Open sourcing policy models: our conversation with Dr. Audrey Lobo-Pulo

We sat down with Dr. Audrey Lobo-Pulo to unpack how open source models are being used in government to inform public policy. Dr Lobo-Pulo is the founder of Phoensight and a policy modelling enthusiast with considerable expertise in the Australian public service. She also has a strong interest in how data sharing arrangements between government, the private sector and not-for-profits may contribute to better social outcomes for citizens.

We spoke with Audrey before her sessions at FWD50 2018 to hear about her cutting-edge work in co-creating public policy.

What kind of role do you think the internet has in public policy?

If you think about what the internet really means, it provides a network and a way for people to connect together. To me, government can better leverage that in an organized and controlled sort of way. For example, we have Wikipedia; that’s a platform where information can be organized in a way that brings together vast views, but you still end up making a coherent document at the end of the day. Github is another example where multiple developers can contribute to a common project.

In the end, you have one specific project that’s been contributed to by a vast number of people. I wonder if government can use this concept and leverage it when they’re actually delivering public policy. I think that’s something that we can actually make a lot of strides in.

If I want to get a bit more futuristic, I would love to see the day where we could have virtual reality systems actually test out different policy environments. If we were coming to an election and there were two different policy ideas being put to us, wouldn’t it be awesome if we could put on virtual reality goggles to actually immerse ourselves in a few different virtual policy environments and make a more informed decision on that?

That would be quite interesting. I noticed that you’re very enthused and excited by government open source models. Can you explain more about what those are?

Of course. One of my biggest passions is government open source models — they are policy models that are developed by government and released to the public. Then the public can use, modify, and develop the models further. The biggest advantage of government open source models is that you are engaging with the public not just on the development side, but actually also inviting them toward participating in creating a better model ecosystem. The advantage of that is you can have much more evidence-based and meaningful debate.

A simple example that I like to give is: I have no idea how the GPS on my phone works. I still have a lot of confidence in using it. I might not always agree with the method that it chooses to get me from point A to point B, but I trust it to a large part and I will rely on it while travelling. I don’t see government open source models being very different from that. It’s not that important that the public understands exactly how it works. It can certainly provide feedback so the public understands what these things mean and where they’re heading. Part of our work in the evolution of these models is to make them simpler for the public to understand and interact with.

How would you simplify policy so that people can interact with it? What would be the first step?

I think it’s using technology as a tool to provide that level of simplicity. You may not understand fully how the policy works, but if you keep using the virtual environment you’re gaining the tacit knowledge underlying that system.

Another analogy would be driving a car. I know if I push on my accelerator I am going to speed up. If I go over a rock it’s going to feel a bit bumpy. I may not be able to calculate exactly what that means to me, but it’s a tacit understanding of the policy environment. What that means is that if someone decides certain benefits should be increased, if I have the right virtual environment, I should be able to relate what that increase in payment means to me. Would it mean that I’d spend more on junk food or put it into something for my future like books for education?

At the end of the day what we want to do is create a feedback mechanism where the public understands the policy implications. I’m not sure we’re demystifying policy, but we’re immersing people in an environment where they can understand the implications of those policies and make better decisions or vote more in alignment with their needs and desires.

Right. The public sees what the potential outcomes might be, and policy creators could also use the models to see if they’re being adapted in ways they didn’t anticipate. Who is currently doing this? Are there actually governments that are testing these out?

We’re in very early days. When I put forward the concept at the Linux conference in 2015, I don’t think there were any at all. Since then the New Zealand government has released a model that they call their Social Investment Analytical Layer code. That cost them like $140,000 in initial code development. What they have claimed is about one million dollars and nine months of work was saved by having the public contribute to this model.

Not only do they end up with a better product, they’ve saved money on the development of this code; they’ve engaged with the public and showed they are co-developing and co-creating this model. They’ve also saved on time. This is a good example of how engaging with the public can be a win on all sides.

On the Australian side, the Australian Treasury released Capita. Capita is a personal income tax and transfer model. It’s still in the early days, but we have a couple of think tanks looking at those models and their implications for different policy options in Australia.

We’re not in a place yet where we’re having people participate in virtual policy environments, but that’s certainly where the technology would head once it’s sophisticated enough.

Are there ways that government, the private sector, and not-for-profits can collaborate to share data? Are there things that even the private sector can do to move these models ahead?

I love that question because that’s my passion right now. At the moment, my thinking as to what the big question for policy developers and designers is: where do we get our data from? There are the obvious sources where government does collect its own data. We know of sources where other research institutions collect this information. The big unknown is: what data is collected out there that we simply do not know about?

This year LinkedIn worked with the Australian Treasury on a pilot project about tech entrepreneurship using LinkedIn data. Now, the interesting thing about this situation was that there was no exchange in data. LinkedIn did not exchange data with the Treasury. What we were able to do was have a discussion about the impacts of that data, which was an exchange of information on an aggregate level. So even though data wasn’t specifically exchanged, insights and information were. The Treasury was able to cross-check that information with its own data to verify certain claims and assumptions.

The new space we seem to be going into is having a conversation about what the insights in your information tell you about the social environment. The other big thing is about how we have those conversations in a meaningful way. How do we engage multiple parties that may not even be privy to this conversation?

We may begin to have a public consultation, but a public consultation isn’t usually based around data-sharing arrangements. This makes things a bit tricky. If you look at things from a micro-level, we’re clearly now in an era where we’re collecting enormous amounts of data at an alarming rate. Organizations have amounts of data that they’ve never had before. A lot of them are recognizing the value of that information. They’re leveraging their data assets to give them better insights into their own business needs. From a policy perspective, that’s useful on a micro level. What we’d like is for that information to somehow feed through on a macro level so we can have a better sense of what’s happening with big data from a societal perspective.

For example, company A might have a lot of information that might relate to company A’s business needs, but it may also be useful to address homelessness — useful data to address unemployment, for example. Currently, it’s only being used to operate in the context that they use as a business. They are unlikely to be using this data to address social needs. The question then is what role does government play- not just in identifying where that data is in society- but in how they could leverage it for better social benefit. How do we build relationships with these entities in a way that we can start having these conversations to provide better outcomes with society?

We’re trying to build a platform where other startups and organizations can leverage their data onto this platform to contribute to these societal issues. The idea is that this platform will be organized by issue. We want to have a private, yet safe way, for organizations to contribute to working toward these social issues with their data.

Right. Can you tell me a little bit more about the LinkedIn project?

So LinkedIn and the Australian Treasury presented at the Strata Conference in London this year. We looked at a specific case study of tech entrepreneurship that was done in five different countries over the last several years. The biggest standout takeaway for me was that tech entrepreneurship has a huge gender gap.

We’ve all heard about this anecdotally in the media and the news. What I recall: the best case scenario was out of every ten tech entrepreneurs, at most two of them were women. That begs the question: you’d think that entrepreneurship, being much more flexible in its nature than an employment role, would suit women better. They’d have more ability to leverage their time. The question is: why is there such a big gap? Being able to quantify to some degree how big this gap actually was is huge.

Having said that, LinkedIn more recently has paid a lot more attention to this social issue as we’ve been going forward. I think we can expect more from them in this space as well.

That’s excellent. What are you working on right now that you are really excited about?

Building this insight-sharing platform for not-for-profits and other organisations (Phoensight). There are a number of not-for-profits that collect information, particularly in third-world countries, where that information is not being leveraged to its full advantage. Looking at the problem from the other side, we see that governments, think tanks and academics would love to have this information. It would give them an opportunity to better validate their thinking, but also to develop better policy. On the one side you’ve got demand for this information and on the other hand, you have this supply that organisations collect that they aren’t fully leveraging.

The question then for Phoensight is: we’re trying to get in this market where this exchange can happen much more effectively. Making sure that not-for-profits can benefit not just for their organizations, but they can contribute to these theme-based issues that will be of interest to the government as well. It’s creating a platform where this meaningful exchange can take place on a macro level.

Right. Gathering information from the grassroots and figuring out how to leverage it.

Exactly. You might have a not-for-profit organization that’s in the market for looking at micro-finance loans. Now that might have some data that would be really excellent to combat poverty. They may not recognize the potential of that data for other things that may be quite useful.

Data collection is an issue. But once we get that data, it’s not just about applying it to the specified problem, it’s about getting it to the bigger issues that government and other think tanks are interested in. Each not-for-profit organization may be contributing a piece to that puzzle so that we get a bigger picture at the end of the day.

Is there anything else you’re interested in adding?

I love the concept of scaleable stakeholder engagement. Governments have done stakeholder engagement for many years. I think using technology to scale that would be excellent — like, the example that I gave about Wikipedia. Github has also been able to really reach developers that they otherwise wouldn’t have been able to reach, and still create coherent singular products.

I think that governments can really leverage this idea of scalable stakeholder engagement to engage with stakeholders in a way that we’ve never seen before. For me, that’s absolutely incredible. We might be able to capture niche thinking that we otherwise might have missed, purely because of the vastness of the internet and our ability to reach more people than we ever have before.

Would you see it as a tool like Github where people are all contributing and building something together?

How I see it in the first phase would be a government open source model that’s put out that everybody and anybody could work on. Ultimately the government would get a say on the final product they use.

It’s about tapping into the capabilities of individuals and capturing viewpoints that we would not otherwise have been able to. The ultimate in all of this is transparency. This is why I’m such a big open source advocate. I believe that if the technology is transparent and open, it allows the opportunity for people to check and validate the things that they want to. That makes it a fairer society.

FWD50 returns Nov 5–7, 2019, in Ottawa.