Let’s get started. Hello.

My name is Łukasz. I come from Warsaw. I work for “Gazeta Wyborcza” daily of Poland. I want to ask you about how you counteract fake news from China.

That’s right. Disinformation. Anything…

No. I’m just very interested in hearing about just the work that you’re doing generally, your priorities right now as a part of this government.

I’m Ankit Panda, by the way, from New York City. I’m an editor with “The Diplomat.”

Awesome.

I come from Germany. I’m a writer and journalist. I’m working especially in the branch of the small and medium enterprises…Digitalisierung in German.

Digitalization.

Digitalization, very good.

We are very happy to be here and hear something of what you are doing, Minister, on digitalization.

Would you mind if we switch places? It’s easier for you if I’m not in the line with the presentation.

Yeah.

Just as a brief introduction, I’m Audrey Tang, Taiwan’s Digital Minister in charge of open government, social innovation, and youth engagement.

Disinformation is indeed one of the issues. I gave a talk at this year’s IS about this. I’ll talk in very broad brushes. Then maybe you can just ask follow up questions.

This is my real office, by the way, the Social Innovation Lab.

It’s an open space. We tore down the walls. People can walk in and have 40 minutes of my time every Wednesday. That’s my office hour. It’s a very creative space.

The basic idea, very simply put, is that when countering disinformation, we want to strengthen the social sector and journalists. We do not want to alienate journalists from the administration.

There’s many nearby jurisdictions that counter disinformation through encroaching on the rights of the journalists, on the freedom of press, and so on. According to CIVICUS Monitor, Taiwan is the only jurisdiction in Asia that has not encroached any civic freedoms along this two year or three year of counter disinformation, especially on social media.

We intend to keep it this way, which is why we don’t use the F word, fake news. In Taiwan, news and journalism translates to the same Mandarin word, xinwen. Journalism is literally news work. Both my parents are journalists. Due to filial piety, I cannot use the F word, because then it will be like talking bad things about journalists. So we use disinformation.

Disinformation, we solve it through radical transparency, meaning transparency at the root, making the state transparent to the people. There’s less room for the rumors to grow. Whereas, of course, some nearby jurisdictions think transparency is making the system transparent to the state, which leads to a very different strategy which I will not go into.

There are three approaches. The first is what we call humor over rumor. We counter rumor with humor. In all the ministries, we have a team of five people of different skills. It’s like a standup comedian’s team. They are charged to produce within two hours, now usually within one hour. For each trending rumor we have an equally trending humor that is rolled out within an hour or two.

This is an example. There was a popular rumor that said perming your hair will be subject to one million dollar fine next week. Within one hour, the Premier, Su Tseng chang, rolled out this picture that says it’s not true. A young version of him says, “I may be bald now, but I will not punish people with hair.”

The fine print that says, “What we’ve introduced is labeling requirement for hair products that takes effect 2021.” A small print from the Premier as he looks now that says, “However, if you perm your hair many times a week, it will not damage your pocket, but it will damage your hair. If serious, you can look at me to see how it becomes.”

It’s pretty good humor because he makes fun at the expense of himself, not some other people. This went viral, like absolutely viral. If you search for perm hair fine, whatever, you find this humor. You don’t find that rumor.

We found that people who laugh about it is then vaccinated so that people will look at this and have a matter of fact debate rather than being incited from and go to outrage like “This is outrageous that the government is fining people for perming their hair.” That kind of emotion doesn’t happen if you laughed about it. That’s the first thing.

The second thing is that we have an ecosystem around collaborative fact checking. Taiwan has a member of the International Fact Checking Network called the Taiwan Fact Checking Center, or the TFCC. They publish fact check to the public.

They publish it so that on Facebook, for example, when people share any of these stories that’s already fact checked as false by TFCC, which is an independent social sector organization, it gets buried, meaning that it doesn’t show up on timelines anymore.

It’s not a takedown. It’s like moving an email to the spam or junk mail folder. It doesn’t consume people’s attention. If you go there, it’s still there with a link that says, “This has been fact checked false. Click here to learn more.”

How do TFCC know what to fact check? You need a signaling system, an end to end encrypted systems, like LINE, which is like WhatsApp. We found that a lot of the prototypes of the disinformation is prototyped on the end to end encrypted systems, because then it evades the public scrutiny.

Only when it evades to be particularly viral do the people who run interference campaign move it to a more public place. We want to know the virus when it’s still incubating. We rely on people to voluntarily…Just like you can flag something as spam, you can flag any LINE end to end encrypted message as spam or scam or disinformation by long pressing it.

It’s supported by the official LINE system. It then goes to all its partnering organizations. That’s how TFCC know what’s about to trend and where they can focus their journalistic energy on. That’s the second thing.

Finally, the third thing is about during election and a referenda we have honest advertisement as a norm. All the campaign donation and campaign expense is already published in the raw data form not in the statistics, but rather in its raw data by the separate branch of government called the Control Yuan, the control branch.

Because the Control Yuan published this, we see that many people do social advertisement around political campaigns that doesn’t count as either political donation or expense. Basically, it’s a way for foreign actors to work around our domestic campaign donation laws.

We talked to all the social media advertisers and say, “This is the norm in our Control Yuan. You either publish exactly as transparent as the Control Yuan data, or maybe you don’t want to run political or social activities as advertisements for hyper precision targeting during the election. Otherwise, you may face social sanction. We’re just saying.”

Basically, they then agreed, Facebook agreed, to publish exactly the same raw data in a real time way as the Control Yuan. Google and Twitter decided to not run political or social advertisements during election season because they have not yet perfected such a radically transparent system. That’s the three pronged approach. I’m done with that.

I’m just going to talk a little bit about one specific example, which is this fact check report. That was last November. There was a disinformation that says, “Hong Kong’s thugs’ compensation exposed. Kill a police and earn up to 20 million.”

This is obviously in a way to paint the Hong Kong protesters as actually anti democracy, not pro democracy because that’s becoming the defining issue in the Taiwan presidential election.

The Taiwan FactCheck Center see it’s trending, did this source search and found that in this disinformation package which was viral in Taiwan the photo actually came from “Reuters,” but Reuters says nothing about being paid to protest or murdering police.

Rather, it is an account from the PRC. Actually, it’s Central Political and Legal Commission, Zhongyang Zhengfawei, their main political propaganda unit’s Weibo account, plagiarizing the Reuters photo and adding this very deceptive caption. From there, a lot of remixes then became aware on the social media, which is why we call this a notice and public notice.

A takedown only hides the issue. What we want from this public notice is rather they understand that when they see this in their social media, they see a really good link that correctly attributes this to the Zhongyang Zhengfawei Chang An Jian so that this account and its associate account lose credibility when people view its propaganda items.

This is just one of the many examples, but I hope this paints a broad picture.

Any questions? OK, it’s pretty clear.

Good. More about my general work. I’m a horizontal minister. In Taiwan, in the administration, there’s 32 ministries, each with a vertical minister. Above or with the 32, there’s nine horizontal ministers, who are like ministers at large that handles cross ministerial issues.

My office is literally one delegate from each ministry, at most, that assembles my office. Theoretically, I can have 32 colleagues, but at the moment I have maybe 20 means that some ministry didn’t send delegates, such as the Ministry of Defense. I wonder why.

I know nothing about defense issues, but I know something about diplomacy because the Foreign Service did send people and things like that. People who are dispatched here, they are usually around Section Chief level. There’s also Director General from the National Communication Commission.

These people collectively decide what kind of mechanisms they want to foster that can promote trust in the public sector so that we can trust citizens more. Just one very quick example.

There was a disinformation around the mask distribution. There was a viral disinformation that says, “Press share and get a box of surgical mask for free.”

Of course, it’s a scam. Nobody really got mask by sharing it, but this is kind of thing that can go viral if you don’t have a transparent expectation of what really is the surgical mask distribution issue.

The delegates in my office worked with the National Health Insurance center and basically said that anybody can just go to any nearby pharmacy, show your NHI card and get three masks if you’re an adult, five if you’re a child. Then it’s locked for seven days. You have to wait seven days to get another batch.

Then we worked with National Health Insurance center. They opened all the stock level data of the pharmacy of surgical mask. Anybody can look at any of the 100 or so tools, like a Siri voice assistant chatbot or whatever, and find where within their vicinity is a pharmacy with sufficient stock so they can go there and collect without standing in a very long line.

This requires coordination from the National Government Council, which has the open data index and things like that. It requires coordination from the pharmacies and the NHI, Ministry of Health and Welfare. If people want to order it online, then it has to work with the Ministry of Taxation, of Finance, and things like that.

All these need to work together. This horizontal collaboration is how we show the state’s working transparently to the citizens so the citizens can trust each other more. That put an end to the speculation around surgical mask.

That’s just one very recent example, but there’s many mechanisms like that that were designed. Any questions?

No questions?

I’m overwhelmed with information. [laughs]

I’m just curious about some of these. Are you sharing these best practices with partners in other countries, for instance?

Yes. We work quite closely with many research organizations. There’s one called CrowdLaw that we work in partnership with the GovLab that’s at NYU. This is CrowdLaw. The best practice has been written like a how do you do crowdsource policymaking.

We contribute to the problem definition and solution identification stages. You can see some videos featuring yours truly from around the world of how to work this kind of crowdsourcing legislation and regulatory mapping.

We are not only contributing our best practices. We’re actually sending people overseas so that we now currently have two designers that are our colleagues, but one based in Bristol, one based in London. They’re based in UK.

They actively work with the international organizations and the governments there to share best practices but while being also our colleague. They regularly share their experiences applying our methodologies to the commonwealth counterparts. That’s one.

I’m also personally a board member of many international nonprofit organizations, such as the Digital Future Society, which is a think tank of the Mobile World Congress. Although that event is canceled, the think tank meeting did not. I just returned from Barcelona a couple weeks ago. That’s the think tank.

Just this afternoon, I’m going to meet with the CONSUL Foundation. CONSUL is this Madrid participatory budgeting and petition platform. They’re very popular in Spanish speaking countries. I’m also a board member there.

I’m also a board member in the New York City organization called RadicalxChange with Vitalik Buterin, the Ethereum inventor.

Basically, I’m wearing multiple hats. I’m working with the Taiwan government. I’m not just working for the Taiwan government. That is another thing.

What are some of your other priorities over the next year, let’s say?

Just today, I’m working on a tweet that basically talks about the Presidential Hackathon. The Presidential Hackathon is a priority because this year we’re really expanding to include all the 17 Global Goals. Of course you do know the 17 Global Goals, so I don’t have to iterate.

The basic idea is that this is just too much for a lot of people, like 169 goals. No goals left behind. Every year, we choose five concrete solutions. For example, these are people who detect the water pipe leaks. On average, they used to take two months from a leak that happens to where it’s detected.

It’s very difficult for them to recruit young people to join this kind of boring job. They worked in the Presidential Hackathon for three months with AI researchers. They co created a chatbot that these repair people can wake up and ask what are…

Just like what are the nearby pharmacy that still have mask, they can ask what are some nearby leaking points that has a high probability of leaking soon or now so they can go there.

With a 70 percent or more detection accuracy rate, if they visit three sites a day, it’s bound to be somewhere that’s actually leaking. They can work on the creative part of the job, not the routine part. They reduce the two months to two days.

In many jurisdictions, you have this kind of prototype fund. The problem is that if they solve for a small region, there’s no funding to scale it up or scale it out because it requires coordination across municipalities and across ministries.

Every year, the President give five awards to five teams that are really good ideas. These five teams each corresponding to one specific target in the Global Goals has this trophy with no money attached. It’s a micro projector.

If you turn it on, it shows Dr. Tsai Ing wen giving you that trophy and promising you whatever you did in the past three months we agree to help you get a data collaborative going about your idea and become national policy within the next 12 months.

It’s political binding power as the trophy. Every year, we have more than 100 teams proposing anything from measuring air quality together, measuring water quality together, increasing the marine debris detection offshores not until they hit our shores, and things like that.

We use quadratic voting, again a new voting method, to let people vote on the priorities in a way that reflect their true social preferences. The top 20 receive our coaching to become cross sectoral teams. We’re actively promoting this idea not only for domestic teams but also for people overseas to join.

Last year, we did a pilot. The Honduras team and the Malaysian team were the top winners. This year, we’re working with the AIT as a co organizer so that more people can join in this kind of data for public good activity.

If you’ve some questions.

You seem to be so much focused on very technical and specific issues. Do you take part in creating some long distance strategy for Taiwan?

Yeah. The long term strategy for Taiwan that’s a great question is basically crowdsourced. What we are trying to do is to make sure that at any given time hundreds of small experiments, each breaking one regulation or more, can co create at the place.

The strategy is basically a swarm strategy or open innovation, in a more beautiful term. Anyway, it’s an open innovation strategy. We rely on the crowd intelligence to tell us what are the rough consensus of the people. We use this to solve especially, what we call, wicked problems.

Like UberX first entered Taiwan in 2015, we asked them, “Why are you working with drivers with no professional driver’s license?” They’re like, “Oh, because it reduce carbon pollution.” Whether that’s true, we don’t know. We can crowdsource the data, but most importantly crowdsource what people feeling is toward that data.

This is a real conversation we run in 2015, where everybody can see the clusters of people holding different reflections about the same data. Data is the same, but allow for three to four weeks of people expressing their sentiment around the data. Then we crowdsource the idea. The best idea are the one that take care of most people’s feelings.

The UI looks like this. Somebody think passenger liability insurance is important. Now that’s a strategy. There’s a value statement. It’s not just about the technicality of insurance. It’s about we should protect the passengers even if they are UberX rider.

You can agree or disagree. As you do, your avatar moves toward the people who feel like you. It doesn’t pay to troll. First, there’s no reply button. There really is nowhere to troll. The second thing is that after pressing agree or disagree for a few times, you will be prompted to offer your own suggestions.

Even if you mobilize 5,000 people in, if they vote exactly the same, maybe there’s an extra zero here, but there is no change in the landscape. The landscape measures diversity, not headcount.

Every time we run the conversation, we always see this picture. This picture may be the most important takeaway. It’s that most people actually agree with most each other on most of things most of the time. If you only look at social media, you may think that those five divisive ideological positions are all there is — because that’s where people put most of their activity, energy toward.

Actually, when we run this…This was in Bowling Green, Kentucky. It’s a real conversation, a virtual town hall. No matter whether they identify as a Democrat or as a Republican, all of them actually agree that everybody need to pay attention to science, technology, engineering, and math in the light of the arts. The arts are important component in STEM education.

Nobody is against that whether they are for or against Democratic or Republican ideas. This is a way to find the low hanging fruit that we can just very easily make reality, to change our regulation to include the arts. It doesn’t cost anything, really.

Because of that, then people feel more like they’re a polity. It’s more like a pro social social media rather than anti social social media. It’s not my strategy. My strategy is more mechanism design, more participation, and for people to understand more that we’re the same polity and we’ve got to solve things together rather than pulling additional.

Are you already on the way to paperless government?

To what?

Paperless. Without paper.

Yeah, of course. We can do all the tax filing and things like that online. It’s an opt in. We’re not forcing people to use the paperless system. For example, there’s more than 23 million people here. Like three million people or more have filed their taxes and things like that online successfully. The satisfaction rate is 98 percent, which is unheard of.

The tax filing system itself is actually crowdsourced. It was a creation by people who complain about the tax filing system. This is in Mandarin. It says the tax filing software is explosively hostile.

This was a petition. Anything that has 5,000 petitions, we meet in a ministerial meeting to answer them. This is the petitioner, a professional user experience designer. We worked together with them, all the petitioners, on the user journey of before and after and during tax filing. We see this mess. We don’t harmonize our contributors.

There may be opinion like, for example, the words are explosively verbose or that the interface is baroque it confuses people and so on. We don’t censor these words. We make sure that if 5,000 people say the same thing, it’s just one Post It note. This is exactly the problem landscape. Through co creation workshops, four of them, we co created with the most toxic commenter online. The actual vendor and the civil servant…

They did accept your invitation to…

Of course. The minute we post this invitation. It used to be 80 percent saying that the Minister of Finance should resign. That was 2017.

After we say “Anybody who complain get a free ticket to the co creation workshop. If you cannot make it to Taipei, you can watch the live stream,” then it became more than 80 percent constructive feedback, less than 20 percent trolling. It’s really interesting how a simple invitation can do.

Basically, the first year, we only roll it out for Mac and Linux users. That’s 96 percent approval. The year after that, last year, it’s 98 percent approval. Not because it’s particularly pretty or something, although it is prettier than the old system, but because there’s thousands of people who feel that they have contributed at least one Post It note into this system.

They voluntarily work with their families and friends to coach them into the new system because they feel they own it. Now, with thousands or tens of thousands of owners, that’s how we get this approval rating. If we pay a very famous designer but doesn’t involve the citizen, even if we run something exactly the same, it will not have this high approval rate.

I understand that social engagement is priority for the government.

That’s my job.

Could we get a photo of your original bureau?

Of course.

Original bureau?

You showed us before. [laughs]

The Social Innovation Lab workplace.

Yeah.

There’s the one with the self driving tricycles around.

It’s really way before this was legal. It was already going on. This space itself is a sandbox. That’s where people bring barely legal new things to try it out so that the lawmakers there with 12 ministries get a firsthand experience of what the future is like. We can make a future to future conversation about it, not a past to future conversation about it.

Like we talk about alternate different future, but we’re not bound by our old experiences. This is important because these self driving tricycles, they’re co domesticated with the nearby market and community. They were originated in Boston, in MIT.

In MIT, when they have to yield, like two people you have to yield to one, they always yielded to children. Here, the consensus was we need to yield to the eldest first and then handicapped people and then pregnant women and maybe children. [laughs]

It’s important that they agree with the social norm when entering the society. Otherwise, people will feel like we’re being colonized. If we co create the norm first before we decide on the policy and the code, then people will feel, “Oh, we own it.” Again, this is a principle of social innovation and engagement.

A photo?

Ah, a photo.

Good photo. Maybe you have a file?

Sure.

Milky. [laughs]

I can send it to you actually. Maybe easier. Just maybe you have an email or something.

I have an email. [laughs]

Everything is online anyway. I’ll just send a link to you. Thank you.

Let me just grab my name card. My name card is also quite interesting. As I said, I’m working with Taiwan. There is no country name on the name card.

Thank you.

It’s a very cosmopolitan card. The other side is provided by our foreign service. It’s a free advertisement space for the MoFA.

I’ll send you the files.

Thank you.

Thank you.

We’re obliged.

Thank you.

Thank you so much.

Thank you.

All the best to you.

Thank you.

All the best. It’s all really impressive. Thank you.

Thank you.

Thank you for your time. It was very instructive. Thank you.

Thank you.