97. How to fight fake news and strengthen democracy (Audrey Tang)

0:28Thanks for listening to The Rest is Politics. Sign up to The Rest is Politics.

0:38The spirit of innovation is deeply ingrained in America, and Google is helping Americans innovate in ways both big and small. From improving emergency response times to protecting against cyber attacks, local, state, and federal governments are innovating with Google AI. This is a new era of American innovation. Find out more at g.co slash American Innovation.

1:26Sign up for your$1 per month trial at shopify.com slash special offer.

1:37Welcome to The Rest is Polity's Leading with me, Alistair Campbell. And with me, Rory Stewart. And today we are interviewing a member of the Taiwanese cabinet, Audrey Tang, who is the first global minister for digital and who has done the most astonishing things. in particular, did a very unique COVID response in Taiwan, which had a completely different approach and achieved spectacular results, transformed the way the Taiwanese government engaged with citizens, really by changing democracy through technology, creating forms of virtual citizens' assemblies and much more. But Audrey is also a fascinating figure in her own right.

2:18And her journey into government, I think, is also something worth exploring. So on that, over to you, Alistair. Now, Audrey, a couple of things looking at your childhood that strike me as being really, really interesting. First is that you had to really fight against pretty serious illness as a child. So I'd like to ask you about that. But also, you do seem to be extraordinarily, ridiculously clever. You were reading classics at five. You were studying advanced maths at six. You were programming at eight. Then you dropped out of school at 14 and started your own full-time programming business, age 15.

2:56So just take us through your childhood of those two elements of it. Sure. I was born with a heart defect. That simply means that whenever I get too upset, either too happy or too excited or too angry, I would just faint and maybe wake up in a hospital or something. And so I learned very early on to be moderate in my emotions, practicing Taoist breathing and Qigong and many other things. that's one part but that also means that I get to focus much more on programming when I was eight I was very interested in mathematics but not in math not in calculating and so I first learned programming in basic without a personal computer just writing on a piece of paper but after a while my parents did agree to get me a personal computer and then I spent a lot of time just experimenting on the computer, all sort of ways that I can make it visualize or interactively display all sort of the neuromathematical ideas that I was learning at the time, like chaos theory or strange attractors and things like that.

3:59So that was fun. But I went to three kindergartens, six primary schools, and one year of junior high before finally dropping out with the blessing of the head of my school, who said that since who are already doing research on the internet with the people who publish their preprints anyway. So she would not stand in my way to do more advanced research. And so after a year, I started a company with a few of my friends. If people listening to this, some people will have also been, not many, but some will have also been very talented children. Are there lessons of some of the difficulties of being more of a prodigy as a child?

4:36What were the positives? What were the negatives and what lessons would you share with other people who've been through a similar experience? That's a great question. I think there's two things I would share. One is that it now takes almost no effort to find one's intellectual counterparts online. Whenever you feel like contributing to open source, to the internet commons, to co-create something online, and so on, it is not actually difficult to find the counterparts. So the earlier you get to the internet and to, as I mentioned, get blessings on your environment to participate in research and collaborative development, the better.

5:16The other part I would say is that even though that somebody may look like that they're very mature beyond their ages, it doesn't mean that they're comprehensively so among all the different areas. And so it is quite easy, for example, to see somebody very erudite and like saying very complex sentences very early, just like, you know, Chagibity or some other language models, and assume that they're equally mature when it comes to emotional intelligence, when it comes to social settings and things like that. But actually, these areas take longer to mature. And so very clearly know oneself and also communicate clearly that which parts of you are not as mature as your other intellectual pursuits.

6:01I think that also helps to set and manage expectations. Sometimes when you were a child, did it feel a little bit isolating? Because presumably when you were much younger, when you were eight or nine or ten, you must have been confused and sometimes frustrated by the fact that other children or even other adults were not able to focus on the issues that you were able to focus on. So you must have felt at some level a form of difference between you and other people. Yeah, indeed. When I was eight, I was bullied in school and it really perplexes me. Why would the second in the class to basically wish me death so that he could become the first in the class?

6:38Because to me, it does not actually increases anything that he learns or anything like that. Because I work with intrinsic motivations more, I cannot fathom why would people behave that way. So I actually read a lot of Piaget Winnicott and the other child development psychologists just to get myself to a frame where I can understand, you know, the family pressure, the relative lack of self-confidence, the relative status that they're anchoring this child's self-worth on and so on before I can finally see, oh, that is why they're behaving this way. Tell us a little bit about your family's story as a way of thinking a bit about the bigger story of Taiwan's development.

7:20So my father's mother, my grandma, was born near Lu Gang in Taiwan. And she speaks like Dai Yi and also Japanese, Nihongo, growing up. And my father's father comes from Sichuan and so speak Sichuanese and Mandarin. And he came to Taiwan after the relocation of the ROC armed forces and the government to Taiwan in 1949. At a time, the island was facing severe challenges, not just food shortage, but also lack of industrial infrastructure, among many other things. In the 50s, the government implemented land reforms, redistributing large landowners to tenant farmers and so on. And so the 60s, we saw Taiwan shift toward export-oriented industrialization in a stronger agricultural sector and leveraging the low labor costs to attract foreign investments and so on.

8:15And in the 70s and 80s, we continued to diversify our economy, moving to a higher value industry such as chips, which would eventually become a global leader. And in the 90s, we democratized, lifted the martial law and became so-called one of the four Asian tigers. And that's the time when I started learning personal computers and there's widespread democratization costs to directly elect the president, for example, in 96. And eventually that affected quite a few constitutional amendments that added a lot of things like universal healthcare and so on to our constitution. As you know, this podcast is mainly about politics.

8:54And you became effectively a politician, but in a very, very different route that most politicians that we talked to took. You were part of this movement that kind of used tech to challenge the government, occupy parliament, and then became part of government. And you actually, you've had many firsts in your life, but one is that you became the world's first digital minister. So you took all that stuff that you developed and learned through childhood, and then took that into government. So just talk us through that process of how this protest movement became the heart of a government. Certainly.

9:33So after working with the open source community and many Silicon Valley companies on what's called enterprise social, which is productivity software, bringing people together with wiki or micro blocking or things like that to foster the sharing of knowledge within enterprises. We were caught into action, so to speak, in 2014. Now to set a stage, there was a widespread demonstration by students mostly, but also people who were worried about a trade deal with Beijing at March that year. And we have more than 20 NGOs, each on a side of the parliament. And before they actually broke in, the students broke in and occupied the parliament peacefully for three weeks.

10:17The reason why it is one of those very rare occupies that even with half a million people on the street, it did not escalate to violence, neither did it go nowhere. That was because it was live streamed throughout. And so G0B or GovZero, which is part of the movement that fosters the civic technology and helping the civil society organizations with digital tools, helped to facilitate conversations among those people who went into each corner of the occupied parliament. On one corner, they may have a deliberation about the 5G, sorry, 4G at the time infrastructure by the likes of Huawei or ZTE and so on.

10:59Another corner may be the impact on labor conditions. Another corner may be on environment. Another may be on the publishing industry and so on. And so with the digital tools, we radically trusted the people who participated on the street and online so that every day you literally see those uncommon grounds being built. people. So it's like a very large scale citizens assembly, although it is not randomly chosen, but rather self-selection in a sort of self-selection among like open space technology voting by walking to which booth. So yeah, so long story short, the people who occupy the parliament open up the conversation in a nonviolent way with half a million people on the street.

11:41And so people could every day see the rough consensus that was reached among the occupiers. So in many senses, it was not a protest in as much as a demonstration that shows the Speaker of the Parliament that people do actually coalesce into something like a rough consensus after three weeks of nonviolent Occupy. And Speaker once simply said at the end of Occupy, yes, that is the direction we're going. And so the trade deal was cancelled and people embraced this idea of crowdsourcing at the end of that year. Audrey, please help us understand how practically this system works. I mean, how on earth do you get 500 ,000 voices to reach a consensus and generate anything workable?

12:24Certainly. So in each corner that deliberates one particular issue, you can, for example, talk about a 4G infrastructure, whether we open up the core network to the likes of Huawei and ZTE. Another corner may talk about labor or some other things. And so when people have a conversation around that issue for each table, for each corner. There may be a facilitator or one or more people who record this in real time. And in any case, it was always live streamed. And so the facilitator, many of them trained previously by the youth development agency in Taiwan, is experts in what's called the focus conversation method, which is one of the facilitating techniques that tries to get people to agree on the common facts and then reflect on each other's feelings.

13:15And so after a while, you start to see common feelings and the common acknowledgement of common facts. And those were then registered and published online to serve as agenda for the next day's deliberation, which tends to be more focused on specific issues. And Audrey, what lessons did this teach you about democracy? Because often when we look at politics in the United States or the United Kingdom, we think about politics in terms of two political parties attacking each other. But you seem to be describing somehow 500 ,000 people with a facilitator finding common ground much more easily. So what are the structural differences?

13:54Why does it work in this online platform better than it works in a parliament or a senate? So first of all, that was a hybrid exercise. Most people participated on-site, not online. So when I say half a million people on the street, I mean at the peak of the demonstration, like literally that many people went to the places around the parliament. But of course, people who actually had time to stay a full day for deliberations is a fraction of that. But nevertheless, we have many people who were not just interested, but also augmented with digital tools. They're willing to first check the facts, like they can enter their company name or company number, registration number to see exactly how does the cross-trade deal affects their particular business.

14:42And also the fact that it was registered, like transcribed and put on the record online also made sure that people spoke in a more civilized fashion as opposed to, you know, off the record, which many occupies tend to escalate or spur out of control. So I think the main lessons are first that facilitation in a way that de-escalates the situation and focus on the issues at hand is very important. But equally important is the live streaming and the fact that there's no escalation that would go unnoticed. The tech tool that was at the heart of this is called POLIS, P-O-L-I-S. And Rory and I are talking to you, we're both in Scotland at the moment where polis means police.

15:26And I just wonder whether you've presented a very kind of positive citizens assembly type view of this, but is there not also a risk that this can be used for bad as well, used by authoritarian governments, used by bad players in the tech field? So how do you weigh up that balance between technology that can be used as a force for good in decision making and the same technology that can be used for bad. There's many implementations of this online deliberation idea. In 2014, during the Occupy, I believe it's mostly Hackpad and Lumio. And we started using Polis the next year in 2015 when we deliberated about the Uber case.

16:09And you're absolutely correct that any tools like this can be abused. And I still remember that when we started to deliberate the Airbnb case after the Uber case, Airbnb Taiwan literally sent an email to all its members in Taiwan saying that, please go on the police platform and vote exactly this way to support the Airbnb line and so on. So it is not completely unheard of that people would like to mobilize and to kind of game the system somehow, and so on. There are two things that we have done that really improves the resilience against such ways to capture it. One is that on the Polis platform in particular, when you see people's feelings, like how do you feel about Airbnb or about Uber, it is freeform.

17:00So it is not just people voting on the predetermined questions. It is actually people sharing how they feel. And so when they see each other's feelings, which tends to be more nuanced than the standard questions, then they get into a much more deliberative mood. And so from our understanding, less than one third of the Airbnb people, members in Taiwan actually voted along the Airbnb lines. And second, the system, unlike traditional social media, it does not highlight the divisiveness or hate, but rather it acknowledges the divides, but surfaces the kind of sentiments like in Uber. People would say search pricing when the demand is high is fine, but undercutting existing meters is not fine.

17:44So it tends to surface those ideas that gets broad acceptance among the people who are on different sides of the ideological divide and just say that, oh, we will sort this to the top and so lead by example. And so other people would then propose more nuanced, eclectic ideas. And so these two designs means that even if 5 ,000 people gets mobilized and vote exactly the same, they're just one dot in the visualization, they still have to bridge across the differences in order to surface the ideas that speaks to more than one science, essentially. And Audrey, we're obviously in a world in which we're very troubled by social media, troubled by X, troubled by Facebook, troubled by the ways in which these algorithms seem to encourage polarization and populism.

18:30Do you have any reflections on, firstly, whether you agree with that? And if you do, what could be done to improve the way that social media interacts with politics? Yeah. In Taiwan, we've learned long ago that this kind of town hall or online alignment assemblies or this hybrid citizen assembly or citizen's initiative or participatory budgeting need to take place on an actual public square supported by the civil society and the government instead of taking place on private sector advertisement field, so-called social media. And so we built our own pro-social social media platforms, like join.gov.tw, which is the one-stop shop for all sort of citizen engagements online.

19:16So separate from Facebook or x.com. But I also see that, for example, community notes on x.com, they specifically learned from the polis exercise that Taiwan did. And so nowadays, you can actually co-create the notes that gets attached on those viral X posts. And only the notes that speaks across the different party lines, different ideological lines, except by both sides, gets amplified into visibility. So I do see that these bridging algorithms are also being used by more mainstream social media, but much more pressure needs to be done from the civil society and from the regulators before we can see the social media that prioritizes this kind of pro-social ranking in Finland.

20:03Audrey, how was this all used when it came to COVID? And how did that impact upon the way that Taiwan handled the COVID outbreak? So a lot of the co-created policies that were deliberated on the Join platform, on the VTawan platform, ended up getting the investment from the state as digital public infrastructures. So many years before COVID, we already invested in the so-called civil IoT system, the Internet of Things system that established real-time monitoring with citizens' participation, air quality, and water quality, and on many other things environmentally. And so when COVID happened, people repurposed this real-time API infrastructure.

20:51But instead of showing the air quality and so on, it shows the next available mask for purchase when we started rationing out masks very early on in 2020. And the same group, the G0V, Gov0, Civic Tech people, as the Taiwan COVID response mandated contact tracing, also came up with a way that is much more privacy preserving. They came up with the idea of SMS-based contact tracing, where you just generate a random code as a venue owner, and the person who visits you scan that random code and send the SMS to 1922, and that's it. So the venue doesn't even learn the phone number or anything, and privacy is preserved.

21:34But that held forth until Alpha and Delta, until the first wave of Omicron. And Audrey, can you, for international listeners, explain in simple terms what this meant for Taiwan during COVID? What type of lockdown did you end up having, and how was it different from other people? And what was the consequence in terms of public health? So Taiwan, for the first couple of years into the pandemic, achieved, I think, among the lowest global death rates during that stage. And during all these three years and a half, we've never had a citywide lockdown. We, of course, had quarantines in border control, but we did not impose anything intercity.

22:15And all the TSMC and other companies, manufacturing sector, kept working throughout the pandemic. And so that led to, I think, one of the fastest rates of economic growth in the world during those few years. And it also led to a kind of solidarity between our different sectors that led us counter not just the pandemic, but also the infodemic without government-mendated takedowns on social media platforms. People, by and large, using community node-like techniques, inoculated each other's mind by contributing to collaborative fact-finding. In addition to having been the digital minister, you also minister, this is a great title, Minister for Social Innovation.

22:54What does a minister for social innovation do? I was ministering in charge of social innovation, youth engagement, and open government. So social innovation to me means that it is a larger sector, not just the more traditional co-ops or the social enterprises, but it also includes the advocacy groups, and it also includes decentralized autonomous organizations and many new forms of organizations. As long as there is a clear way to measure the social impact of such organizations, we worked to make sure that the social sector, as we call them, have agenda-setting power when it comes to, for example, public infrastructure.

23:36So social innovation means that people closest to the pain, no matter how they're organized, are given the voice to not just voice out the pain, but also share the kind of solutions that they prototyped on the local scale. So that through mechanisms like the presidential hackathon every year, we can uplift those local ideas, those local innovations into national infrastructures. Do ministers in Taiwan have to endure the same sort of abuse online and vile treatment from members of the public that most politicians in most democratic parts of the world have to endure? Well, I mean, I practice troll hugging.

24:17And so during my tenure of seven and a half years serving in the cabinet, there's no shortage of trolls for me to hug. And so yes, in a nutshell, but I think two elements are different. One is that because I'm nonpartisan or non-binary partisan. I do not actually go to any party rally or identify with any party. So in this sense, I can serve the constituents without the more like party colored kind of attacks that other ministers in other political systems tend to kind of have to draw the line of which party you are in. And the other thing is that because we have a very mature way of turning these complaints and so on into invitations of co-creation so that many trolls that engage me online, I make sure only to respond to the constructive parts.

25:08And they ended up contributing quite a bit into social innovation. And when I invited them, for example, to my office hour, used to be like every week, the social innovation lab, many of them actually do come and contribute meaningfully. And so I think the way to kind of harness this conflict into co-creation is also quite interesting, quite unique in Taiwan's high trust society. As a very smart person coming from outside government into government, what were the things that surprised you about government, both negatively and positively? The most surprising thing to me was that how many innovators are there in the Korea public service, how many innovators there are, like usually just around the section chief level, a little bit above and below, they actually are very close to the pain.

25:59They actually do know how to solve wicked problems. They may not have the blessing of the directors general or the ministers to implement their plans, but they actually have it figured out like years ago. So that was the most surprising thing. It was just because of anonymity, they could not talk about it. And so a lot of my work as minister is to design safe spaces where they can pseudonymously participate, like in the presidential hackathon, or that they can work with their civil society counterparts. And if the ideas didn't work, they can say, oh, I'm just collaborating with civil society. But if their idea wins the top five place in the yearly presidential hackathon, the trophy is a projector.

26:41If you turn on, it projects the president giving you the trophy. And so if they get a presidential blessing that their idea will become public infrastructure project in the next fiscal year, then they basically change their career, right? So So by designing ways that lowers their risk and save their time when it comes to public sector innovation, we were able to engage many innovators across the different parts of the departments as well as in the local government. And Audrey, give us two or three really good ideas that came out of these hackathons. One of the good ideas is to invest in the digital public infrastructure of sharing data in such a way that does not compromise privacy, but allows people to analyze the trends when it comes to public health, when it comes to like sports data, people like me that has specific conditions, like what kind of exercises tend to be more useful and not threatening my health condition and so on.

27:39So in many jurisdictions, people were still stuck in the old ways where you have to go to an enclave or you have to like trust people to process the private data. But very early on during the presidential hackathon, we realized that these analysis can be done in a way that does not involve the raw data leaving the silos, but could instead using like zero knowledge and advanced cryptography to have the researchers put the algorithms to those individual silos so that they can get useful analysis for either environment, public health or sports or anything, and do not compromise the privacy in any way.

28:19And so this kind of investment of computations over private data without compromising privacy as an infrastructure, I think that's one of the very good ideas that came out of the presidential hackathon. Maybe one more non-tech example of an idea that was generated by people that you thought was good. For example, there was another team that came from one of the remote islands in Taiwan. And prior to the presidential hackathon, because the people who suffered armed at one surgery and so on, the local practitioners, the local medics were not fully trusted by their family. They insisted that they fly a helicopter back to Taiwan for the full treatment.