What can UK data folks learn from Australia?
I've just been to visit a contact at CSIRO's Data61 and my head is buzzing with all sorts of learnings! I wanted to answer the question "what can UK data folks learn from Australia?" – and here's what I found.
1. A mix of private partnership and Government funding is working for Australia
CSIRO is the premier Australian science institution and is a highly regarded authority and innovator in many areas of research and development. They built part of the tech stack that is used globally for Wi-Fi and also developed the first commercially viable plastic banknote, for instance.
Going into the conversation, I assumed the CSIRO was an Australian public body fully funded by the state, and I wanted to know what made them so successful at innovation in that context. What I learned was that CSIRO's funding is actually only half provided by the government – they are required to make up the rest of their budget by delivering solutions for private industry. I think this funding model is a contributor to their success.
By interacting with the private sector, their need to solve real problems is far more acute and pressing than an institution in a pure research role. But their government funding allows them to follow less profitable paths that might be more in the public interest and gives them space to develop their industry solutions into public goods. Anecdotally, I was told that CSIRO try to only pick projects that will likely end in a public good such as a research paper and only go for pure consultancy as a last resort.
What's more, because CSIRO does still have a certain closeness to the Australian public service through things like cross-government professions, there is a natural osmosis of knowledge and technique from the private sector into the public sector via this channel.
Compare and contrast with Ordnance Survey, who despite being 100% government-owned have struggled to be a champion for innovation because of the need to profit from their most useful mapping data. Clearly, the ownership is less important than the funding model – which is probably not a epiphany for anyone familiar with the OS.
There is also DSTL, which is meant to play a similar role but only for military and defensive applications but its funded 91% by the MOD anyway. One could perhaps argue that universities play this role too, but as a sector they focus less on applying research and consequently only receive about 6% of their funding from commercial activity.
So we don't have an equivalent institution in the UK that I'm aware of. I'm interested as to whether a delivery-focused, part-funded science establishment could be a useful force for UK innovation.
2. Many of the grass-roots data problems are the same
I was somewhat unsurprised by the answer when I asked about practical issues data practitioners face on the other side of the world! The shortlist of issues contained a lot of the usual suspects – linking identifiers across datasets, properly anonymising personal data, and dealing with data quality and cleansing issues were all mentioned.
As I say, not surprising, but interesting to have it confirmed. But maybe there's a silver lining: if the problems are the same, maybe we can work together on the same solutions too.
3. An integrated approach to data sharing is working
I also learned about the Australian Bureau of Statistics (the ONS equivalent) and it's integrated data service called DataLab. It's a service that takes some of the highest value datasets from around government, links them together into a high quality data asset and then makes them available via a secure system. The idea is that ABS (and ONS) are the people with the right sort of skills and jurisdiction to be doing this work and are better placed than any individual department.
Given that the ONS has also been building a service like this, I was interested to find out whether the model was working for them because I have been personally sceptical of the ONS' plan. It involves centralisation and sharing of other people's data – something that is well known to be fraught with economic and governance barriers. It also is a multi-year piece of technology infrastructure with multi-stakeholder investment and high cybersecurity requirements, which is difficult to deliver in the current climate.
The Australians are several years ahead of the UK, and anecdotally the results are positive. The service is well-regarded and has been used to deliver some meaningful use cases. They've managed to overcome the challenges with data centralisation and sharing. Interestingly, they use the same Five Safes framework that has been used by ONS and UKRS to control use of sensitive data. All in all, it seems to be regarded as a success.
This is good news for the ONS' similar project – perhaps they are already collaborating with ABS, and if not, they should be!
4. The UK is doing more on data literacy
When I asked about data literacy in the Australian public service, I was a little surprised to find that there doesn't seem to be much mindshare for it. No doubt there are initiatives in central Australian government to promote data literacy, but they don't seem to have hit the headlines or advanced as far as they have in the UK yet.
I told my contact about One Big Thing – where every civil servant in any role must complete a day of data literacy training per year. One day may not sound like much but for many UK civil servants it was the first time they had ever had any formal data training. It's a big result and I think it's a sign that the UK is taking data literacy more seriously. My contact wasn't aware of any initiative like it in Australia. So the UK can pat itself on the back that it's leading the charge here!
And so that's what I learned. I was really grateful for the opportunity to visit CSIRO. Hopefully I'll be able to learn more from them at some time in the future!
Author
