Australia will conduct a census on August 9th and for the first time will retain name and address details in the data set created by the nationwide data dredge.
That's got privacy advocates worried that your data could be linked to multiple other government data sets, so much so that friend of The Register Jack Skinner has decided to absent himself from Australia on census day. His post explains why: the Australian Bureau of Statistics (ABS) has a form telling porkies about how it uses data, and has been less-than-forthcoming in explaining how it will use census data. Which turns out to be to offer it to just about every arm of government for cross-referencing, without disclosure of when and how it is used.
The ABS counters by pointing to its long and successful incident-free record of collecting and retaining data without incident. It also argues that the Census' policy-development-and-improvement aims are pure.
There are two things to consider here.
Firstly: most of us have already given away vast amounts of personal data by joining social networks, or by agreeing to use apps that can use our phones' cameras to watch us and their GPS kit to track us around the world. Some of us wear heart sensors that shunt data into the cloud. The providers of those devices and services have a breadcrumb trail of our lives, literally (thanks to location services) and metaphorically.
It's scary that we gave away all that data because the tax affairs of the companies you gave it to show they have few scruples. The data is also probably stored offshore, beyond the reach of the government of wherever it is you live. The data you gave away is also at least as sensitive as Census data and is probably stored under flimsy accountability rules so the app or social network you signed up for can exploit it.
Why else do you think Verizon bought Yahoo ! and AOL? The reason's right there in the press release: targeted ads.
Don't forget that third parties mine public data: check out my social feeds and you can probably predict the books and movies my kids have seen and want to buy next just based on the Harry-Potter-derived name of our family cat.
Which brings me to my second point: while noble in aims, data matching across government is worryingly pervasive.
That's certainly how I felt last week when I encountered the Australian State of New South Wales (NSW) Minister for Innovation and Better Regulation Michael Dominello.
At a Fujitsu event in Sydney, Dominello explained he's a big fan of data-driven policy development. And interventions based on data.
He explained that NSW has a scheme called “Home Warranties” that pays compensation to customers of builders who fail to complete jobs. Dominello wants to figure out leading indicators of project failures – probably missed milestones – and find a way to politely ask builders what's going wrong before things go pear-shaped. The outcome will hopefully be a reduced number of claims on the Home Warranty fund, which means lower insurance contributions for all.
That sounds great until you realise that local government and private enterprise supervise building sites and check on progress. So this worthy policy goal is going to require data-sharing across a diverse group of stakeholders. Which just makes for a larger attack surface, a larger number of potential leakers and an obvious potential for more complexity.
And let's not forget that the building trade has a vast web of interdependencies. If a material or key supplier essential to a job just isn't available and milestones are missed, that's not a red flag for a future warranty claim. Those interdependencies also mean that news of a probe will travel through the industry at speed. Which won't be great for reputations.
Then there's the honeypot a database of red-flagged builders represents. Gee it would be a shame if nobody turned up to work on your site, Mr Builder. But we can make that risk go away ...
Dominello tempered his remarks with numerous insistences that privacy must be done right.
But that call needs to be more than a promise. We need to know what data is gathered and what criteria create a red flag. We need to know exactly how and when agencies will be able to use data. We need to know how security will work for data sharing among agencies and private sector third parties. And we need oversight and frequent, transparent reporting of compliance with those data-sharing rules.
It would nice if this stuff were explained well in advance, rather than dismissively in a short campaign on the eve of a census.
We need these better explanations and assurances across all levels of government, because Dominello is far from alone in believing in data sharing. In recent weeks data sharing has been advocated as a way to curb welfare fraud and stage targeted interventions (sometimes executed by the private sector) in the lives of welfare recipients.
Yet as The Register recentlyrevealed, hackers need just your address to commit identity fraud or worse, using entirely public datasets.
Sleep-walking into acceptance of well-meaning government data use is therefore just as stupid as the sleep-walking into surrendering personal data to social networks.
And now that we've seen the latter in action, surely we should demand better of the former. ®