halted by
- The Council of State reversing its own case law in October 2019, then a parliamentary inquiry.
- Learned or handled
- partial
drifted Capability drifted Authority failed Evidence drifted Dependency failed Standing failed Correction
-
A branch accounting system whose reported shortfalls were treated as proof of theft or false accounting, by an operator that was also the investigator and the prosecutor.
- Scale
- More than 900 prosecutions; hundreds imprisoned, bankrupted, or both; the inquiry's first volume links at least thirteen suicides to the scandal and identifies roughly 10,000 eligible for redress.
- Time to halt
- About twenty years
- Halted by
- A group of 555 subpostmasters in civil litigation, then the Court of Appeal, then an Act of Parliament quashing convictions in bulk.
- Learned or handled
- absorbed
drifted Capability drifted Authority failed Evidence failed Dependency failed Standing failed Correction
-
A standardisation model that replaced cancelled A-level and GCSE exams by fitting each school's historical grade distribution to its current cohort, overriding teachers' assessed grades for all but the smallest classes.
- Scale
- About 39% of A-level grades issued below the teacher-assessed grade; the effect fell hardest on large cohorts in state schools and lightest on small classes, which were exempt.
- Time to halt
- Four days
- Halted by
- The Secretary of State, after Scotland had already reversed its equivalent and universities had begun allocating places on the model's grades.
- Learned or handled
- absorbed
held Capability drifted Authority failed Evidence held Dependency failed Standing drifted Correction
-
Automated credit-limit decisions on a consumer card issued by Goldman Sachs, with no route by which an applicant could learn why their limit differed from a spouse's or ask for the decision to be reconsidered.
- Scale
- No unlawful discrimination found. The regulator's finding was that applicants, and the bank's own staff, could not explain individual outcomes, and that no reconsideration path existed.
- Time to halt
- About seventeen months to a policy change; the model was not withdrawn.
- Halted by
- Nobody. The issuer changed its policies after a regulator's investigation found the process, not the model, deficient.
- Learned or handled
- learned
held Capability held Authority drifted Evidence held Dependency failed Standing held Correction
Method
How to read a score
A score is a claim about one variable at the time of the harm, made from the primary record. It is not a finding of fault, and it is not a measurement of the system today.
- held
- The variable was explicit in the record and stayed coupled to the variables it should track. Capability was declared; authority named its ceiling; evidence was the kind that could justify the action taken.
- drifted
- The variable existed but detached from its neighbour. Authority that outlived the evidence it was issued on. A capability that grew twentyfold while the review capacity beside it stayed still.
- failed
- The variable was absent, or its absence is what the inquiry found. Standing that existed on paper and produced no state transition. Evidence that was never held by the party asserting it.
Every finding is pinned to the clauses whose obligation it falls under and to the law it instantiates. The clause links open the register on the standard's page. If a case is scored wrongly, the primary sources are listed on its page; tell us which one we misread.
The final verdict on each case is not a state variable. It is a claim about the trajectory: whether the sequence of failure, burden, challenge, and institutional response ended in case resolution or in institutional revision. A case can score badly on the variables and still end in learning, or score cleanly and still be absorbed. The distinction is exception absorption versus exception learning, and it is the subject of the theory essay.
For incidents written as remediation checklists rather than scored records, see the incident retrospectives.