Skip to content
  • Blog
  • 15th Dec 2022

Australia: A government learning how to learn faster

Thoughts from a recent visit

BIT’s CEO David Halpern recently visited Australia to meet with various leaders and officials there. Here are his reflections on the behavioural challenges and responses taking place there.

After a week of events and meetings in Australia, one thing is for sure. The new Federal Government is in a measured, but determined hurry. 

Don’t wait around…

Prime Ministers and Premiers do not know how long they have. Both Blair and Cameron, Britain’s longest serving living Prime Ministers, have both spoken of how power declines while knowledge increases across a premiership. A key challenge for a new Prime Minister is to accelerate the latter: acquiring knowledge – and building a government machine to deliver – before the power and the moment ebbs away.

There is an abiding piece of advice for new Ministers and Prime Ministers – figure out your priorities, and early

Muthurkrishna’s work on what makes us smart shows the power of learning from those previously tried to crack a problem, even if they didn’t succeed. One newly appointed Minister, in the turmoil of the UK this year, had the wisdom to bring together every previous Secretary of State – from left to right – to have held the position to ask them what they learnt. None refused the invitation. It is a lesson others would do well to follow.

Australian Prime Minister Anthony Albanese has brought in some bold thinkers and experienced reformers to help. These include the much respected Glyn Davis to head the machine, and Gordon de Brouwer to help drive a rejuvenated program of public service reform. 

Clues to Davis’s thinking can be found in a recently published book, On Life’s Lottery. It conveys a deep passion for harnessing innovation and community mobilisation to find more effective ways of helping the most disadvantaged. His book starts with the image from Ursula Le Guin’s short story of a society whose prosperity painfully rests on the misery and isolation of a single child. This is not to be the model of Albanese’s Australia.

Intent into action…

As with so many early governments, there remains a gap between the ambition and the detail of how to get there. The new Australian Government has the framework of the Thodey review on which to build. But it is a framework that still needs filling, and success rests on bringing an army of public servants with you. Even if they are willing, the chasm between the intent of Prime Minister and Cabinet and the communities you wish to help – especially in Australia’s massive geography and cultural divides – is enormous.

But, Government does not need to go it alone. I was very struck by the energy and imagination of many of Australia’s leading Foundations to support innovation and change – to make this moment matter, for example in discussions with The Paul Ramsay Foundation, Vincent Fairfax Family Foundation, Trawalla Foundation and Minderoo.

The truth is, many Foundations and Third Sector organisations, with the best intent in the world, become caught in an endless struggle to plug gaps in service provision. They become further drawn in through being commissioned by the state itself. These forms of provision often become fossilised and stripped down. Despite the huge sums spent, both commissioners and deliverers are left with a lingering doubt that much of what is being done isn’t really helping. They are picking up the pieces, rather than finding deeper solutions. But some of the major service delivery providers are actively looking to work together to change this.  

Forward thinking foundations, like forward thinking Governments, can have a particularly powerful impact because they can create room to do something different. They can break out of mental and institutional priors. The Susan McKinnon Foundation, set up by tech entrepreneur Grant Rule and Sophie Oh, may especially be one to watch. 

McKinnon is planning to focus its efforts on what we would once have called ‘public administration’. Fix the machine, that’s there to fix other machines. It may not be fashionable, but it is clever. Here’s some quick reflections, echoing the three-fold focus of McKinnon, and the path that the APS will have to tread.

Every government needs Ministers and advisors who understand the value of smart use of data and evaluation

Effective elected representatives

There is an abiding piece of advice for new Ministers and Prime Ministers. Figure out your priorities, and early. There are things your Department will do just fine: let them get on with those. There are things that you’d like to do, but where the battles are just too big: leave these for tomorrow. But there are things that you are elected to do, that your Department won’t do by itself, yet you could make happen: these are your target.

Yet how would you know what these are? Here steps in the plague of ‘prediction errors’ and over-confidence (ie we predict that we are better at predicting the future than we are and the more senior we become, the worse these prediction errors become), and the deepest secret of government – that most of the time, we don’t really know what works.

The good news is, at least some of these gaps can be plugged by learnable skills. BIT (and Philip Tetlock) has shown that specific exercises can enable us to be better ‘calibrated’, and prediction errors reduced. Similarly, along with learning from war stories and the experience of others, leaders can learn the methodological techniques that can answer the ‘what works’ question. 

It’s not that we need our leaders to be running regressions in time they don’t have. It’s that knowledge of these techniques can sharpen their ‘bullshit detectors’. Crucially, they can then also push the public servants around them to up their game. 

I sharply remember when Nick Clegg, then Deputy Prime Minister, intervened in an argument between two major Departments over a proposed innovative idea. ‘The thing is’ he said, ‘neither of you know if this will work. So why don’t we run it as a – what do you call it – a Randomised Control Trial.’ He was right. But no one expected him to say it.

The new Australian Government is lucky. It has Andrew Leigh, an exceptionally sophisticated methodologist (and Randomista) as a Minister in Treasury. It should use him. Every government needs an Andrew Leigh. Or at least Ministers and advisors who understand the value of smart use of data and evaluation… 

Robust state institutions

Government Departments are, in general, massive legacy machines overseeing even more massive legacy public services. They do almost no R&D to find better ways of doing what they do. So we shouldn’t be surprised that they will do tomorrow what they did today.

There are of course innovators. (The hugely talented Victor Dominello being one in NSW.) But no Minister should rip up everything at once. People depend on the services governments deliver, imperfect though they are. Rather, governments and public service systems need more ‘innovability’, as system modellers call it.

This is the characteristic of systems to try out alternative paths, but without catastrophic failures. The genetic code has such characteristics (as I’ve learnt, rather appropriately, from my research scientist son and the work of Nick Lane and others) – but most of public service does not. This is why this type of radical incrementalism was intentionally baked into the DNA of The Behavioural Insights Team. 

Ironically, robustness comes from the ability to break and remake. We need government Departments with serious innovation teams and budgets. Departments that map what they don’t know, and plug those gaps. Departments – or partners – that ‘mine the variance’ to identify teachers, social workers, police officers, and community workers who achieve more, and figure out how they do it. And we need Departments that restlessly evaluate, scale and evaluate again – just as the UK’s Evaluation Task Force is learning. (For more details on the 10 things that government evaluation task forces need to do see a forthcoming blog based on my reflections as the UK’s National What Works Advisor). 

Well-designed institutions don’t divide the pie – they make it bigger

Quality policy dialogue

Finally, governments need better ways of engaging, learning, and partnering with their citizens. This includes getting well beyond stale, conventional polling – which fails to draw out trade-offs – or small sample focus groups – which lack the statistical representativeness to bear the weight of decisions placed on them. There are so many more tools, but most are barely used by governments. These include, digitally enhanced ‘delphi’ and crowd voting methods that enable group interaction. And the whole family of deliberative mechanisms that BIT and others are continuing to evolve.

My last day in Australia this trip was spent in Melbourne. It was there, a few years ago, that BIT worked with VicHealth on a deliberative forum to ask Victorians what they thought should be done about Australia’s growing obesity problem. They got to hear a range of evidence from experts and industry. They got to call forward more evidence where they wanted it. And for two days they wrestled and developed their own list of what they thought should be done – including the recommendation for a 20 percent sugar tax (yes that’s right – citizens asking for a tax rise).

It was too radical for the Australian government of that time. But a government should be working with citizens and communities, not ‘doing to’ them. It should dust off such mechanisms and be prepared to trust its citizens – in discussion with each other – to find solutions that they own. 

Power isn’t just a pie to be divided – where if leaders get more, others get less. Power is fragmented. But well-designed institutions can assemble that power, identifying common purpose and shared interest, that enables disperate actors to move together to get something done. Well-designed institutions don’t just divide the pie, they make it bigger.


I’m human too. Perhaps it’s just optimism bias. Australia has dabbled with the ‘What Works’ agenda, established internal Behavioural Insights Units and invested in other elements of public service reform and innovation. Now it’s getting serious. 

Australia today seems like a country at the precipice of bold reform. I hope it succeeds. I hope it provides a case study we can all learn from.