 Welcome, everyone, to today's Institute for Government event, How Taiwan Became a Coronavirus Success Story. I'm Gavin Freeguard, Program Director for Data and Digital Government at the Institute, and it's a good morning from me here in London to all of you. And a very good evening to Taipei from where I'm delighted to say we're joined by Audrey Tong, the Taiwanese government's digital minister. We'll hear from Audrey shortly, but first some digital housekeeping. If you'd like to tweet along with today's event, you can do so using the hashtag IFGDigital, and we'll be live tweeting from at IFG events. If you'd like to ask Audrey a question, you can submit them to me in three different ways. First, by using our Slido, where you can also vote on the questions that you like the most, you can get there using bit.ly slash ifgtong, as well as the longer link you'd have had in the event invitation. Second, you can use hashtag IFGDigital on Twitter. And third, you can drop a question into the chat on this livestream broadcast. And if you use Twitter or livestream, one of my colleagues will drop that question into Slido. Video and audio of this event will be available on the IFG website afterwards. So how did Taiwan become a coronavirus success story? Here in the UK, debate rages as to whether the death toll is closer to 40,000 or 60,000. Whereas in Taiwan, a country of 24 million people with lots of travel to China, there have been fewer than 450 cases and only seven deaths. While many countries have struggled to negotiate their way into and out of lockdown, Taiwan has kept its schools and businesses open. Whether UK has had debates about personal protective equipment procurement and wearing masks, donating them as humanitarian aid, even the Taiwanese response to panic buying toilet roll was rather different to ours, something I'm sure we'll come on to. So how has Taiwan done all of that? Well, answering that question, and many more, is Audrey Tong, digital minister in the Taiwanese government since 2016. Audrey dropped out of junior high school and had founded her first business by the age of 16, later becoming an entrepreneur in Silicon Valley. She returned to Taiwan and joined its vibrant civic technology community, a community responsible for projects including Gov Zero, an open source open government collaboration, and V-Taiwan, a pioneering public engagement platform. Since Audrey became digital minister, even more of that civic hacker expertise and way of working has been brought into government, with Taiwan seen as a world leader on everything from open data to open consultation. So I'll start asking Audrey some questions. I'll then start taking your questions and putting those to her as well, and we'll be finishing at around 10.30 British summer time, so just under an hour. So without further ado, Audrey, thank you very much for joining us today. How did the Taiwanese government first learn that coronavirus might be providing a problem that you needed to deal with? Yeah, hello, and have a good local time, I guess, everyone. Today marks eight weeks since our last locally transmitted case, meaning that as of today, it's the start of our new way of life post-COVID. So I first would like to express my gratitude for best and foremost to the collective intelligence system. In Taiwan, we have three principles of social innovation, fast, fair, and fun, which all contributed to the counter-coronavirus efforts. And collective intelligence relies on the society to report on our new way of happening, and we have in particular Dr. Li Wenliang, the PRC whistleblower, to thank, because Li Wenliang shared that there may be new source cases, actually seven confirmed source cases back in last December, and whereas many jurisdictions began heating that early this year, Taiwan started last year. So as you can see in the Taiwan equivalent of Reddit as called the PDD, someone with the nickname Normal Pipe reposted Dr. Li Wenliang's message on the very early hours of December 31st. And because of that, we immediately, it went viral on the PDT, meaning that people upvoted it. And because of that, our medical officers noticed this post and issued an order the very next day that says all passengers flying in from Wuhan to Taiwan need to start health inspection the very next day. And so this says to me two things. First, the civil society trust the government enough to talk about possible new source outbreaks in a public forum, because we're according to civics monitor the most open society in the whole of Asia. And also the government trust citizens enough to take it seriously and treat it as if such happened again, something we've always been preparing since 2003, including setting up at the central epidemic command center even before we have the first locally confirmed case. So they run this daily press conferences until yesterday when we officially declare we're now in post pandemic. Excellent. Well, congratulations on being post pandemic and something I know those are set in the UK would welcome happening quite soon. You sort of found out via the sort of Taiwanese Reddit. And how did the how did the government sort of swing into action? What what did you put in place to be able to deal with everything? Sure. So the CECC, the central epic command center, epidemic command center is the core, the cornerstone of this rapid fast response system. Because for many months, ever since January, they hold a daily press conference, which is always live streamed. And we work with the journalist community and making sure the CECC answer all the questions. And this is a new structure. Back in 2003, when sauce happened, the municipal government, the local officials, the medical officers from the central government all said very different things. And I think that's a response response. So post sauce, we're like, yeah, 37 people dead is 37 people too many, we need to run yearly drills. And as if starts has happened again every year, and increase our response system. So anyone with a telephone can call 192 and learn about the latest CECC announcements. And so any new idea to the CECC, which always gets responded, the very next day, for example, there was one day in April, where a young boy said, they don't want to go to school because their schoolmate may laugh at him for wearing a pink medical mask. And the very next day, everybody in the CECC press conference started wearing pink medical making sure that making sure that everybody learned about gender mainstream, which is again a social innovation. So this kind of guaranteed fast response, users trust between the government and the civil society, making sure that must use skills around hesitation around physical distancing, and our value and so on. People watch these daily 2pm press conferences and bit by bit become kind of amateur epidemiologists that understand the underlying science, which is again very important if you're going to mobilize the whole society without relying on top down lockdowns, which we never did. So how did you avoid those lockdowns? I think your sort of slogan for your response is fast, fair and fun. I wondered if you could talk us through that. Sure. So the fast part I talked about and the fair part has a lot to do with the GovZero movement, which you already mentioned. The idea of GovZero movement is very simple. For every government website, such as join.gov.tw, people who don't like this service can build a shadow service just by changing O to a zero. And that's it. And so without paying for advertisement or anything, people learn that if you change Gov.tw to Gov.tw, then you get use of the shadow government that uses the same data but presented in a much more interactive and participatory way. And so in fairness, for example, when we ramped up the face of mass production, making sure everybody can use their national health insurance card to collect masks from nearby pharmacies, fairness is of course our guiding principles. However, even before the government figured out how to make sure that people can see that this is being fairly distributed, this is a civic technologist from Thailand with the name Howard Wu, who builds a map of all the nearby places that sales masks and relying on citizens to reply to the current stock level, whether they have sold out or not. So this is not unlike Ushahidi or other open source crowdsourcing platforms. However, because he did not anticipate that a lot of people, millions of people would end up using this service, he very soon owed Google about 20K euros in API usage fees and have to close down that operation because he could not fund it himself. But the two days of this map running is sufficient to build a social sector consensus that this is exactly what's needed for ensuring fair distribution. So I showed this map, Howard Wu's work to our premier, Su Zhenchang, and he immediately saw the value of it and said that we need to dedicate our government resources to make sure that the citizens don't need to pay for API usage fees by themselves. We should provide the underlying maps, underlying longitude led to all the APIs and trust the citizens with these urban data. So not only do we publish the stock level of all the pharmacies, as you can see here, the green ones are ones that still with a lot of masking stock. We also have a real time updated every 30 seconds at a time for all adult and all the children's masks available in all the pharmacy and it's all completely automated. And so for even people who don't like viewing maps, maybe people with blindness, they can use voice assistants, they can use chatbots and all of them get the same inclusive access and because Taiwan has more than 99.99% of health coverage, people who show any symptom would then be able to take the medical mask from a nearby pharmacy, go to a local clinic, knowing surely that they would get treated fairly and without incurring any financial burden, but it doesn't stop there. The civil technologist also made a list that lets people see that our supply, for example, when we started rationing like three masks per week around this time, we increased to be nine masks per two weeks for adult and 10 for children. And this gets visibility from everybody because this is not a government website, this is just a civil society's contribution. And this also showed us where in Taiwan do we have an oversupply or under supply and we co-designed this experience with the pharmacy and we show this to our premier every week. So we can see him smiling happily here because according to analysis, the accessibility of masks at that time peaked at 70%, meaning that there's like 20% of people who never collected masks. It turned out that they are in large municipalities in the north, they work very long hours, they started working when the pharmacies have not opened, they stopped working when the pharmacy have closed. So we had to work with convenience stores which opens 24 hours a day, which can use the same national health insurance card to go there and collect your mask anytime. And at that time then, we're 23 million people, right? And now 21 million people have used the mask service one or another, which means that more than 90% of people not have access to medical masks and thereby ensuring they are not valued, that is under one, that is controlled. So because of that, we ensure the fairness through the feedback from the social sector and also collaboration from the economic sector. So it's a sort of harnessing the collective intelligence to improve your responsibility. One of the things that we've been talking about quite a lot in the UK is around contact tracing and particularly whether that sort of manual contact traces and sort of doing things via interview and people sending their contacts to our national health service or whether it's using an app. What's your approach being to contact tracing in Taiwan? Yes, in Taiwan, we make sure that all the information that we collect, there is a civil society website called Taiwan Can Help That Us, that this is not a government website, all of it is crowdsourced and crowdfunded, that you can look at our designs. Basically, we rely on data that's already collected, for example, the cell phone strength data, to make sure that people who return to Taiwan in an airport, they have the choice of going into a quarantine hotel, in which case they're physically bought from leaving that hotel for 14 days. Or if they don't live with vulnerable people, like very old people, they can also choose home quarantine, in which case their phone is basically put into a digital fence. And if their phone leaves the 50-meter radius, that is the triangulation resolution of the perimeter, then SMS is sent to the local household manager wardens or the local police station who will then check of what happens to you. So basically, we don't collect new data, we reuse existing data and use it in a way that sends SMS automatically, kind of like how before an earthquake or after a heavy rainfall with an automatic warning to a geofence. And this is called a digital fence that retains the data for 14 days, and after which, of course, there's no constitutional basis for us to run the data. But that data is already collected by all the telecoms anyway. So we see this as proportional, it's ruled already as constitutional by the constitutional court after a source. And this, of course, beats the 2003 response, which is to barricade the entire hospital unannounced and with no fixed termination date. And this is also important because we then never had to rely on application level tracing, which only makes sense if a majority of people start installing it. The cell phone tower, triangulation, and digital fence works regardless of which phone you're using. Again, we've had a lot of discussion in the UK and elsewhere about the sort of ethics and privacy debates about people's data and how the government's using it. And obviously, some of that sort of digital fence technology, you mentioned that it's deleted quite quickly, but it can be very intrusive. How have you dealt with those sort of privacy debates? Well, if you don't like the digital fence and you prefer to stay in the quarantine hotel for 14 days, that's your choice. And we even pay you a stipend either way of around, I think, 100 euros per day. But if you break the quarantine is a thousand times that fine. But in any case, so the point is that, of course, being put into a quarantine hotel is also an intrusion on pretty much anything, on the freedom of movement and so on. And because the digital fence is not GPS location, it only knows the kind of general perimeter of the phone. It doesn't know like which room in your home you are in. So the basic idea is that it should be proportional, which is more or less like if you're staying in the hotel, the hotel, of course, make sure you cannot use the elevator to go down. So that's roughly the same plan. But we do understand that not everybody supports these measures. The CCC, the latest numbers, 94% of people support these measures. Previously it was 91% when the digital fence was being rolled out. So we thank the 6% or the 9% of the population, which keep us honest and accountable because we have never declared a situation of emergency. We're still operating entirely under constitutional law limit, which means that every administration protection tactics we take, we need to be accountable to the MPs and to explain it to people. So if people understand how it works, the science behind it, that's how we get the 91% of support or 94 now, but we still thank the remaining 9% or 6% for keeping us honest. How is all the work that you've been doing over the last decade or so to engage the public, opening things up to consultation, how much does that help build that sort of trusted relationship? A lot. A lot of our communication strategy, for example, was developed way before the coronavirus. It was developed as a way to counter disinformation using a tactic that we call humor over rumor. And the idea is that because we cannot do takedowns. Takedowns is against the principle of Taiwan being the most open society in the whole of Asia. So we're forced to innovate, to counter the narrative of conspiracy theories of which there's a lot, especially during pandemic, but also leading up to an election without resorting to administrative takedowns and encroaching the journalist freedom because we want to work with journalists, not against journalists. So what we have discovered is that the viral messages that are toxic and polarizing is a little bit like a virus in itself, kind of a virus of the mind. And the reason why it has a high R value, that is to say a person will look at it and just automatically share it to many people, is because that it provokes a sense of outrage. If you provoke a sense of outrage, people share it much more than if it provoked other kind of feelings. And so how do we vaccinate against outrage become the main question. And we discover that if we can respond within two hours, a fun message that capitalizes on the same keywords, but makes people laugh, then it's mutually exclusive with the feeling of outrage. It's actually impossible to feel outrage about something that you've already laughed about. And so for example, during the pandemic, it's a stressful time. People feel anxious, lots of panic buying, lots of conspiracy theories. And there was a panic buying of tissue papers. There's a rumor that said, well, Taiwan's been ramping up the mass production from two million a day to 20 million a day. It's a quote, same material, unquote, as tissue papers. So people panic buy. So the same premier who smiled happily here with the convenience stores, pushed this within two hours. And now showing his bottom, wiggling it a little bit, and says a very large print that each of us only have one pair of botox. And so meaning that we don't need to panic buy. And a clear table that says, well, the facial masks are made out of domestic material, while tissue paper are imported using South American material. And this went possibly viral because this whole meme, this design, is literally a tissue paper box. And so people, and they are very interested, curious even, because that's the first time that the premier made himself a butt of the joke, so to speak. And so because of that, it has a much higher R value than the conspiracy theory. And people who have laughed about it will stop believing the conspiracy theory. So that rumor died down within a day or two. And finally we found out a person who spread the rumor at the first place was the tissue paper reseller. And this is not just a single shot. Social media work, every day the CEC daily press conference gets translated by the spokes dog of the Ministry of Health and Welfare or Zongchai, the doga CEO, the Shiba CEO. And they translated, for example, this is physical distancing. When you're outdoors, you need to keep two doga away from one another. When you're indoor, you have to keep three doga away or hand sanitation rules. Remember to cover your mask in this when sneezing or wearing a mask reminds you not to put your hands through your mouth as the dog does here. So remember to pre-order your mask. And all of this factual humor that have a scientific basis spreads faster than boom and that is how we make sure Taiwanese people feel calm and collected even during the pandemic. Excellent, thank you. We touched a little bit on some of the work that you've done before the pandemic around digital government and public engagement in Taiwan. I wondered if you could tell us a bit more about your sort of general approach to social digital innovation as digital minister. Sure. So this is my office, literally my office, the social innovation lab. It is in the heart of Taipei, just near the Da'an Central Park in the Jianguo flower market. And this is literally a park. We tore down all the walls so people can see very transparently as I work. And people can just knock the door every Wednesday and have 40 minutes of my time and just chat about pretty much anything as long as they agree to publish a transcript or the video online. And because of this, I'm very accessible. People can see that whenever people feel that there is kind of a tension between say economic development on one side environmental protection on the other or scientific innovation on one side and social justice on the other. Instead of relying on different siloed ministries, each ministry that's participating in the social innovation have a second mint in my office. So my office is literally from like 12 different ministries and each of them agree to work out loud, meaning that all the new ideas that they develop get spread automatically to the people. So when people come up with a new innovation, a social innovation, for example, self-driving tricycles, they can work with nearby flower market to ensure that instead of a technologist dictating the social norm, it is a society working with those self-driving tricycles to say, hey, maybe in the flower market, these can be kind of self-driving shopping carts that follow people around you buy some flowers, you put into it, and it follows you. And that is not the way that it was originally designed. But through the idea of an innovation, we can change a lot of that part and make a co-design, a collaborative design, so that it then responds to the social norms. This is called norm first design, or in SDG terms, 1717, encouraging effective partnerships. And all the best ideas, we choose five ideas every year that receive a trophy from the president. And the trophy is a micro projector. There's a shape of Taiwan and a micro projector underneath it. And when you turn on the micro projector, it shows Dr. Tsai Ing-wen, our president, handing the trophy to you, which is very meta, is a self-describing trophy. And that symbolizes whatever you have built in the past three months in a data collaborative. We do everything we can to make your idea into national policy within the next 12 months, so that ensure the availability of cross-sectoral data and making sure that whether it's air quality, water quality, sustainable education for global citizenship, all of those great ideas are voted into existence using quadratic voting, which is, again, another social innovation. So basically, it's a system to make sure that the best idea gets amplified automatically into the national scale. I've got a quote from your president here, actually, which is, do it bravely dare to make mistakes. That's right. I think people often find governments can be quite risk-versed because the nature of the sort of world they operate in. How do you overcome that? Yeah, it's actually a UK idea that we learned. It's called a sandbox, whereas the UK initially applied this idea to FinTech. We have applied it to pretty much everything. And so it could be a platform economy sandbox. The social innovation lab was a sandbox for self-driving vehicles. Now, the self-driving buses are going to the street of Taipei City. There's 5G sandboxes for, especially above 6 gigahertz, for a limited place to try that use of 5G. There's actually, our sandbox regulations is very such that you can challenge each and every ministry's regulations except for two things. There's money laundering and funding terrorism, because we know what happens. We don't have to experiment on these two things. But otherwise, everything is fair game. And so because of that, the government basically worked with the innovators. I tour around Taiwan to the places that are least connected. That is to say, who are most difficult to travel to my office at the social innovation lab. And the local co-ops, the local social entrepreneurs, the local elders in both sense from the indigenous nations and so on, they just, in a regular town hall, they just voiced their concern. The difference is that because we have broadband as a human rights, no matter where I am, for 16 euros per month, there's a limited 4G bundled with at least 10 megabits per second. Otherwise, it's my fault personally. And so through those cultural translators, we connect to the people in the social innovation lab. As I mentioned, 12 ministries, each of them section chief or higher level, who listen to the local people of what they truly need. And this has two benefits. First, that it doesn't get lost in translation. They hear exactly what the local people want. And second, if the local people want to try something that's not like a great area by the current regulation, they can say, okay, let's try it out for a year. And it doesn't work. Well, we think the investors for paying the tuition, basically, everybody learns something where the investor lose the initial investment, like reverse lottery. But if they wax, then they get a first mover advantage. And then we have a new law or regulation that's co-created by people. But of course, the obvious question is that, then how do we know at the end of that year, it's a good idea or not? How do we listen at scale and get the people's feedback? Well, we also use AI for that. There's an AI power conversation by AI, I mean, assistive intelligence called Perlis, where let's people see what other people feel about any particular issue. For example, during the Uber case in 2015, we shared the data. We asked for three or four weeks of what people feel about it. And the best idea are the one that take care of people's feelings. And finally, we turn it into a regulation. So at that time, no matter whether people is pro Uber or against Uber, actually, everybody agree that passenger liability insurance, their registration, and the taxation are the most important time. And so we made that into our new regulation. So every time we run Perlis, we see that while the social media and some institutional media may over focus on the ideological divisions, we don't actually touch that. We acknowledge that, but we just work on the consensus statements, which are something that everybody agrees with their neighbors about, and we just regulate those into existence. Nowadays in Taiwan, Perlis is regularly used. Every civil servant learns that they can just run a Perlis conversation, wait for three weeks or four weeks, and voila, they have a set of rough consensus that they can base their regulation on. This is actually one of the first questions that somebody's asked. And just to remind everybody, I will be putting questions to Audrey very shortly. So if you'd like to ask any, you can use hashtag IFG Digital on Twitter. You can use the chat on this live stream broadcast, or you can go to bit.ly. So Nathan Young sort of said, he read that you'd use the VTaiwan platform, which uses Perlis to mediate between taxi drivers and citizens. And that was a few years ago. So what new steps have you been taking on crowdsourcing policy since then? Yeah, we've been since combining the use of Perlis of face-to-face deliberations, much as what I showed in this teleconference room, where we bring all the five municipalities and many more cities into the same large virtual room. We make sure that we host such an open collaboration session every once in a while, actually two times per month. And just actually this morning, we're talking about that last year, we used Perlis, five Perlis actually, to talk about how to open up the hiking, the mountaineering, the restrictions about hiking. And now we're talking about opening up the sea and how those sea sports and so on need to be opened up for more people to understand the ocean and preserving the sustainability of the ocean. The Taiwan has like 10% of the world's marine biodiversity. And all these are of course very open-ended questions of which Perlis is perfect to do. And so we combine Perlis with this national join platform. It's called Join the GOV, the TW, which is a single platform, a single-stop platform for e-petitions for regulatory pre-announcements and consultation, like they're opening up mountains and the ocean. And also most importantly, it's also a participatory budget platform. So it's the same platform. And you can see how each and every ministry is doing its budgets. You can see that long-term health care is at the moment the top thing that people are concerned about. You can see the KPIs, how much they're working. And there's comment board, of course. And then there's also quarterly responses that says that how we have changed our ongoing policy. So this, while we Taiwan was mostly about like pre-legislation, we now have completely full life cycle. So that even during the 10-year project, that's a long-term health care project. We're on only the third year. You can between 2017 and 2026 work in real time with the public servants here. And they only answer once publicly, and everybody can very easily discover like how this presidential promise, how this long-term project is doing. And so a single Join the GOV, the TW is our main change after the initial prototype in the Taiwan. Now pretty much all the regulations and most of the draft bills went through this public consultation process for almost always 60 days. And so that's the norm. If they fast track it to 14 days, they have to write why. And so the norm is to do a public consultation for each and everything. And so because of this Join the GOV, the TW has more than 10 million unique visitors out of the country of 23 million people, that's almost half of the people. Excellent, thank you. I'm going to start taking questions from Slido now. So people just remind everybody watching that you can, as well as submitting questions on Slido, actually vote for your favorite. So I'm going to start with the top-voted questions so far. This is from Gavin Heyman of the Open Contracting Partnership, which is, how do Taiwan handle emergency procurement? And how do you keep it transparent if it involves sole sourcing? And how do you make sure that you sort of get the resources through to the front line and make sure that they've made it through to the front line? That's an excellent question. In particular, because we have not declared the emergency situation. And so all of the things that I show you was done using regular procurement regulations. And in our regular procurement regulation, there's something that says if the minister is willing to say that this needs to go to a specific vendor, they can actually do so without consulting anybody, basically putting their name on it. But of course, there's a limit to it. I think this only applies to things that are around or under 25,000 British pounds. And so for each and every prototype like this, they have this kind of discretionary budget use that we use often to explore the various different solutions. So that's one answer in that we can do an early design procurement in emergency. If it works, then of course, we do a larger one with the proper process of multiple bits. And the other thing is that the social sector actually contributes a lot of these things without any procurements needed. Basically, for example, when we did this mass map out of nowhere, the HTC, which is a company that used to make pretty good phones, still makes pretty good phones, but most of them gets bought by Google, HTC makes a chatbot, a line chatbot. And that actually solves a major problem with maps, which is quite bandwidth intensive. So with the chatbot formulation, people can just go to the chatbot, ask, I'm here, where are my nearby pharmacy, and it shows a few cards, and it doesn't need to zoom in or out or anything, it just takes you there. And they finish the development in like 24 hours without any procurement, because we publish the open data every 30 seconds. So for the chatbot developer, that's golden, because then you can go to the nearby pharmacy, swipe your NSI card, procure, sorry, purchase nine masks. And after a couple of minutes, the chatbot can actually tell you that this stock, if you're an adult, then this become like 49. So basically it's instant gratification, something line chatbots really like. And also people are basically participating in a ledger in this way. This is a distributed ledger. If you go to the pharmacy, purchase, and after a couple of minutes, this rather goes up into like 60, then you will call 192 and something bad happened because the ledger isn't working. So instead of the traditional freedom of information act, which usually publishes like every week, or at most every day, this is not open data anymore. This is open API, because it's almost always real time. And once it's real time, there's a lot in those service providers, the chatbot developer's interest to provide this as a value at the service at pretty much no development cost to their customers. And because of that, people just develop it voluntarily. They don't really need procurement money, because this also adds to their bottom line. Excellent. Thanks. We've got a question now from Katie Lachlan. She asks, isn't there a problem with the geo fence that you were talking about earlier that people could leave their homes without taking their phones with them? Yeah, which is why, of course, the chatbot checks on you a little bit, like in random intervals, and asks how are you feeling? What's your temperature? Would you like to take a picture with your thermometer and things like that? There's, of course, that part of that. And if your phone runs out of battery, then, of course, after a few minutes, a police will come and visit you. And so, but people are addicted to their phone anyway. So in practice, we don't, we don't see a lot of that problem. Yeah. I've got a role, a couple of questions which are related together. So Nathan Yang asks, what do you think is the best way to reduce the siloing of information within different parts of government? And that matches quite nicely with this question from Joe Mitchell, which is, have you had any trouble convincing your fellow ministers of the importance of transparency, trusting the citizens and so on? Yeah. In Taiwan, we reduce the siloing of information by making sure that when we procure, we can make sure that the vendor can deliver a machine-readable version, that's to say the open API specification or OS3. I think the GDS in UK eventually recommended OS3, but that's, I think, two years and a half after we did. We recommended that even before OS became the official standard. And I think that was in late 2016. And we basically said that during procurement, we treat open API as a sort of accessibility because there's a section in our procurement template that says if the vendor charges extra to make it so that people with blindness can see the information, then that vendor is not professional and could actually be disqualified for charging extra to make their websites accessible. Accessibility should be universal, it should be the default. And we kind of changed the template to say that machine-to-machine API is a kind of people with blindness. And if you discriminate against bots, well, we didn't quite say that. But if you say that you need to charge a lot more for providing open API for this human visible and human input places, then you could also get disqualified just by doing that. And so that kind of forced all our vendors to speak open API by default. And once you speak open API by default, of course, you would design your front-end in a decoupled way that basically works across all the different like maps, apps, chatbots, and voice assistant, because then they all will talk to the same API. And once the procurement system integrated, getting to the habit of API-first design, then the sideline of information is automatically soft, because your front-end, your back-end is then thoroughly decoupled. And everybody who wants to use the same back-end data for a different front-end application, just like the HTC DeepQ team, they do not need to ask anything about the other front-end like map developers, because they will just look it up on the open data portal and then see where the open API is. So our open data portal, if you refresh quickly enough, then you just take the API box and that ensure that your procurement goes successfully. Excellent. And how have you sort of approached bringing other ministers and other parts of government with you in this sort of journey towards more transparency? Sure. So my theory of change is very simple, right? It's three axes. It first is saves people time, reduce the chores. It reduces political risk, because when it's the people's idea and we just support people's idea, there's no political risk whatsoever. And finally, there's due credit for all the public servants involved, because back in the day before that we had this open collaboration meetings, for these people that you see here, the section chief's career public service, if things go wrong, it's their fault. But if things go right, well, it's the minister's credit, it would be my credit, right? But now, because people see them, I do I, and we keep a transparent record of each and every decision made, it's their credit when things go right. And it's my fault if things go wrong, because I'm the one who came up with this crazy idea of radical transparency. And so this flips the ideas around credit sharing. And so it's like three dimensions. So as I said, less work, less risk, more credit. And we only make Pareto improvement, meaning that we never trade one for the other two. We only make piecemeal improvement on one without sacrificing the other two. And so from the other minister's viewpoint, this is essentially a no brain deal, because who wouldn't want less work, less risk and more credit and more trust. Thank you. Got a question there from Julian McCrae, the director of Engage Britain. What's been most effective in getting civil servants to trust and actually seek out the input of citizens into developing policy? The most effective way as I have found out is that when we have no idea. If you have some idea, then people are either for that idea or against the idea. But if you get into the habit of saying, well, I have no idea, please come up with some idea. And then that always works. And so the citizens really like the agenda setting phase of policy development, because that's where their experience truly counts. When you move forward into in design thinking terms, into develop and delivery, that actually requires a lot of professional expertise. But when you're in the first diamond, that is to say, if you're just discovering and defining, then people are very happily contributing their ideas. And we basically have a code PO that P is the TW. PO is for participation officer that outlines what is participation officer, the directions, how do you choose collaboration topics, and the principle for how to interview people, the process, the toolkit and things like that. There's this whole regulations, if you want to copy it in your jurisdiction, please feel free to do so. And we make sure that people learn about how to choose the proper ideas, the proper processes to make that happen. And so, as I said, if you start with agenda setting, then people can very easily see that they are themselves setting the agenda. So most of our collaboration meetings, why it's so important to have open registration of presidential hackathon, like more than 200 teams this year, or sandbox applications, or the joint platform with e-petitions and so on. That's because when people mobilize to give us those ideas, there's already some stakeholders there. And so, for the civil service, this is guaranteed to be more signal than noise. And when you're in that stage, if you take away the reply button, as we did during Polis, and actually Slido too, there's no Republic button on Slido either, if you don't have the reply button, then people can only add to each other's ideas. They can never attack each other. They can never subtract from each other's idea. If you don't agree with me, you probably have to propose something that other people agree with. If you don't think a Slido question is worth asking, it doesn't pay to attack that person. You have to propose something more interesting for other people to outboat. So also design your interaction platform so that for civil servants, you always get more signal than noise. And because of that, then the civil servants will trust this more if it's early enough in the agenda setting stage. Just a reminder to everybody watching, if you'd like to put a question to the audience, you can use hashtag IFG Digital on Twitter. You can use the live stream chat on this broadcast, or you can use our Slido, which is bit.ly slash ifgtongue. I'm going to take another question from Slido now, which is actually about one of the processes that you mentioned earlier. Could you explain how quadratic voting works? Okay, certainly. So in quadratic voting, everybody, the 10 million visitors have joined the GOV.TW, looks at the 200 or so proposals on that platform. And then everybody gets 99 points. Now, every single project need to be already SDG indexed, meaning that you need to respond to one or more of the sustainable development goals. So they are all problems worth solving. The main thing to ask people is what are the best ideas that's worth coaching, right? So people, of course, there's no one who understand all the details of all the 169 sustainable development GOV targets. So people naturally will mobilize and vote for the one that they feel the most interested in. However, if you give people 99 points, chances are if you elect dot voting, people will just vote 99 points to the first thing that they think is good, or their friends and family call them to vote. Quadratic voting says that you can't do that. If you vote for one vote, that's going to cost you one point. But if you're going to vote two votes, that's going to cost you four points. And three votes is going to cost you nine. And so with 99 points, all you can do is vote nine votes. This particular one, which is SDG 6.5, which is about using a IoT device on the waterways to ultimately detect the pollutions to the aggregate lines by the industrial plants nearby. And for the low abiding industrial plants, also by those water boxes to prove that the pollution came from upstream. It's a pretty good idea. And so, and power by this religion. And so if you really like the idea, you can vote nine. But then you still have 18 points left out of your 99. So you don't want to squander those points. So we'll probably look into other ideas. For example, using computer vision to reduce marine pollution by stopping those marine debris before they hit the shores. And that's a pretty good idea. And you still have what 18. So you will vote four, which costs you 16. And then you have two more. So maybe you look into two other points. At some point, you will discover that some and something have synergy with something else. So maybe you take some of these back and do the seven and seven. And so in mechanism design terms, in mechanism design terms, the marginal cost and the marginal return is equivalent in the quadratic voting, which means that the most strategic voting method is to review your true social preference, which contributes the more to the complete picture of the system of goals. And people are motivated to learn about four or five of the sustainable goals. So unlike traditional one person, one vote or one person thought they think most people feel they have won when we announced the top 24 this year, because other than the people who have never voted, everybody who have voted, they average vote for maybe five or six teams. And one of them is bound to be part of the top 24. So everybody feel they have won, unlike the traditional kind of binary voting, where half of people will feel they have lost or everybody in some other cases. Next, we have a question from John M, which is how have you addressed access to hardware issues for people with lower incomes, where the cost could potentially bar their access to some of these systems? That's called public libraries and also digital opportunity centers. As I mentioned, broadband is a human rights, but also they can go to their local digital opportunity center or public library and just rent for a device that's guaranteed to be made in the last three years. Great. We next have a question from a UK open data expert, Peter Wells. Morning, Peter. Why do you think the UK and indeed other European Union countries have focused on app-based contact tracing rather than using existing data? And do you think anything might change that perhaps looking to the Taiwanese example? Well, the thing is that the digital fence worked the way it is using very coarse grained like 50 meter precisely because we do border control. So we do very strict enforcement. And during those voting days of quarantine, a lot of human rights, not just privacy, but right of movement is encroached. But then it applies fairly to everybody. So it also doesn't have a labeling effect. However, if you don't do defense at your ports, then you have to do defense in the community. And once it's in the community, 50 meters is two coarse grained. And that's why people start getting the idea of we should use Bluetooth or even we should use GPS data if it's stored in centralized databases. So I think it's a conscious choice by Taiwan to do most of our protection, the strict measures at the borders so that we can live in a much more relaxed fashion in communities so that we ensure that hand sanitation, mask use, physical distancing, these three measures together is enough to put the R value of being under one. So that's a conscious choice. But you probably cannot do that unless you have near universal mask wearing, which of course is a culture thing. But in Taiwan, as I mentioned in the Doge CEO picture, we put these pictures together because we say that masks are something that reminds you not to touch your face and wash your hands properly. So mask is a social signal. It's most of the psychological tool. And for this, everybody agrees with that, right? The scientific evidence and so on about masks like blocking filtration or whatever capability that's up to the debate and up to the fabric of the mask. But wearing even a mask made out of a t-shirt reminds you not to touch your face and reminds you wash your hands properly. Probably everybody can accept that. And by a few people, even just a few people in a large crowd sending the social signal, it enabled people to take care of each other saying, hey, why are you not protecting yourself from your own hands? You should probably also wear a mask. And that enabled this idea of wearing a mask to have a high R value in the idea MIM space. And because of that, everybody started wearing a mask. Excellent. I think you've answered a question that Judith Richard said on the slide actually, which was, do you think masks are key, especially in places where we cannot social distance? It sounds like your answer to that would very much be yes as it sort of helps take the rest of the strategy. Just to remind everybody, we've got about 10 minutes left. If you've got any final questions for Audrey, please do put them on Slido. Now that's bit.ly slash ifgtong. So we have a question from Steve Lloyd. There's been a lot of discussion in the UK recently about local lockdown measures. So perhaps not the entire country is around the same rules and it's able to adjust according to the level of outbreak. So he's asking, do you have any advice regarding the moving out of lockdown measures on a local level? Yeah, we have simulated, like if there is a limited local community transmission, there may be places where we will just have a certain area that is self-sustaining, meaning that it has most of the kind of groceries and things like that, so that we can do a very contained local lockdown while the rest of Taiwan is still free of lockdowns. So that we have simulated. But what Steve Lloyd is asking is essentially the other way around. That is to say the rest of the country in lockdown, but only one part out of lockdown. If it's an island, then of course go for it. But otherwise, I think you need to carefully plan, as we did on the border control, how to screen the people and how to make sure that people who transfer between those different municipalities carry the same quarantine obligation, essentially, as we did for our returning citizens. If you can do the quarantine and contact tracing using regular interviews and so on, well, then there's no reason why you could not do that, because that's essentially what Taiwan did, because we're a set of islands. But if you cannot do that with efficiency, if your contact tracers' capacity is not there, then of course they inherit risk in doing that. And that actually is one of the points that we made in Taiwan Can Help That Us, which is this, as I mentioned, crowdsourced website of the Taiwan model, is that we make sure that people who eventually see those measures, such as the lockdowns and now their partial relaxation and so on, understand the scientific reason why. So this person is Chen Jianren, Vice President of Taiwan at the time, trained and John Hopkins, our top epidemiologist, actually wrote a textbook on epidemiology. And so our top scientist doesn't have to convince our top official because he was the top official. In this case, he recorded this crash course on the popular MOOC database, the HAHAU, where a lot of people enroll in online learning. So I think more than 20,000 people enrolled in the first few days, me included, so that he explained all the different models and the different simulations. And if you're interested, you can also, there's a interactive version of that at, I think, Nikki case, ncase.me slash COVID-19. That says what happens next, which you can take Dr. Chen Jianren's courses, but then verify his numbers by simulating all the various lockdown scenarios and things like that. So my main point is that if everybody become kind of an amateur epidemiologist, then this kind of localized measures will work because everybody who travel there and travel out of it will understand why are those measures in place and what effect is it having on the population. But if people don't understand the underlying science, then of course there will be people who do not innovate in advance for those goals, but rather innovate in the opposition of those girls. I think that's the most mild way that they can put this. Excellent. I have a feeling we'll have quite a few people enrolling on that course after this event has finished today. We've got a question there from Terence Eden, who works on Open Standards at NHSX and did a fantastic talk on why making things open makes things better as part of our data bite series and events last month. He asks, we see lots of open source code from Taiwan and indeed from other countries, but most countries are still trying to roll out their own coronavirus-related services. What's stopping the reuse of your code? Well, nothing is stopping reuse of our code. I think this is mostly culture. If we put our code out there as we did in the coronavirus hackathon co-hack.tw, as you can see, everybody who participated see very clearly that from various different countries, including the UK, people design those privacy enhancing technologies that enable, for example, working with contact tracers, but all the data is kept in your own phone and only sends a one-time link to the contact tracer for the information they need without divulging any privacy of anybody else you have encountered and autonomy does that on a community level. Gemini does the storytelling, part visualization of that and so on. And each and every one of them agree to abide by the spirit of the open source and agree to use the MIT license, which is one of the most permissive license for their work. And so as more countries participate in this sort of open innovation, people will get more into the culture of just learning from one another. But if you start with a procurement strategy that doesn't include open source and open API in even template language as Taiwan did, then of course, it was very counterintuitively because after all, you did not pay Taiwan. And why would you then use Taiwan's kit? So there's a lot to do to change the culture around procurement to basically avoid the not invented here culture. I think this is a culture thing and not at all a licensing. The licensing is just the underpinning. The culture is what needs to change around procurement. And how do you change that? Do you think? Well, two things. First, write it into the template language so that open is by default. And if you don't do open, you have to write a reason why. This is like the regulation that needs to be open for open consultation for 60 days. There are still ways to kind of work around that, but you have to say exactly why. And that why is also kind of accountable to the entire population. So when there was a draft bill where people said, where the ministry said, we need to fast track this to seven days because otherwise we will not meet the parliamentary schedule. They drew a lot of flak from the journalist who say you could have published a draft like a couple of weeks earlier, then you will meet a parliamentary schedule and then you will still have at least 30 days of public deliberation. So making the defaults much, that is my first suggestion. The second thing is that the civil service are also citizens. So many of them are also very much willing to use open data from other nearby siloed departments. And you need to encourage that internal innovation or so-called intrapreneurship. So when the HTC, which is a contractor of the Center for Disease Control, uses the national health insurance agencies mask data, I made sure to publicly just give them a lot of credit to make sure that I share a lot of their stories in my talks to their ministers and so on. Because when the minister level encouraged this kind of internal intrapreneurship and even participate like I did as a programmer, I just held fix in their code, then people would get very encouraged to use data across silos, not personal data, open data across silos. But if the minister level doesn't give it's a thumbs up, then they would end up absorbing the risk while getting no credit, and in which case there is no culture possible for the open within the public service. So for public citizens to make sure that their social innovation can amplify or the civil servants make sure that their ministers show an active interest in building an open culture, and then you will have the culture of avoiding the need syndrome, they're not invented here syndrome. So I'm going to squeeze in one final question and it's actually a perfect final question from Joachim von Halles on Slido. What are your ideas and developments for the future? What are you working on next? Okay, so as with every other questions like this, this is time for me to read my job description. So I mentioned about the sustainable goals and how my work is in the 17s, which is building effective partnership. Of course, that is my main goal. And our president promises in her second term, which begins just a month ago, not even a month ago, that we will now have a dedicated cabinet level, digital council or ministry that takes care of this cross silo policy making that enhances even more of reliable data, and also do open innovation, making sure that open innovation is a dedicated council or ministry in Taiwan. Now people ask a lot, why call it digital? Why not call it information and communication technology or ICT? Because that's the more usual term as you use in Taiwan, because Taiwan is you know, Taiwan semiconductor and so on, very good on ICT. But we insist on calling it digital. So I wrote a poem, a prayer that explains the difference between ICT and the digital, and that's my job description, which I'll read to you now. When we see the internet of things, let's make it an internet of beings. When we see virtual reality, let's make it a shared reality. When we see machine learning, let's make it collaborative learning. When we see user experience, let's make it about human experience. And whenever we hear the singularities near, let us always remember the plurality is here. And that's digital. Thank you so much. What a perfect note to end on. And what a great way to start the week. Thank you everyone who's tuned in. Thank you for some brilliant questions that we received from everybody. And please, everyone, thank me. Join me in a virtual round of applause for the fantastic Audrey Tull. Thank you very much indeed for joining us, Audrey. Thank you. Thank you.