The proposals actually are written by the mid-level management themselves. We can tell by looking at the way they use language. However, they just hand it to their academic or social sector friends, who then propose it as a new idea.

That’s the main thing that they think about. Because of that, we always emphasize that it’s a time-limited experiment. If it doesn’t work, it doesn’t work. Presidential Hackathon is a good example, because everybody can just say, “We’re just reacting to the proposals from the social and private sectors.”

That’s the front line. For the middle-level management, the main worry is the political risk. If we implement an idea from the sandbox, from the Presidential Hackathon, from the petition platform, and it just explodes, does it create a political risk for our minister?

They feel like they own this, or at least co-own this new experience. They volunteer to help their friends and families with this new design system, because they feel it as their own. That saves time for the career public servants. They don’t have to play customer support anymore. They ...

We co-created the new tax filing system, which is a major point of this year’s administration, because it won 98 percent approval rate, compared to the old tax filing experience. This is not because we chose very good colors, fonts, or whatever, but because literally thousands of people felt that ...

The complainers, it turns out, can be very talented designers, because they know the most, they care the most, or they complain the most. We just invited everybody who complains in through telepresence, and also through face-to-face meetings, through paper prototype.

It shows that, if we can, instead of having people complaining after we roll out a service that breaks down— that’s actually how the USDS was formed [laughs]—we can bring the users, the citizens, into the very first line. Wherever people complain, we can invite them into our co-creation meetings. ...

It just turns out that they designed it for people who are very well-versed in the tax filing procedure, and people who are using Microsoft Windows instead of an iPad. There’s a large disconnection of what public servants feel as their job well done versus the people who have different ...

On our national petition platform, which is like We the People, there’s someone that says that the tax filing system is “Explosively, unbelievably difficult to use.” There are obviously different thoughts among some citizens.

Anyway, we have six language versions of what we call the PO manga, and we have printed version as well. It’s called “The Very Taxing Tax Filing System.” [laughs] Basically, it shows a front line career public servant enjoying the very smooth tax filing experience that they have.

There are, of course, different levels in the government. There are the front line, career public servants who look at this, and there’s a lot of fear, uncertainty, and doubt about, “Will it increase my workload?” “Do I have to fight trolls in addition of my daily work?” and so ...

The Pol.is system, which is the AI-powered conversation, is now a New York nonprofit called Math and Democracy. We have tight links with all those municipalities that try out similar ways of governance.

Basically, wherever there is a post-Occupy-ish legitimacy of this kind of crowd intelligence, we are part of the network, and we co-develop most of the tools. We certainly don’t develop all of our tools ourselves. The whole transcript system you just saw is from mySociety in the UK.

I am also a board member of RadicalxChange, which is a New York nonprofit that use quadratic voting. Those are the new inventions that we have deployed on the Presidential Hackathon here that we are also sharing with, for example, the Colorado State Congress. That also uses the same new ...

The best practices—or at least better practices, in various different branches of the government—of how to engage with the citizens make collective intelligence part of the policymaking process, instead of some ad hoc add-ons. We are co-developers in these ideas.

More personally, not speaking as Digital Minister, but as a lowercase minister for digital ideas, this is also my personal passion. I am on the international advisory board in the GovLab, which is Beth Noveck’s team. There are a lot of ideas, called crowd.law here, for which I just provided ...

That is to say, it’s a real trait of liberal democracy, or at least it’s moving that way, to share best practices. We are very active participants to the OGP summits of the past five or more years. Starting next year, we are formulating our national action plan for the ...

We are working closely with the Open Government Partnership (OGP), which is the international organization started in the Obama Administration, with a specific aim of uniting countries that pass a value check.

This basically makes the public sector something that incubates, or at least co-regulates, with the private and social sectors, instead of relying purely on the policymakers’ advanced AI knowledge, of which there is a shortage. [laughs] People have firsthand experience on how society reacts to projects.

If it doesn’t work, again, everybody learns something, but if it does work, through the specific act, or through the regulatory reinterpretations, which are all announced in sandbox.org.tw, unrestricted to those three … There are many, like 5G, vertical experiment fields. Projects are all announced here, so that people can ...

The crucial point is that, during those three to four years, the initial experiment, including the business model, is still legal, whereas all other competitors are not yet legal. This creates a first-movers advantage that we use to incentivize people to just go and try it out.

If it does work out, then at most two years after the sandbox experiment, we must make a new regulation for it. Of course, the MPs, at any point, can say, “Oh, it warrants a specific act,” in which case they have three or four years to make an act ...

We have a fintech sandbox act. We have a self-driving vehicle act. By vehicle, we don’t just mean cars. There are cars that fly, boats that drive, and so on. They get a year to prove it to society. If it doesn’t work out, then we thank investors for paying ...

Yes. We have a very comprehensive sandbox laws—because Taiwan is a continental system—so that we have to, as the lawmakers, to specifically carve out parts of the laws that we want to defer to the municipalities or to the space owners to decide.

All this then informs the law that regulates this thing. We don’t jump into the law, because that will inhibit the creative space, the solution space. We almost always just ask people to try it out in a sandbox. There are physical sandboxes. There are time-limited sandbox, like the Presidential ...

Then the entrepreneurs would understand what things actually fit people’s needs and design the market entries correspondingly to participate in the sandbox. That then informed the code, which is very important when we talk about machine learning, especially in decision-making protocols.

This kind of norm then motivated the tweaking, the hacking, of this cyclops-like design into a more two-eyes-that-can-read-human-emotions-and-emote-back design. This norm then informed the market policies.

Then the sandbox usually gets a year or so of experimentation. During the experimentation, people take it in whichever way, like the self-driving tricycle: people kept thinking it’s a shopping cart, so people actually tweaked the program into a shopping cart that follows people around. You can buy some flowers ...

You can see in the Pol.is conversation that there are a lot of interesting ideas around e-scooters. Then we asked the Ministry of Transportation, the Ministry of Communication, and so on, various lawmakers and regulators, to try it out in the sandbox, right before the consultation, so that people have ...

E-scooters, another great example, are being tried as a sandbox in the National Taiwan University as we speak. The vTaiwan platform, which is a civil society platform that initially deliberated Uber, asked all the different stakeholders to participate in the Pol.is conversation.

They’re self-driving, but they’re very slow. They don’t harm people. Because it’s open source and open hardware, people can tweak it however they want. This is our basic idea of our norm-first approach, of first asking the society, “What are the social norms around those emerging technologies?”

Once we set the norms through our fintech sandbox, our self-driving sandbox, or our platform economy sandbox, people have the idea of what it looks like to have such a conversation. As a real example, my office was a sandbox for those self-driving tricycles. [laughs]

Excellent question. We have a draft bill in the Parliament, probably to be passed soon, called the Digital Communication Act, or the DCA. In that act, it stipulates that all the multistakeholder issues in our era of, as the UN call it, digital interdependence, must be done in a way ...

Also, on the platform, people compete on resonance. The trolls must also compete on resonance, which makes it uninteresting for most of the trolls. That’s the basic design criteria.

Instead of focusing our attention on those five divisive statements, we should just ratify whatever people think is important across the different aisles. That was the idea of keeping the trolls at bay, basically: you can’t make personal attacks.

For example, for the The Bowling Green Civic Assembly, it turns out that, no matter which partisan side they are on, everybody agreed that, instead of STEM, the arts should be part of the K–12 education, making it STEAM. This is what they have not yet done, but this is ...

If you have a very divisive idea, you will not get agenda-setting power. People can still compete, but compete on getting more resonating ideas.

This is the real root of the UberX conversation in 2015, which is the first time that we deployed this technology. Again, you can see that this is not proportional to the people who vote the same way. This is not a representation in the sense of counting the number ...

In this way, if 5,000 people get mobilized, like fake accounts or whatever, and vote exactly the same way, it actually doesn’t affect the area. One area here is just 16 people, but it’s actually a larger area representing a more diverse point of view.

Once you click agree or disagree, your avatar moves toward the people who share your opinions, but there’s no way to directly reply to that particular statement. After agreeing or disagreeing for a while, you can then contribute your statement for other people to like or unlike.

With the reply button, people who have the most time on their hands win arguments by virtue of having more time on their hands. They can also make personal attacks. Without the reply button, which is very consistent in all the spaces we design, all we show people is the ...

The second question, which is the main thing that we are researching, if you look at New York Times article—this is basically what we’re saying: We’re keeping the trolls at bay by a mix of artificial intelligence and crowd intelligence. The basic idea, very simply put, is that we take ...

This is not a specific-to-one-ministry item. This is basically our national auditing office publishing their information and sharing it with people. It’s part of their work. It’s not an additional track of work.

The platform itself is managed by the National Development Council (NDC), which is a merger of the Economic Planning Council and the Research and Auditing Councils within the administration. Once these two merged, they are both in charge of long-term planning, as well as the national auditing day-to-day, and whether ...

Every quarter, there are good suggestions from people, and every quarter, the Ministry of Health and Welfare can say, “We relaxed these reimbursement rules,” or whatever, “based on your feedback.” This is an ongoing conversation with career public servants, instead of through specific ministers.

Public servants can now say: You just go to the Join platform, and look at our long-term healthcare, which is our number one watched policy. You can see the KPIs, the budget deliverables. This is a 10-year policy. We’re in the third year.

Most of the g0v innovations started just as these kind of experiments. Once the public service realized that this is actually a much better way, compared to the traditional way—which is each public servant has to pick up the phone, and the person is speaking and asking questions, not realizing ...

That is much more interactive, much more participatory, and is always open source, meaning that the creators relinquish the copyright claims. Because of that, people can then delve into the specific part of the budget that they care the most about and have a real-time conversation among citizens themselves.

If you go to budget.g0v.it, it looks just like budget.g0v.tw. there really is no trademark of g0v. The basic idea is that, for each government website that people complain is not useful or not interactive enough, we change the O to a 0, and you get into the shadow government.

Both excellent questions. The participatory budgeting, which is part of join.gov, is actually a civil society initiative. If you go to g0v.it, which is Italy, you will see the Italian people in the very beginning, like in Taiwan in 2012. Instead of asking the government’s permission, they just parsed through ...