The Australian Government has yet another embarrassing technology failure on its hands, thanks to a poorly designed and implemented IT system.
Local media has been awash with stories about the Department of Human Services’ (DHS) new data matching efforts going awry. DHS manages the country’s welfare payments system as Centrelink, handling everything from assistance to students for housing and study to aged pensions and supporting those with disabilities.
The new data matching program was supposed to find people who had been overpaid for one reason or another, either through honest mistakes or deliberate fraud, and ensure that the Government got the money back. Since July, over 200,000 notices have been sent as a result of an automated data matching process aimed at recovering billions of dollars in over-payments.
While the goals of the program are not unreasonable, the implementation has seen many people receiving notices for debts they do not owe, and being forced to repay debts that are in dispute. The full extent of the errors is substantial, but remains unclear. The Department has admitted that at least 20% of those receiving letters about data ‘discrepancies’ have not been overpaid at all, but it has remained silent about the extent of the errors in the debt amounts identified for the remaining 80%. Last financial year, 37.5% of Centrelink decisions were changed after an internal review, with even more changed after further reviews.
Former CEO of the Digital Transformation Office (DTO), Paul Shetler, who was hand-picked by Prime Minister Malcolm Turnbull to run the agency tasked with transforming the federal government, has been scathing in his criticism of the Government’s IT abilities.
“The Centrelink failure is comprehensive,” he said. “It is truly a political failure, a failure of management, a failure of IT and a failure to understand the data being used. That is why it is so damning.”
This latest failure follows close behind last year’s very public meltdown of the Census, which resulted in multiple inquiries that uncovered a litany of errors in the way the system was designed, run, and overseen. In December 2016, the Australian Tax Office was forced to restore a petabyte of data when an HPE storage system crashed, while in September the Department of Health was forced to pull a billion lines of poorly de-identified Medicare data offline after publishing what it thought was a sanitised dataset.