a/justice· 14 October 2026 · 5 min read

Australia writes the report. Then nobody closes it.

Art: Omxus Press

Aboriginal and Torres Strait Islander readers: this article names a man who has died, as his family has asked the public to do.

On 29 December 2015, in the hospital at Long Bay prison in Sydney, officers went into David Dungay Jr's cell to stop him eating rice crackers. The Dunghutti man, 26, had diabetes. They moved him and held him face down. He said "I can't breathe" twelve times, and he died there (Lee, 2019).

Australia had already held the inquiry that should have prevented it. Twenty-four years earlier.

339 recommendations

In April 1991 the Royal Commission into Aboriginal Deaths in Custody reported on 99 deaths across the 1980s. Its finding was blunt. Aboriginal people in custody didn't die at a higher rate than anyone else in custody. They died so often because so many of them were locked up. It made 339 recommendations. Make arrest the last resort. Take hanging points out of cells. Call an Aboriginal legal service when an Aboriginal person is taken in. Stop locking people up for being drunk in public (Johnston, 1991).

It was one of the most thorough looks a government has ever taken at itself, and the diagnosis was right. What happened next is the story.

27 years later

In 2018 the federal government paid for a review. By governments' own reports, 64% of the recommendations were complete (Deloitte Access Economics, 2018). Researchers at the Australian National University went through later inquests and said even that was too generous: key recommendations were still missing, and people died where they were missing (Anthony et al., 2021).

Check the root cause the Commission named. In 2017–18, Aboriginal and Torres Strait Islander adults were imprisoned at 10.2 times the rate of other Australians. By 2024–25 it was 14.5 times (Productivity Commission, 2026). The rate for Aboriginal and Torres Strait Islander adults rose from 1,718.9 to 2,112.6 per 100,000, while the rate for everyone else fell. The rate of death per prisoner still looks much like everyone else's, as the Commission found and later monitoring confirmed (Gannoni & Bricknell, 2019). More people locked up means more deaths.

The Commission said people died in custody because too many were in custody. On that measure the problem is growing. A society that investigates its failures and never closes them isn't learning. It is performing learning.

Not a story about bad people

Most people in this are doing their jobs. The shape of the system does the rest. Calling an inquiry looks serious and gets rewarded on the day; one study of British governments found the choice to call one is driven by avoiding blame, media attention and popularity (Sulitzeanu-Kenan, 2010). Carrying out the recommendations is slow, costly and rarely news. Nobody owns them. Nothing dates them. No public list shows which are still open.

Victoria tried the obvious fix. Since 2008, anyone who gets a coroner's recommendation must reply in writing. Researchers followed 164 recommendations over 33 months. 37% were accepted and implemented, 27% were rejected, and for 36% the action was already under way (Sutherland et al., 2014). A duty to reply is better than nothing. It isn't a duty to close.

How engineers do it

When a big website goes down, the people who run it write a "postmortem": what happened, why, and a list of fixes. The rule that matters comes last. Every fix gets a named owner and stays on a list until it is done. A postmortem whose fixes never happen counts as wasted work (Beyer et al., 2016). It's also "blameless". It looks for the conditions that let the failure through, because blaming the nearest person leaves the system ready to fail again (Reason, 2000).

So we hold a cat video to a higher standard than a death in custody. Hospitals borrowed half of this and skipped the rest: of 1,137 recommendations from root-cause analyses of serious incidents in Victoria, 8% were strong enough to be likely to work (Hibbert et al., 2018).

The missing half

Give every recommendation four things: a named owner, a public due date, a status anyone can look up, and an independent check that the harm actually stopped. If a recommendation is wrong, close it with "rejected, because", in public. What it never gets to do is sit quietly on a shelf. This costs almost nothing. It's a spreadsheet with consequences.

What you can do

You don't need a royal commission to start. Pick one recommendation from the 1991 report, or from a coroner's finding in your state. Public drunkenness is a good one: a recommendation from 1991, and Victoria only ended the offence in 2023. Write to your state MP and ask three things. Who owns this? When is it due? Where can I see the status?

Australia is good at finding facts and naming causes. Lives depend on what comes after, and a question from a constituent is how that starts. Prevention isn't a slogan. It's a closed loop.

Sources

Gannoni, A., & Bricknell, S. (2019). Indigenous deaths in custody: 25 years since the Royal Commission into Aboriginal Deaths in Custody. Statistical Bulletin, 9. Australian Institute of Criminology. https://doi.org/10.52922/sb09418
Anthony, T., Jordan, K., Walsh, T., Markham, F., & Williams, M. (2021). 30 years on: Royal Commission into Aboriginal Deaths in Custody recommendations remain unimplemented (Working Paper No. 140). Centre for Aboriginal Economic Policy Research, Australian National University.
Beyer, B., Jones, C., Petoff, J., & Murphy, N. R. (Eds.). (2016). Site reliability engineering: How Google runs production systems. O'Reilly Media.
Deloitte Access Economics. (2018). Review of the implementation of the recommendations of the Royal Commission into Aboriginal Deaths in Custody. Department of the Prime Minister and Cabinet.
Hibbert, P. D., Thomas, M. J. W., Deakin, A., Runciman, W. B., Braithwaite, J., Lomax, S., Prescott, J., Gorrie, G., Szczygielski, A., Surwald, T., & Fraser, C. (2018). Are root cause analyses recommendations effective and sustainable? An observational study. International Journal for Quality in Health Care, 30(2), 124–131. https://doi.org/10.1093/intqhc/mzx181
Lee, D. (2019). Inquest into the death of David Dungay. Coroners Court of New South Wales.
Johnston, E. (1991). Royal Commission into Aboriginal Deaths in Custody: National report. Australian Government Publishing Service.
Productivity Commission. (2026). Report on government services 2026, Part C, Section 8: Corrective services, dataset. Australian Government.
Reason, J. (2000). Human error: Models and management. BMJ, 320(7237), 768–770. https://doi.org/10.1136/bmj.320.7237.768
Sulitzeanu-Kenan, R. (2010). Reflection in the shadow of blame: When do politicians appoint commissions of inquiry? British Journal of Political Science, 40(3), 613–634. https://doi.org/10.1017/S0007123410000049
Sutherland, G., Kemp, C., Bugeja, L., Sewell, G., Pirkis, J., & Studdert, D. M. (2014). What happens to coroners' recommendations for improving public health and safety? Organisational responses under a mandatory response regime in Victoria, Australia. BMC Public Health, 14, 732. https://doi.org/10.1186/1471-2458-14-732

Keep reading