The ACT Government Mandates ISO 31000 for Risk Management. Its Domestic Violence System Meets 0 of 17 Requirements in Full.

Procurement gets structured risk management. Staff travel gets it. Public events get it. But a process capable of removing someone from their home, separating them from their children and suspending their firearms licence does not.

The ACT Government has a risk-management policy.

It is not obscure.

Government functions are supposed to manage risk using a recognised international standard: AS ISO 31000:2018, Risk Management — Guidelines.

Government procurement is subjected to structured risk assessment.

Public events are.

Staff travel is.

Information security is.

Yet according to a detailed clause-by-clause assessment of the ACT’s July 2022 Domestic and Family Violence Risk Assessment and Management Framework, one of the Territory’s most consequential exercises of state power does not satisfy a single one of 17 tested ISO 31000 requirements in full.

Eight requirements were assessed as not met.

Nine were partly met.

Zero were fully met. A_System_That_Cannot_See_Itself_Technical_Volume_7x10_interior.pdf

That should trouble anyone concerned about domestic violence, irrespective of politics or gender.

Because risk management is exactly what this system claims to be doing.

This isn’t an argument about paperwork

ISO 31000 does not exist to make organisations complete prettier forms.

Risk management exists because consequential decisions have to be made under uncertainty.

Domestic and family violence is almost a textbook example.

A police officer may arrive at a home at two o’clock in the morning.

Two people give incompatible accounts.

There may be little independent evidence.

The immediate consequences of getting the decision wrong can run in either direction.

Leave a genuine victim exposed and somebody may be seriously injured or killed.

Misidentify the aggressor and a victim may themselves be charged, removed from their home or separated from their children.

The ACT’s own death review demonstrates that the second error is not theoretical: 8 of 13 deceased victims it examined had previously been misidentified as the predominant aggressor by police or another agency. A_System_That_Cannot_See_Itself_Technical_Volume_7x10_interior.pdf

This is precisely why a functioning risk-management system matters.

The missing ingredients are fundamental

The technical review identifies what the 2022 ACT framework lacks:

No defined risk criteria.

No likelihood scale.

No consequence scale.

No risk-tolerance threshold.

No structured comparison of treatment options.

No residual-risk step.

No named risk owners.

No framework-level monitoring.

No framework-level reporting.

The resulting assessment was eight ISO 31000 requirements not met, nine partly met and none fully met. A_System_That_Cannot_See_Itself_Technical_Volume_7x10_interior.pdf

These are not cosmetic omissions.

Without risk criteria, two people assessing the same facts have no defined basis for deciding whether they mean the same thing when they say “high risk”.

Without consistent decisions, reliability cannot be measured.

Without measuring reliability, error cannot reliably be detected.

Without detecting error, the system cannot learn from it.

And the ACT’s own independent service review subsequently found that agencies were using different risk processes and could reach different conclusions about the same case. A_System_That_Cannot_See_Itself_Technical_Volume_7x10_interior.pdf

Now compare that with an interim family violence order

An interim family violence order can have immediate consequences.

It can exclude somebody from their home.

It can restrict contact with their partner and children.

It automatically suspends a firearms licence.

Contravening the resulting order can become a criminal offence carrying up to five years’ imprisonment. A_System_That_Cannot_See_Itself_Technical_Volume_7x10_interior.pdf

Yet the interim order may be made before the respondent has been heard.

The Magistrates Court need not comply with the ordinary rules of evidence in proceedings under the Family Violence Act.

That flexibility exists for a good reason. Family violence often occurs privately, urgent protection can be necessary, and demanding conventional evidence before intervention could leave genuine victims in danger.

But relaxing evidentiary rules makes the quality of the decision-making methodology more important, not less.

The weaker the evidentiary filter, the stronger the need for a consistent, documented method for assessing uncertainty.

Instead, the public record discloses no defined likelihood and consequence scales, no stated risk threshold, no structured comparison of treatment options and no documented method for recording what is unknown. A_System_That_Cannot_See_Itself_Technical_Volume_7x10_interior.pdf

The absurd comparison

This is where the situation becomes difficult to defend.

Before an ACT Government agency undertakes procurement, holds a public event or approves staff travel, structured risk-management processes may require criteria, likelihood and consequence scales, treatments, ownership and residual-risk acceptance.

But an interim family violence order can remove a human being from their home, suspend a professional or firearms licence and disrupt contact with their children.

There is no equivalent documented risk structure publicly demonstrated for that decision. A_System_That_Cannot_See_Itself_Technical_Volume_7x10_interior.pdf

Think about that for a moment.

The ACT applies more recognisable risk-management architecture to administrative and operational activities than it can demonstrate for a decision capable of radically changing somebody’s life before they have been heard.

That does not mean interim orders should become harder to obtain.

It means consequential decisions deserve better decision support.

And the consequences of error are already documented

This is not a hypothetical standards-compliance exercise.

The ACT’s own 2025 death review found that more than half of the 13 deceased victims it examined had previously been misidentified as aggressors.

The consequences documented included criminal charges, employment loss, children being removed and incarceration.

The system therefore already possesses evidence that the classification can fail catastrophically.

What it does not possess is an established error rate.

No sensitivity, specificity, inter-rater reliability or error analysis has been published for an ACT family violence risk or classification instrument. A_System_That_Cannot_See_Itself_Technical_Volume_7x10_interior.pdf

You cannot credibly call something a risk-assessment system while refusing to ask whether two trained assessors presented with the same case reach substantially the same assessment.

That is not an ideological objection.

It is measurement.

The same failure appears at the system level

The problem does not stop with individual assessments.

ACT households have paid the Safer Families Levy since 2016.

By 30 June 2024 it had raised more than $46 million.

Yet in November 2024, the ACT Auditor-General found that the responsible office had not developed a performance monitoring strategy or evaluation framework capable of measuring the outcomes and impacts of initiatives funded through that levy. A_System_That_Cannot_See_Itself_Technical_Volume_7x10_interior.pdf

The ACT’s 2026–2036 Domestic, Family and Sexual Violence Strategy finally establishes six outcomes.

But the technical review found that neither the Strategy nor its First Action Plan contains the words “baseline” or “indicator”, sets a numeric target for an outcome, nominates a data source or identifies a lead agency responsible for an outcome.

The monitoring and evaluation framework that could supply those things is itself future work. A_System_That_Cannot_See_Itself_Technical_Volume_7x10_interior.pdf

Ten years of reform.

More funding.

More services.

More legislation.

More institutions.

But still no scoreboard capable of demonstrating whether the underlying risk is going up or down.

The verdict isn’t that the system doesn’t work

That would go beyond the evidence.

The more disturbing conclusion is:

We don’t know.

Some programs show promising results. Frontline services unquestionably do valuable work. The 2026 Strategy improves substantially on what came before it.

The technical review does not conclude that the money has been wasted, that family violence orders cause suicide, that false allegations are common, or that officials are acting in bad faith. A_System_That_Cannot_See_Itself_Technical_Volume_7x10_interior.pdf

Its finding is narrower.

And harder to dismiss.

The ACT has constructed a large domestic and family violence system without constructing the measurement architecture necessary to demonstrate its aggregate effectiveness, detect all of its errors