• Let’s get started.

  • Thank you very much for seeing us. This is my first visit here to Taiwan zone. I’m a generalist on my island, as you’d imagine. This is Ed McBride, who’s our Asia editor. He runs all our Asian coverage. And David Rennie, he’s our Beijing bureau chief, and writes our China column. And Jane, who’s been with us, who’s here as our correspondent.

  • I wondered if I could start by asking a question about digital Taiwan and digital China, and how much China infiltrates digital Taiwan. In other words, how active in the cyber area China is, and whether that’s a threat to Taiwan in digital at all?

  • Would you like to be more clear about the so-called infiltration? Do you mean on the cybersecurity layer or on disinformation layer?

  • There are various layers. There’s the cybersecurity layer. It would be good to cover all of them, but I was thinking initially the misinformation, fake news, and stuff like that.

  • I don’t use the F-word, but you’re free to.

  • (laughter)

  • Both of my parents are journalists, so I never use the F-word to describe their work.

  • (laughter)

  • As journalists ourselves, we’re often accused of it. [laughs]

  • I’m happy to share my views on misinformation. We do distinguish between misinformation and disinformation. In Taiwan, all our government communication has changed to use disinformation to mean something that is both intentional, objectively untrue, and harmful to the public, not just to one specific individual.

  • It has to be intentional, it has to be objectively untrue, and it has to be harmful to the public. If it only satisfies one or two of the three criteria, it may be just misinformation. People are misinformed, or maybe it’s just political satire, which is protected by freedom of speech, and so on.

  • Disinformation is the ones with all the three pillars. [laughs] Misinformation is everything else. Misinformation, to me, is something that affects all the different modes of cyberspace communication.

  • Back in the year 2000 or so, I was involved in the concerted effort to tackle what we called the Spam Wars back then. At that time, people’s email inboxes were flooded with Nigerian princesses offering you lots of deposit or something like that.

  • For a while, people panicked thinking maybe email has stopped being useful. There’s various crazy proposals like we should charge postage stamps for every email. I don’t know...

  • That was a Bill Gates proposal, wasn’t it?

  • You still remember those bad old days? [laughs] It turns out it took us maybe three or four years to solve junk mail. Nowadays, we don’t think about junk mail much anymore.

  • That is only because we in this Internet multistakeholder community made a concerted effort to essentially deploy what we call IA nowadays -- back then, it’s just Bayesian learning -- into people’s flagged suspected junk mail.

  • We built a pattern database called the Spamhaus. In the Spamhaus Project, people lowered the virality of any sender that matches a certain message so that they won’t reach as much people. Then all the email providers agreed to use a secure protocol to identify the domain of the sender and so on.

  • Along each step, it lowered the expected reward by a little bit and increased the costs by a little bit until the economic incentive is no longer there to participate in spamming. It took us maybe four or five years, but we eventually got it resolved without hampering the fundamental freedom of people on the Internet to send each other email without prior solicitation.

  • I use this analogy just to say that we’ve seen this before on a different media, of course, on a different configuration of cyberspace, but it is just like that. It is not something new.

  • It is just a reincarnation of the same configuration of expected reward, but now on what we call social media, which makes it easier to share a piece of news on a smaller screen before actually even finishing reading that something on the screen.

  • This configuration is new because around the turn of century, the screen are large screens and it’s easier to read stuff than sharing it. Nowadays it’s the reverse, and so that’s the background.

  • Taiwan, of course, is affected. I wouldn’t say it’s just by the PRC, but also by all the different neighborly authoritarian or going to be more authoritarian jurisdictions.

  • When faced with the same social media disinformation issues, there is a strong incentive, if you start with the authoritarian governance system, to use this as a easy, I’m trying to think of a neutral word, as a easy excuse to shrink the civil society’s space, the basic freedoms of speech, assembly and so on, by essentially passing more draconian cyber control rules around.

  • It is, of course, a very attractive proposal to any authoritarian regime. The PRC for example spends a lot of energy on what they call [Mandarin] which I don’t know how to translate that -- to stabilization? I don’t really know. To keep the stabilization of the social media and the Internet there.

  • I’m sure there’s some figure that says they spend more on this kind of internal policing than even some outward endeavors, but even as such, disinformation is not gone from Weibo, or from the social media of the PRC.

  • They do pass some, one would say, regulations and endeavors that shrinks the civil society space online, but still even after that, I wouldn’t say they completely solved the disinformation problem on their cyber space communities, and so Taiwan is perhaps alone.

  • If you look at the CIVICUS Monitor, which tracks the ongoing condition of citizen action, then you see very quickly, at a glance, if you choose...It’s a rank from closed, repressed, obstructed, narrowed and open.

  • It decreases a few points of score whenever something new development happens. It’s a constantly updated monitor that tracks around the world all the newest developments. If you go to Asia and select "completely open," you only see Taiwan.

  • That is to say, in the recent years in Taiwan, the civil space is actually expanding. You don’t see any similar news as compared to other jurisdictions that makes the freedom smaller or reduce the freedoms.

  • Britain and France are narrowed.

  • [laughs] The UK is just not scoring top marks.

  • No, it’s not, but Ireland is.

  • No, it’s not that law. It’s the...

  • ...RIPA, isn’t it? You know the RIPA Act in the UK?

  • It’s considered by some as draconian. I think we’re a favorite. [laughs]

  • They say France is narrowed, but Sweden is still open, and of course I don’t maintain or have any relationship with this website. [laughs] I just want to say there is a strong polar in our region, in a world actually that extends all the way to Africa to use disinformation as a excuse to shrink the civil society space.

  • In that sense, yes, it is a threat, but it is a threat not because of disinformation itself, but rather because...

  • Because of the response to it.

  • ...of the response to disinformation threatens to shrink the civil society space.

  • Would you say that because you think that there is an evidence that misinformation can change public opinion, mislead and distort public opinion?

  • Of course we have pretty good numbers that shows exactly how...For example, here is the Cofacts project. Do you read Traditional Chinese?

  • There’s an English version of it, but anyway, this is a simple bot on the end-to-end encrypted instant message channel, called LINE. LINE is pretty popular in Taiwan, similar to WhatsApp and other places, and it’s end-to-end encrypted, meaning that even the LINE company doesn’t know what’s being transmitted, aside from the stickers, which they do know.

  • In any case, this is a bot that anyone, when receive a piece of misinformation or disinformation, the bot is literally called Is It True or Not? Its English name is Cofacts, as in Collaborative Facts, and this is an entirely social sector, civil society project, not government funded.

  • Just by looking at the recent trending rumors, we can very quickly see which rumors are being spread, how quickly in end-to-end channels, and whether they are actually true or not. At the moment, it’s mostly about referendums. It’s pretty clear. Even though, as always, there’s always what food plus what food, and seeing what cancer, or whatever. That never goes away.

  • In any case, yes, a lot of it is election-related, maybe because we’re in an election week. The good thing about this is there’s a volunteer group just surfacing, these to the public searchable web and also just in junk mail, providing white or blacklist, that clarifies whether this is a rumor that is true or whether it is a rumor that it actually has some evidence in truth in it.

  • Here are the evidences and the bot, once it accumulates more than one report, starts on the third attempt, where you share this rumor to them. The bots start giving you back these clarified responses. Yes, first we do know which ones are trending, even in end-to-end encrypted channels.

  • Second, we’re taking actions by essentially surfacing them to the public web, and making each rumor a URL, around which the people can have a real discussion on. There’s the Taiwan Fact-Checking Center, for example, that can then look at which one are trending the most and spend their journalistic energy to provide a full account.

  • Cofact is government?

  • Cofacts, it may look like government, but it’s not. The g0v.tw, the g0v, pronounced gov-zero, is a movement starting six years ago, that consistently look at all government services and find the ones that are either broken, unusable or missing, but systematically build civil society alternatives to it.

  • Just by changing an O to a zero, you get into the shadow government, so to speak, where the civil society people, the "civic hackers" contribute things that they think the government is not doing a perfect job on, and just fill in.

  • Are you confident that by surfacing rumors, allowing people to have debates about it, and making it clear that things are contested, that that blunts their ability to mislead and distort people? There is a sense, as expression that the lies have got seven leagues before the truth has put his boots on, or whatever.

  • I do agree with the sentiment, although I would want to make a distinction. Sometimes, misinformation, or just rumors, they get spread, and they were not that harmful, in either intention or the reach. After a while, think AB testing, right?

  • Things will naturally get more escalated, just because in the Darwinian methodology, things that are more sensational or viral automatically gets more people spreading it. It mutates, so to speak. If, before it mutates to a more harmful version, people already surface the less harmful version, then that is actually how vaccines are made.

  • Vaccines are essentially virus that are not that potent, and that gets into our bodies, so we become immunized over it.

  • I think whenever misinformation that has not yet evolved or manufactured into disinformation gets discussed in this public way, then people builds in their mind a vaccination against that particular topic, so that they would not be affected by polarized information, once it evolves into the disinformation stream, so to speak.

  • I’m not saying this solves 100 percent of the cases, I’m saying this is part of media literacy.

  • Going back to the analogy of spam, it’s one of the things that reduces the virality.

  • It’s one of the things that reduce the virality, increase the cost. Of course, the Taiwan Fact-Checking Center does something that’s even more. They actually look to all the trending ones that are on the public web. For example, this is a classic one.

  • The Taichung Social Bureau reportedly said that in our ID card, our gender field, which used to be female and male, will be canceled and everybody has to pick between LGB and T. OK, you smile, but it’s something that’s trending.

  • The Taiwan Fact-Checking Center did a full journalistic work of checking the source and demonstrating how exactly they did a fact check, and of course, proclaim it as false. That’s their 21st report. The TWFCC is part of the International Fact Checking Network, the IFCN, at the Poynter Institute.

  • That gives them a kind of special status, because when clarified this way, the signal is taken into account by large multinational platforms, as one of the input sources to their algorithm, to reduce the virality of, not criminal, but misinformed content. Just by producing this report, it has the effect of making this particular item spread to less people’s news feed.

  • Just on the Cofacts, because fact-checking, that’s a very familiar concept, I get how that works. With the stage before with Cofacts, when they’re identifying something as a rumor, what do you do about something...Obviously, if it says Tsai Ing-wen is actually a communist agent, that’s a rumor.

  • What about if it says, "Such and such a politician gave a racist speech"? How do you know if that’s a rumor, or whether that’s just your point of view? It’s a political judgment.

  • If it’s your point of view, then here, it says it’s a point of view. That’s actually one of the four categories.

  • Which one is point of view then?

  • There’s many. For example, the point of view...There’s a lot of it.

  • The dot, dot, dot is point of view.

  • The dot, dot, dots are the point of view.

  • I don’t know which example, like the fourth nuclear plant is unsafe and is a bad deal, after all. That seems like a point of view to me. It’s a judgment call, essentially, but if it says the fourth nuclear plant requires such and such years to complete, that’s an objective fact that could be fact-checked.

  • The Cofacts do make a distinction between things that are personal opinions, which there’s nothing to fact check.

  • What’s the yellow triangle?

  • Outside of scope. For example, this one is just a regular chat. They think the chat bot as their friend, maybe. This is a future prediction. This says what will happen the day after the election. It’s a novel, essentially. You can’t fact check a sci-fi novel. It’s just not possible.

  • (laughter)

  • That’s the four categories they work on.

  • How do you detect rumors in LINE messaging system if everything is encrypted? Is that a violation of privacy? How do you detect some of the rumors in transit?

  • People voluntarily share tat to Cofacts. One of the way Cofacts works is everything that’s shared through it gets posted to the public Web.

  • The analogy you made with spam and email, are you saying that in some similar timeframe, the world, or at least Taiwan, will get a handle on disinformation, and it won’t seem like the threat that it currently does to all the people who are wound up about it?

  • What I’m saying is that email is also a point-to-point, and if you use Pretty Good Privacy, that’s even end-to-end encrypted. I’m saying, from a technical standpoint, they’re extremely similar in configuration. That’s all I’m saying.

  • Of course, nowadays, there’s more people using social media than people using email back around the turn of the century. There is a larger fan base, so to speak, of disinformation, and so the reward is higher, so to speak. The cost, because of machine learning, is also lower.

  • I would say it’s harder in absolute numbers to tackle, but actually, people who are participatory, who are civic-minded, who build systems like these are also machine learning experts. The people fighting it also have better tools compared to the simple bias and filtering.

  • I wouldn’t say in relative numbers whether this is easier or harder, whether it will take longer or shorter. In absolute numbers, this is both a more rewarding thing to do compared to sending spam 18 years ago, but people working on it also have much better tools to work with.

  • When we’ve first heard about this, we’ve also understood it in terms of attention hacking. In other words, systems that are optimized, social media companies optimizing their systems in the way they distribute information in order to hack your attention, which is something that email doesn’t do.

  • It kind of does. There were chainmails and things like that, but not at this scale, no, which is why I always recommend this wonderful plugin to people. There’s a mobile version as well. If you install it for Chrome, Firefox, or Safari, it replaces your newsfeed with a inspiring quote, in this case by Atler.

  • (laughter)

  • It just makes the attention hacking go away, because then, everything on the Facebook after installing this plugin -- I use it for years now -- is intentional. I type in a keyword, I look at the result, not unlike Google.

  • I click on the page or a live stream, I watch it, not unlike any blog. I post blogs and interact with people, not unlike Medium or any other system. It takes all the surprising part of Facebook away, but keep all the intentional parts.

  • In other words, if you want to use them, there are solutions out there.

  • It’s just like in the bad, old days of email before Spamhaus. People still installed a local filter called SpamAssassin. I worked personally on that.

  • Also, before this whole movement of conscientious advertisement, people already developed ad blockers -- advertisement blockers -- and some people installed it on their browsers. Before we have a systemic solution, there’s always edge-based solutions.

  • We’re very familiar with the idea with the Russians, the Mueller investigation told us how the Russians...With the indictments, we learned how they hacked Facebook and planted these ads.

  • The sense is that the Chinese don’t do exactly the same thing when they’re trying to interfere in other people’s politics. What is your sense here? Clearly, Taiwan is often caught on the frontline of Chinese influence operations. Do you see them doing similar things with targeted ads on Facebook and trying to sow dissention in the way that the Russians do?

  • A lot of the Russian stuff, the indictments describe as race-based news trying to get the two different races across from each other in relative action. I haven’t seen detailed reporting suggesting the Chinese do that thing. What are they up to, if they are up to Internet-based disinformation here in Taiwan?

  • You just described it as attention hacking, to saturate the topics of discourse, agenda-setting power. If everybody in Taiwan talks about things that are not actually related to the democratic institution and processes, but rather around ideologies, then it sows discord by default.

  • Can you give us some concrete examples of the kind of things that literally sticking up in Taiwan? Who do you think is doing it?

  • In the Russian case, what was so fascinating was that a group of people working out of the GIU were buying ads. This is an ad. This is the kind of ad they were buying. I’d never seen anything similar in the Chinese context.

  • I haven’t seen anything similar either, which is why I can’t give any example. I’m just saying people own their cyber space, based on what we’ve seen from the Cofact rumors. There are people doing a lot of concerted work to spread disinformation. Whether they are PRC-based, PRC-funded, I don’t know. I seriously don’t know.

  • It’s pretty clear that the ones that are the most viral are somewhat related to the idea that all these referendum topics and things like this, they all attract a huge of number of disinformation that diverts attention from the substance of the referendum topics, but rather along particular ideological lines that tries to get people into a knee-jerk mood around such referendum topics.

  • That’s the kind of attention pollution, or attention hijacking, or whatever I’m trying to get across. First, I don’t know whether they are PRC-generated.

  • Second, it’s very clear from the format and framing of the message, what they really want to do is to engage everybody in a way that gets rid of people’s space and more into a half-rational discussion, by getting people to see part of the society as the other and somehow reduce the space for substantial collaboration or, at least, a useful, rational debate.

  • It’s clear that there are people manufacturing messages to this end, but we don’t know -- at least, I don’t know -- whether they are PRC-based or PRC-funded.

  • I’m going to first make a contention, which I’d be very interested to hear your views on, but then ask how it plays into all of this stuff.

  • It seems to me that Taiwan has a uniquely vigorous civil society -- the Sunflower Movement, all these campaigns on the environment, on gay marriage, all of this stuff -- and degree of public engagement that many longer-standing democracies would love to have, and don’t.

  • We’re new to this. [laughs]

  • Maybe that’s right. Does that make Taiwan more or less vulnerable to the same degree as those other democracies to this kind of stuff?

  • I would say in the government side -- because I work open government -- I think it does make it easier for us in the government to say, if we want all the ministries, after detecting there’s a popular disinformation campaign, now every ministry is committed within five hours -- I think it’s going to be four hours real soon now -- to give a clarification.

  • It turns a Web-kind, like a real-time shooter game into a time-based game, more like chess or bridge. The analogy is that if you see something that is partial information on morning, people learn to wait until noon, and there will always be a piece of clarification.

  • I’m not saying that the ministries always have all the information, but at least, they can complete the part of the gestalt so that people don’t project literally psychologically into the partial message. Same, when it goes around in noon, then by dinnertime, there will be a piece of clarification from the respective authorities and ministries.

  • The very fact that I’m personally easily accessible -- anyone can just come to my office hour every Wednesday from 10:00 AM to 10:00 PM and have a face-to-face chat -- and the fact that I publish everything that I’m a chair of, including internal meetings and interviews such as this one 10 working days after each conversation.

  • People know after I become digital minister, I talk with 3,000 people, 159,000 speeches. It all gives a radically transparent way of how exactly does digital minister work, and what kind of policy we’re working on, and why we’re working on it. It gives the context, not just the what, of the protection.

  • None of this would be possible without a thriving civil society that almost always have higher legitimacy than the administration itself. When we talk to the ministries, saying we really have to respond in time, the ministries don’t say the civil society would misinterpret our results, or that it’s better if we just keep people in the dark, or things like that.

  • Each ministry understands that compared to the most well-known civil society counterpart, that counterpart often has higher legitimacy than the ministry. It’s in our best interest to be radically open, because then we can collaborate with people in the civil society working on the same thing.

  • If I can ask this slightly awkward political question, do you think this sort of thing is only possible in a DPP government in the sense that the DPP has closer ties to those protest movements and that organizing? Could you imagine this in the previous government, for example?

  • This was many of these, including building the Join platform, join.gov.tw, which is our national e-participation platform, 5 million users out of 23 million people, is actually founded in 2015 in the previous cabinet.

  • It evidently has happened under the previous cabinet, but with the caveat, of course, mostly it’s led by Simon Chang and Jaclyn Tsai, previous cabinet members who are non-partisan. They’re not DPP, but they’re not, strictly speaking, KMT either.

  • Simon Chang was Director of Engineering at Google Asia. Jaclyn Tsai was Director of Legal Affairs, IBM Asia.

  • Like me — who worked with Apple — we are multicolored.

  • (laughter)

  • I would say that, while this work did not start within the DPP, having Dr. Tsai as the President does make it easier for the civil society to engage more with the government apparatus.

  • The current apparatus was first put together in 2015, in response to the Occupy the year before.

  • The other aspect of this, apart from misinformation and disinformation is cyber attacks, DDoS and all sorts of things. How much of a threat is that and how do you deal with that threat?

  • Unlike disinformation, which we really have pretty good idea of the stem-like status of it, on server security, it’s many, many, many layers. Disinformation is predicated on the current generation of social media and its interface. Structurally, it’s, I wouldn’t say, entirely isomorphic, but it’s very similar to what we already know.

  • Server security on the other hand, the attacks are inventive. Like the BGP hijack thing, which is forcing through misconfiguration a certain number of traffic to route through certain points of presence that used to be a one-shot trick.

  • The people using that technique have escalated and perfected that particular technique so that it could be used both as a surveillance route, but it can also be configured, as in a few years ago, as a great cannon kind of way where people configured popular websites to DDoS the website GitHub. That is very well-documented.

  • What I’m saying is that the configuration changes every day, literally. It is not as determined as social media and its rules of information dissipation and dissemination. Rather, it is acting on the physical law itself of cyber space. The leading physicists, to use that analogy, gets very inventive.

  • My answer is two-fold. The first is that Cyber Security Act is the priority act. There’s no other budget priorities that I’m aware of that has a higher priority in our new budget cycle. With all the government, large projects have to dedicate five percent to cyber security, and medium projects, six percent, small projects, seven percent.

  • Penetration testing to all new systems is now a norm, like the system that I personally set up when I joined the cabinet was subject to six months of penetration testing by top-notch white hat hackers before we even rolled it out to other departments and so on.

  • This is, of course, also to build relationships with the white hat community so that they meet the digital minister and the president every once in a while. They get paid really well. There’s a thriving cyber security service industry, so they don’t go to the dark side, which has cookies.

  • Building the relationship with the white hats, as well as through laws and regulations, making sure there’s plenty of budget and personnel in critical infrastructures and so on, that is evidently our highest priority, and it’s already expressed in laws, budgets, and regulations.

  • We are literally on the frontline, and we behave as such. People don’t have to take training courses. They just go and face the real situation.

  • Can you give us some scary stats or anecdotes about how frequent, or how severe, or how accomplished the kinds of cyber attacks you’ve experienced are in government?

  • I can’t give you any specific cases, because as part of my compact, I have three conditions joining the cabinet: Radical transparency, voluntary association, location independence.

  • Part of the deal of radical transparency is that in a meeting I’m a chair of, I publish everything. In the Freedom of Information Act, that means the drafting stage material, which is supposed to be confidential, there’s an escape clause, but if it’s for the public good, then it may be disclosed. That’s my working condition. I disclose everything.

  • As a counterpart to that, I cannot access state confidential information, because according to our State Secrets Law, if any system’s input contains the secret, the entire system becomes a state secret. I make a physical isolation.

  • If there’s any particular cases that are state confidential, that is handled by the Cyber Security Department, the NCCST, and the National Security Council, the NSC. I don’t know anything about that.

  • I only work on the general mechanisms, education, and things like that, and also, of course, part of the regulation planning, and things like that, but I cannot access specific incidents.

  • When they, for example, run military drills, I take a day off. I still don’t know where the bunkers are, so there’s no risk of me disclosing that. I really don’t know what 衡山指揮所 looks like.

  • That’s a good segue to another broad topic that I’d like to ask about, but I don’t want to stop if you guys have more to ask about.

  • You must get this all the time. It seems your appointment, the conditions you just described, it’s unusual. I assume it’s a first in Taiwan. It would be a first almost anywhere.

  • To have a politician...There’s some people in Iceland that will say that they invented this first, but yes.

  • There you go. Iceland and then Taiwan. It seems to me related to what I was saying about the incredibly vigorous civil society. Can you explain how you think that came about here exactly? Why is it that Taiwan is so blessed with that when so few countries are? What’s your explanation?

  • I would cite two factors. The first is that the lifting of the martial law, which is ’87, and the first presidential election, which is ’96. There’s a decade in between. There’s a decade where I grew up, where I was 6 to 15 around this stage. It’s my formative years.

  • I remember those years as the CSOs, the civil society organizations, the social sector, the social enterprises, the co-ops gaining legitimacy really, really quickly. After lifting the martial law, they started working on what we will now call sustainable development ideas. They really gather around a large amount of people interested in betterment of the society.

  • At the same time, there is still no presidential election. The legitimacy of the administration will not be established until ’96. There’s a decade of head start for the social sector.

  • The legitimacy, especially for people my age or older, when the disaster happens somewhere in Taiwan, and the local office publish a number, and Tzu-Chi publish a number, people tend to believe Tzu-Chi’s number. It is a fact.

  • That’s the first reason why that we have such a thriving social sector. It’s because they were given a decade of head start, compared to the administration, to gain legitimacy.

  • You mean that because people weren’t fully participating in electoral democracy, but they still had that urge...

  • There was no electoral democracy to speak of.

  • They channeled that energy into...

  • That energy into co-ops, into social enterprises, and so on. That’s the first reason.

  • The second one is that the year ’96 is also the beginning of the dot com boom. People associated instinctively the democracy for real with cyber space, with free access to information and viewpoints. I remember personally working on the first presidential election campaign by building websites and things like that.

  • Representative democracy began at the same time as what we will now call collaborative governance, which is what World Wide Web brings to everybody. Instead of 200 years of proud tradition of representative democracy, they had the same beginning. From the very beginning it’s mingled together.

  • Internet and democracy, they’re not two things in Taiwan. The generation that starts to do presidential election campaigning is also the same generation that has access to the World Web.

  • Instead of two kinds of people, people working on public administration and people working on digital, and design, and cyber, we have the same generation of people pretty much balanced in the pursuits.

  • How does that lead to more civic participation necessarily? I understand what you’re saying, but I don’t see the connection.

  • Think Estonia. There’s less legacy. There’s less established ways to channel your political visibility through existing associations or existing representative council members. All these were fairly new.

  • While these are building, at the same time, the PTT or other visibility organizing methodologies are also building on the World Web and on the Internet in general. These two are keeping reinforcing each other. Unlike in other older democracies, it’s evident that if you have access here, it’s worth more than having access here, until very recently.

  • In Taiwan, these two are always mingled, so that people in the representative democracy think they have also to learn some digital communication, otherwise they don’t survive. On the other hand, the people who are digital activists, they also have to learn something about the newly budding representative democracy, otherwise they don’t get political way.

  • You mean the fact that the digital and the political were mixed...

  • It’s the same generation.

  • It in effect provided better organizing methods that were instantly plugged into politics.

  • Of course. It’s woven into the fabric of politics, just like Estonia, after they have their post-Internet constitution, the paperless way of working digital-first is part of their identity now. That really increases the participation.

  • When Dr. Tsai Ing-wen said, "Broadband is a human right. Everybody everywhere in Taiwan need to have 10 megabits per second, otherwise it’s our fault," it’s not seen as outlandish. It is seen as natural. You can’t say that in many other jurisdictions.

  • I wonder if I could ask you about digital skills and brain drain.

  • How worried are you about the brain drain, and what can you do to reverse it?

  • You mean brain drain in Taiwan? We are seeing a reverse brain drain. It was the opposite before it. I’m not worried at all.

  • You can see, around the time that we rolled out AI Taiwan plan -- which is just a year before, I think -- we can see pretty clearly...Let me see if we can increase the WiFi situation a little bit by switching to a better hotspot, or not.

  • We have such kind of one-page describe portals to our national endeavors, such as when you Google AI Taiwan, the first hit is going to be ai.taiwan.gov.tw. I think it’s getting out there now. You see the AI for Industrial Innovation, Innovation Hub, Pilot Project.

  • The very fact that if you think you can break the law for a year and show everybody your AI idea is better, you get to actually break the law for a year, and things like that. We do have a lot of tailor-made-for-AI regulatory programs, where literally, you can break the law for a year to experiment with AI technologies.

  • Like Ethan Tu, previously Director, Cortana Technology, Microsoft, when he went back to Taiwan, he is not reverse brain draining himself only. He recruited his teammates in the Cortana team and brought a lot of people back to Taiwan to work on AI embodiment, which is a very thriving field anywhere in the world.

  • It’s also around the past year and a half where Microsoft said we’re going to set up a 100 to 200-person AI research center in Taiwan. Then Google followed, and IBM, Nvidia, and Amazon, Uber, you name it, all started rolling out their programs.

  • I think the reason why is AI embodiment. When AI was still purely cognitive, like playing chess or Go, it doesn’t need the various verticals that Taiwan provides. It can be purely algorithmic. Everybody now sees AI as actually assistive intelligence, not just artificial intelligence, which is better for the AlphaGos of the world.

  • For it to be assistive intelligence, it need to interact with the society somehow. It need to embed itself into a workplace somehow. For that, you need experienced designers.

  • You need people who design the hearing, the seeing, the body placement, the social norm around the AI interaction with people, both in the individual scale, the edge CI, and the city scale. Smart city as well as how it handles personal agency and data agency. All these are cutting-edge issues that requires a very fast turnaround to all related verticals to work with.

  • For those multinational teams, their choice is either go to Taiwan, where you get access to all this expertise that I just mentioned there, within one hour and a half of high-speed rail of each other, or you can separate those five expertise into five time zones. That limits how fast you can iterate. We are seeing a reverse brain drain. AI is a large part of it.

  • The other thing, when we’ve written about AI, that we’ve focused on a lot is the utility or essential to have access to very large datasets to be able to use those.

  • If you compare the size of Taiwan’s datasets to the size in China’s, and the willingness. I don’t what privacy laws exist here. We assume that in China, not only is there a large amount of data, but the freedom to exploit that data for generating AI...

  • They’re happy to share it with Google how? They’re maybe happy to share it with, I don’t know, SOEs, but I’m not sure at all that they’re willing to share it with oversea teams.

  • No, I agree. I’m saying, compared to Taiwan, if you’re...

  • In Taiwan, we’re going to share. That’s the main difference. Open data is actually a national policy. It’s part of our FOIA now. If you have a FOIA request or anything that the government should disclose actively under FOIA, and that there is structured data related, then the government is actually compelled to release the structured data, not just the human-readable data.

  • We also put in our procurement laws that all the vendors need to start evaluating what we call open API standard, which makes machine-to-machine talking much easier without going into a machine-to-machine translation adapter.

  • That also is the ground on which we select vendors, the vendors who say, "Oh, I cannot implement open API," or vendors who say, "Oh, I can build a website, but it’s going to be inaccessible to blind people."

  • They get disqualified for unprofessionalism. We don’t see that in other nearby jurisdictions. Also, I would say the citizen-science scene in Taiwan is really spectacular. People caring about air quality just set up these less than $100 AirBoxes in their balcony, in their schools, in their homes.

  • Just started collecting atmospheric data, and sharing it to a distributed ledger, a DLT. Then we download from that DLT into the collective intelligence platform, the CI platform, which you can find ci.taiwan.gov.tw, which unites the meteorological, water quality, earthquake, disaster prevention data.

  • Just recently, the water quality team was in Wellington and working with the Wellington government on their data to detect water leakage quickly and to have a higher, better turnaround on water shortage problems.

  • I’m saying it’s all about intention. It’s not about data in itself. In Taiwan, we see data as a beginning of a relationship. Of course, the EU, with the GDPR, is now gradually reckoning this view. The GDPR also somewhat sees data as a relation, where they define the relations as the permission to have a portable, explainable, and usable way of understanding accountability of data.

  • Always in Taiwan, we see data as a beginning of a relation. When a citizen scientist contribute this data and store it in the distributed ledger, they know that we in the supercomputing center will not change the numbers they contributed. Otherwise, everybody on blockchain sees that we’re changing the numbers so we’re not going to change the numbers.

  • We decide that since we can’t beat them physically, we join them. [laughs] We are committed to join the AirBox network by setting up similar microsensors of atmospheric data in the places where there’s a higher digital gap. The citizen scientist also tells that they really want a station here in the middle of the Taiwan street to tell where the wind blows.

  • Again, they cannot set an AirBox station there. Even if you have a drone, it eventually run out battery. Because we’re going to set up wind power there, on the turbines, where there’s plenty of electricity, I’m sure.

  • The idea is that we can put it into our procurement template so that the people who provide service to the public through electricity or whatever it means agree to a data sharing plan, so that we can get atmospheric data from that point also into this collaborative, collective intelligence network.

  • Because it’s all open innovation, people around the world just download this open source stuff and put it on open hardware. By default, they report their numbers back to Taiwan. It’s actually a huge amount of data, and it doesn’t have to come just from Taiwan.

  • That’s how we contribute to climate science and other ways to mitigate climate change and things like that. In sustainable development goals parlance, that is the 16.18, the 17.17, and the 17.6, which is just reliable data through effective partnerships that is open to all.

  • It is a very, I would say, island network way of thinking about data and innovation. That is, again, not comparing absolute numbers, but just using the relationship that is formed by data diplomatically, socially, and also through innovation.

  • When you said companies can literally break the law for a year, can you explain what you meant by that?

  • Yes. This is a really new thing. It only introduced this year. There is a website called sandbox.org.tw, where you can fork the regulation. Fork is a computer science term that says you see the regulation going this way, but you think there is a social environmental need, so I want to bring it this way.

  • Whether this works or not for the public benefit, we don’t know. Let’s try for a year and see what works or breaks. If it works, it merges back into the regulation system. If it breaks, well, we think the investors and also everybody learned something.

  • Because it’s open innovation, people try a different angle the next time. This January, it’s the platform economy. This April, we now have the fintech sandbox. We expect in December the UV sandbox. There’s also a 5G sandbox that’s rolling out any time now.

  • In every place where a ministry used to whitelist experiments, they now say, if you can say your idea may have some merit for address of social need, then regardless of what the minister of transportation think you are doing, it could be a drone, a car, a ship or anything in-between because it’s owned by ministry of economy, today it’s all the same.

  • You can propose such self-driving vehicles, and you get to operate after simulation in a publicly visible space, then you get to operate for a year, essentially breaking any law whatsoever. You can challenge any regulation by municipality or even other ministries as well in your proposal of sandbox.

  • I’m always obliged to say, at this point, other than money laundering and funding the terrorism, because we know how those experiments go, other than AML and funding terrorism, everything is fair game. You can challenge any regulation from any ministry if you think it’s better for the society.

  • I didn’t follow in that explanation, if there is a moment for the government to intervene and say, "Oh, hang on a second. That’s a fork too far," if you see.

  • It start with a multistakeholder panel. Of course, if you want to work on, say, AI banking for financial inclusion, an innovator may say, "I think it makes no sense for people to use their ID card over-the-counter to open a bank account."

  • They think that when they had their first SIM card in a Telco, that’s already KYC enough, that people should be able to open bank accounts purely from their mobile phones.

  • Even they have no credit history, they have, after all, paying telecom bills for a year or more. Even for people who are 18 years old or 20 years old, which has no credit history, they think their AI banking algorithm can calculate a reasonable credit history for them for financial inclusion.

  • All of these are of course breaking the ministry of finance regulations, but of course the ministry of finance need to vote on it. There is a multistakeholder panel of which there’s many ministerial seats as well.

  • I think they’re just over half, so by voting it into the sandbox, it still says something about the ministries. Although cautious, they think it may have a chance of succeeding.

  • If this is breaking the law for breaking the law’s sake, which is vandalism, I’m sure that a committee with more than half of officials, public servants, they will not vote yes on such vandalism.

  • If the fintech company came along and said, "Actually, I think the capital requirements, we can get by without, you know, the starting capital," then the committee would presumably say...

  • The committee may say, "For the first year, for the year of experiment, I agree that you don’t need a capitals, but after the experimentation period, I’m going to require that you have such and such capital requirements."

  • There could be strings attached, like the mobile banking money, which is literally the case in the sandbox.

  • The committee said, the first phase which is one year, 4,000 people, I think, "If there is five cases of your algorithm misidentifying the KYC, that means that it’s worse than the current banking record, then the experiment terminates and it’s not ready for the prime time. I’m sorry, but it’s not ready for the prime time."

  • The KYC was a real example of the...

  • It’s a real example, as is the credit history through telecom bills. These are, of course, for our social benefit because they increase financial inclusion, but it’s also of course very good for the startups as well, and there’s many other examples.

  • Minister Tang, you were a sort of entrepreneur in Silicon Valley, and you came back to Taiwan. Was that just purely out of patriotism, or did you think that there were actually real...?

  • The food is better here.

  • (laughter)

  • Literally. Also, I don’t like long meetings. I worked with Apple for six years on Siri and other language technologies, but I’ve never visited the Cupertino spaceship now. I’ve never visited spaceship.

  • The reason why is that, while I really like telepresence, and the time zone difference enables me to say, with my team, "I only have one hour," which is midnight, Taiwan time, and beginning of the day, Silicon Valley time, to have standard meetings, and any meetings whatsoever.

  • If the meeting goes longer than hour, then I’m sorry, but I have to sleep. Otherwise, they have to wake up real early. They don’t do that, either. Because of that, I think the time zone difference makes it easier for me to work on things in a self-describatory fashion.

  • We used to configure our QA, development, and PM teams in eight-hour apart chunks, just like GPS, so that people are encouraged to produce self-describing artifacts. That increases the institutional memory.

  • That’s with another Silicon Valley startup called Socialtext, which is now Peoplefluent. In any case, I worked with them for eight years also. I visited Palo Alto maybe exactly four times. That’s the way I prefer to work.

  • I wouldn’t say I go back to Taiwan. I’m physically located in Taiwan, but I’m still working with the Silicon Valley culture and folks, and also Oxford University Press, which is another eight hours, but in the other direction.

  • Government after Silicon Valley, how is it?

  • I’m still working with the Silicon Valley. I always stress I’m working with the cabinet, not for the cabinet. I’m at a Lagrange point between the citizenry and the government. It’s a lot of fun. Because of my compact, which is location independence.

  • I get to work in places like this, and entertain the company of pursuers of electric vehicles, which are self-driving tricycles that we just try out in Social Innovation Lab. It is a lot of fun. Back when I started a movement around a new programming language, I had this rallying cry, "Optimize for fun."

  • That is still the rallying cry as the digital minister. Basically, I make sure that people, when participating in collaborative meetings in this space, stop seeing things as a zero-sum game between the organizers and the career public service in-between, which is anonymous, but bears all the tension.

  • Rather, a creative space, where given the different positions, people can work out common values. It’s worth breaking the law for a year for those common values. That is, I think, a very fun-oriented way of policymaking, if you can even call it that.

  • Location independence also means I don’t wait for people to come to Taipei. I tour around. For example, this is Bali, and even more remote places, like Taitung and so on. All the 12 ministries are here.

  • We make sure that they see through my eyes, of my ethnographic interactions, which is just a cool phrase for hanging out. They see me hang out with the local people, indigenous, rural, or whatever tribes.

  • They see how their national policies function well or not regionally. They have to answer to those questions immediately. Within 14 days, everything is radically published and radically open. This really changes the civil service.

  • Previously, if they figure out some way to interpret for innovation, and it works, the minister get all the credit. If it fails spectacularly, the ministry always blames them, so there’s nothing in it for them. They still do it, out of patriotism, I’m sure.

  • Now, with this radical transparency and location independence, it’s the other way around. People really see the professionalism of career public service when they brainstorm to find innovative way to solve local issues.

  • Their full name is on the transcript. They get the full credit. When I talk to journalists, I always highlight the particular names of career public servants, like Yang Chin-Heng, who is the Participation Officer of the Ministry of Finance.

  • That address the co-creation issue of the tax filing system, by working with the petitioners to cocreate this year’s tax filing system, and so on. They get all the credit. If things don’t work out, if it doesn’t make it to policy, then the civil society and the private sector just carry on their idea. There’s nothing lost.

  • The full context is available, anyway. If it fails even more spectacularly, as far as I know, I am the only minister doing this in the world, anyway. They can always blame Audrey. For the risk to be absorbed, for the innovation to save everybody’s time, and also to have the full credit.

  • We don’t always deliver on all three concurrently, but we make sure it’s Pareto improvement, meaning that we don’t sacrifice one for the other two. That’s the theory of change.

  • Related questions, how did this come about, and how has it gone down? It just seems to me that, as you say, you’re the only minister doing this in the world. I’ve got limited experience of the Taiwanese government and bureaucracy, but it seems fairly like bureaucracies everywhere, a staid, buttoned-down kind of place.

  • Silicon Valley culture, a transgender minister, all these things that must seem radical to people. How did you come to get your job, and how have other people responded to it?

  • I think people are always happy there is somebody absorbing the risk. [laughs] I get unconditional welcome. That’s also, voluntary association means I don’t take orders, nor do I give orders. I don’t go to the defense minister and say, "Oh, tomorrow, you’re going to do things my way." [laughs]

  • This is totally not happening. Basically, my office is unique, perhaps, in configuration. It’s literally, I can poach one person -- but they have to be a volunteer -- from each ministry. I’m a horizontal minister, and there’s 34 vertical ministries.

  • Theoretically, I can have 34 staff members. At the moment, it’s 22. It means that it a true multistakeholder team. Every ministry that participates brings a different value. Each ministry is a value in itself.

  • I don’t give them command. They rank and score themselves. All I ask, there’s only one ask that I do to them, which is to work out loud, and nothing else. When we only have four ministries joining, we used to have physical Post-It Notes, the legacy of which you still see it on the window.

  • Now, with 22 people, paper-based Post-It Notes no longer works. [laughs] We’re fully digital now. Because I can’t touch state secret, I can always bring this up to journalists without any...This is the real work that we are working on, sorting it into waiting, doing, and done.

  • In each item, you can see which ministries are interested in this, how they sign up, which are the milestones, artifacts, and things like that, which is not unlike any other Silicon Valley startup. This is how the entrepreneurs manage our work nowadays.

  • This is essentially internal startup that involves one person from each ministry that are willing to join. Back in their ministry, there’s also a team of what we call participation officers that also amplifies this way of working, of horizontalism, into branches of that ministry.

  • Also, the Tainan municipality has signed on as well. After this interview, I’ll give you gifts of comic books that we use to explain this whole process. The way we got into this way of work is by design of voluntary association through a entirely cross-silo way of teaming up the ministries.

  • Oh yeah. This is the open government comic of the very taxing tax filing system. It’s in six languages, including indigenous. That’s the basic idea. That’s the training material, literally the training material that we give to people. It’s also in Taiwanese, Hoklo, and Hakka. That’s the theory of change of my work.

  • I meant more of how did the president come to...

  • I’ve been doing this since late 2014. I’m around for long time now, for four years now. Right after the Sunflower Movement...

  • You took part in the Sunflower Movement, right?

  • Oh yeah. Sure. I was a petal...

  • (laughter)

  • ...in the Sunflower Movement, mostly working on the communication infrastructure. After Sunflower, there was a mayoral election at the end of that year. There was also a national forum on economic development where people asked for a open government platform. join.g0v.tw was the end result of that particular national forum.

  • Those two combined into the end-of-the-year election result in 2014, where all the candidates that supported the Occupy gets elected as mayors, sometimes without preparing any auguration speech.

  • All the candidates that are against the Occupy lose their seats. The political will is very clear. Anyone who advocate for transparency, it’s not progressive, it’s baseline. Otherwise, you don’t get to be mayors.

  • Then the premier, Jiang Yi-huah, resigned. A new premier -- I think it was Mao Chi-kuo -- appointed the deputy premier, Simon Chang, Chang San-cheng, in charge of the open government roadmap.

  • All the neutrals during the Occupy, the facilitators, the civic-type people, the communication people, and so on, were invited to the cabinet -- I think there’s an English word for this - as reverse mentors to the ministers.

  • I’m kind of a understudy minister to Minister Jaclyn Tsai at that time. We worked together for a couple of years. While she is still the horizontal minister of cyberspace law, I am in charge of the more digital or technical part of the consultations, for example.

  • We worked together for two years. They made sure that I trained more than 1,000 career public servants in the art of open government. That’s all before Dr. Tsai Ing-wen became president.

  • I think this continuity is also very important because Simon Chang, when independent non-partisan, told each ministry to checkpoint their working progress to the public Internet for Dr. Lin Chuan, also independent non-partisan, to download as the transition plan.

  • This benefits me enormously because I joined the cabinet five months after its forming of Dr. Lin Chuan, but I still get access to the transition documents. I really hit the ground running. Without such a open non-partisan hand-off, that would be very difficult to me to do the continuity thing.

  • I know you say you don’t get involved in state secrets, but clearly you have a giant neighbor that is presumably reading all of this. Is there...

  • Yeah, my radical transparent records. Yes.

  • Is there any kind of qualm that you’re next to a very big, aggressive China and there’s stuff here that would help them?

  • Yeah, I’m sure Taiwan can help.

  • No. I mean is there anything they could use to harm you? Can they learn anything from this...

  • No, I literally think Taiwan can help. They also signed the Sustainable Development Goals, supposedly. 16.10, 16.7, 16.6 is also something the PRC has agreed on, to reach by year 2030.

  • Yeah, but they don’t mean it.

  • I’m not sure about it. I would like to challenge that assumption. I remember the martial law days when I was five years old. That was before lifting of martial law. The Kuomintang was never wrong. It’s the martial law, so they can be never wrong. They can never be wrong.

  • Still, there are factions within the KMT working on 16.10, 16.7, and 16.6. Even within a political environment that doesn’t allow for a opposition party, there’s still people working on the rule of law and due process and access to information.

  • I don’t think these two are completely contradicting each other. I don’t think openness and participation requires a representative democracy. They of course help each other.

  • You can have that, but you could also look at today’s China and judge by most objective standards it’s not heading in that direction. It’s locking people up who are doing that kind of stuff.

  • We’d like to help. We’d like to help not just the PRC, but also all our neighbors who are not painted green on the CIVICUS Monitor.

  • I’m sure that’s great. I’m just wondering, do you have colleagues who worry that you’re making Taiwan vulnerable by opening up the inner workings of government?

  • How so? How so? If there are information that would make Taiwan vulnerable, they will be marked confidential. If they’re marked confidential, I don’t see it. I can’t, actually.

  • What percent of the government’s work do you think you see?

  • I don’t know. I don’t know how many things are confidential. They never come to me and say, "Hey, minister. Here’s some confidential information. I would like you to publish it radically."

  • You must have a sense of the proportions of government work you’re not seeing, no?

  • No, actually, because the cabinet and the president’s office has a very clear kind of firewall. The cross-strait defense and foreign, these are the purview of the presidency. They’re mostly handled by the National Security Council, which is not a cabinet member, by the way.

  • In the cabinet, we talk about, of course, more domestic issues. Even the foreign ministry within the cabinet, it’s separated into the confidential branch, which is NSC-led, and the non-confidential branch, which is more Twitter accounts, more communication-oriented.

  • I, of course, work only with the communication branch of things. I don’t know how large that branch is. I really don’t know. It’s just like I take a day off on military drills. I don’t know how large that command center is.

  • Presume you can’t attend the cabinet because your colleagues wouldn’t want their deliberations made public, but you can’t not make them public, right? How do you...?

  • How do I work? Very simply. There is a radical transparency protocol. visit.pdis.tw talks about this.

  • I mean when there’s a cabinet meeting, can you go?

  • There’s no confidential information in cabinet meeting. We publish the entire discussion of cabinet meeting in the press conference right after each cabinet meeting. All the slides that we see, you also see.

  • Literally, with the transcript?

  • Not the full transcript, of course, but the same slides, the same cabinet meeting agenda. Kolas Yotaka will recite what she just heard from the ministers.

  • Is that because there are cabinet sub-committees or something? There’s stuff that isn’t secret, but it’s highly sensitive.

  • Then I don’t get invited to those. I don’t know how many there are because I don’t get invited to those. This is a DMZ...

  • (laughter)

  • ...demilitarized zone, which is always open to deliberation.

  • There’s also sensitive in the sense of not strictly speaking confidential but controversial or could be made sensational, which is why we allow people in the public service to edit for 10 working days and people from outside, like media, for 10 days to take out the part that are in-jokes that are bound to be misunderstood or to supply with additional information for context and clarification.

  • This is not like we live-stream every meeting. This is within a series of deliberation, people can change the wordings of what they said in the meeting to sound professional. They do do that.

  • The additional benefit is that in many meetings, the different sides are just talking without listening. Sometimes we say, "There’s still a controversy. There’s still a dispute. We will have to schedule another meeting."

  • After 10 working days of reviewing, they actually have to read through the other side’s position. More often than not, they come to me and say, "Minister, actually we don’t need another meeting. We see where the other side is coming from." It’s also very useful internally as a coordination tool.

  • Fascinating. Thank you very much, minister.