Australia's Department of Social Services pushing ahead with data-matching plans

The Centrelink “robo-debt” debacle hasn't dimmed the Australian government's enthusiasm for data-matching as a policy tool.

The Department of Social Services (DSS) has confirmed to The Register that it plans to go ahead with its own big data project conceived in the February 2015 “McClure review” (full name: A New System for Better Employment and Social Outcomes).

Largely lifted from a controversial program undertaken in New Zealand that willingly trades privacy for service outcomes, the review suggested applying big data to welfare recipients to identify and pursue individuals at risk of long-term unemployment or other forms of welfare dependency.

Part of that, as the department explains here, is the data release: “access will be provided to de-identified, DSS longitudinal welfare data which are one of the data sets used to underpin the Priority Investment Approach. The whole model will not be made publically [sic] available”.

As well as data matching, the program touches another potential flashpoint: anonymity. When academic researchers warned last year that Medicare data released at Data.gov.au could be re-identified, the government slapped a ban on such research without ministerial permission.

The Register asked the DSS if the early 2017 deadline for its data release would be met. We were told by a departmental spokesperson that the “limited longitudinal data” will be released “to eligible stakeholders” within six months.

“The first stage is through a password protected online research gateway, allowing researchers to access the data remotely. This will initially enable Australian Government researchers, and then other accredited researchers, to analyse the data in a Secure Unified Research Environment. Strict security and confidentiality protocols will apply. The gateway will be managed by the Australian Institute of Health and Welfare.”

The Department has a formidable task to properly anonymise the data set. As last year's Medicare data release showed, it's hard to anonymise point-in-time data such that it can't be traced back to an individual.

Longitudinal data captures an individual's interactions – in this case, presumably, with the DSS – over time, and is consequently potentially much more revealing and individual.

Given the controversies already engulfing the government in its handling of data, The Register also sought comment from the minister for social security, Christian Porter, about whether the government will re-think the program. We received the following response:

“Providing access to a subset of de-identified longitudinal data aligns with the Government’s commitment to optimising the use and re-use of public data under the Australian Government Public Data Policy Statement, released in December 2015. Data is integral in designing evidence-based policies and programs and allowing eligible stakeholders to access subsets of de-identified data will assist in developing and tracking innovative policy responses. Security protocols are in place to protect the data, and the handling of data is governed by the Privacy Act as well as relevant portfolio legislation.”

Vulture South has approached the opposition Australian Labor Party's shadow spokesperson for social services, Jenny Macklin, for comment.

+Comment: The devotion to data-matching is troubling on at least four levels.

First, it's become clear in recent years that decades of decimating the Australian Public Service has left it short of skills, either for in-house IT development, or to manage external contracts. The multi-million-dollar Census debacle, for example, was easily avoidable.

Second, as The Register noted in December, Australia's tech policies are years out-of-date, and the government's commitment to them is inconsistent at best.

Third, the Centrelink debacle suggests the government ignores bad news and is unwilling to abandon a strategy that's problematic at best.

Announcing his cabinet reshuffle this morning, Prime Minister Malcolm Turnbull dug in, saying that the debt recovery process was “entirely responsible and appropriate”.

Fourth, when policies are formulated, they seem to be explicitly hostile to citizens. Centrelink has been beset by IT issues, but the data matching swept up innocent people by design.

It's therefore hard to be confident that the DSS project will manage to avoid either controversy or failure. ®