Noema editor-in-chief Nathan Gardels spoke with Audrey Tang, Taiwan’s first-ever digital minister, at the Athens Democracy Forum held late last year.
Nathan Gardels: The historian Yuval Noah Harari has commented to me that the most enduring consequence of the COVID-19 crisis has been the decisive shift to a digital society. Everything will have to adapt, including how we practice democracy and design its deliberative institutions.
You are at the forefront of practicing online deliberative democracy in Taiwan. Would you explain how this has worked?
Audrey Tang: First of all, I rarely use the words “deliberative democracy” because it has so many syllables to spill out of your mouth. I simply say that participation by citizens in such a process should be “fast, fun and fair.” Also, participation only works if there is a real effect on power.
Most of the time, people agree on most of the issues around which they can reach a “rough consensus” as the basis for formulating policies that constitute and reflect the social norm. Polarization occurs on very few issues.
As a result of our practices of online deliberation, Taiwan’s president, Tsai Ing-wen, has said: “Before, democracy was a showdown between two opposing values. Now, democracy is a conversation between many diverse values.”
Our thinking is based on the four-stage, focused conversation model:
- Objective: Facts — “What do we know?”
- Reflective: Feelings — “What are our reactions?”
- Interpretative: Ideas — “What insights do we get?”
- Decisional: Actions — “What should we do?”
The idea behind vTaiwan (virtual Taiwan) is that government needs to respond here and now when citizens express concern about a given issue by inviting them to set the agenda. To do so, information must be transparent and available to all. The government and its citizens must have the same information so that there is a trustworthy basis for a public conversation.
Let me use the example of how such transparency was key to dealing with COVID. During that early period in 2020, Taiwan National University’s open-source website listed all incoming flights from Wuhan and uploaded information from health inspectors on those passengers and where they went after arrival. And our Centers for Disease Control set up a toll-free line and website so everyone could get a response in real time to their COVID-related questions.
The point is that when you have a completely open civil society, and people have a free flow of information, they can triage that information to act themselves. This creates a kind of collective intelligence in which responsibly informed citizens can make decisions that not only protect themselves, but others as well, in a circumstance such as a pandemic.
This open information ecosystem enables the government to have quick, real-time and often innovative responses to citizen concerns.
One example: A young boy posted on our CDC site that he did not want to go to school because his classmates made fun of him for wearing a pink face mask. And so the next day, everybody in the livestream CDC press conference, including our minister of health, wore a pink medical mask. The minister even said that the Pink Panther was his childhood hero. As the images from the press conference spread, that student became the hippest boy in the whole class because he was wearing the same color mask as the medical officers.
My point is that radically trusting citizens accelerates the iteration cycle, especially in a crisis. That is to say, measuring government responses to citizens’ concerns in hours is important because it demonstrates to citizens that they have agenda-setting power. That is the beginning of each and every deliberative space. It’s also important that when we are delivering what citizens want, they must have the information necessary to audit that delivery in an interactive way to trust that it stays on course.
Another example of the importance of open information and trust: At the beginning of the pandemic in February last year, the government ramped up the production of medical masks from 2 million to 20 million a day — in a country of 24 million people — and ensured their ready distribution. We did this by coordinating with 6,000 pharmacies across the island. Each and every one agreed to display on their websites their mask availability. From there, we created a real-time map both of where masks could be readily found and also of who needed them most.
When I clicked on the map, I could see in real time the number of older adults looking for masks as well as the addresses, opening hours and quantity of masks each pharmacy had. If a person was queuing in line, they could check the mask inventory every 30 seconds. Once each person obtained a mask, the number would go down. New shipments were also immediately registered as well. The government providing open-source information to citizens to act on their own enabled the rapid and wide diffusion of masks.
Gardels: You have also institutionalized the agenda-setting power of citizens through what you call the annual “presidential hackathon.” Would you tell me about that?
Tang: This involves what we call “data collaboratives,” in which people across the country’s nonprofit, public and private sectors are invited to join with others to address an issue of concern — for example, infrastructure and pipe repairs, novel approaches to climate action, measuring air and water quality, and so on.
During the hackathon, over a three-month period, the public is asked to vote online for the best proposals. Sometimes, the votes involve 10 million people, almost half our population.
The top five proposals out of the mix are chosen as “champions,” and a trophy is handed out. When you turn on the trophy, there is a micro-projector with a message from the president that says something like: “Whatever you did in the last three months will become public policy with all the personnel, budgetary and regulatory changes necessary to be implemented within the next 12 months.” So, that’s presidential executive power with its own regulation written into it. It demonstrates the importance of the power of deliberative space.
We also use a program called Pol.is — which is powered by assisted intelligence (the most basic level of AI) — in our deliberative online practices. It is now integrated into our government institutions.
We first used Pol.is in 2015 when we deliberated about the role of Uber — whose drivers had no professional licenses — as part of the transportation ecosystem in Taiwan. You could see comments all over Facebook and Twitter with differing opinions on whether this was a good or bad thing. In order to pull that together in a systemic way so that there was an objective basis for a discussion, we used Pol.is to crowdsource the facts about Uber’s presence as well as the many comments about the experience of both consumers and competitors. This information was then open for all to see on our government website.
Typically, as in this case, we started out by creating our own social media entry point for discussion with the question: “What do you feel based on these facts?” As obvious as it might seem, this reflective question soliciting emotions is often missing from political discourse, in which people are asked to fit into ideological boxes of preset solutions.
So, in this online space, people were asked to write down statements about their feelings, such as: “I feel that liability insurance is very important.” On the basis of these stated sentiments, we set an agenda for both face-to-face and livestreamed consultations with the public to see how the feelings of some resonated with statements of others.
In our experience, the best ideas — even small ones — are the ideas that address people’s feelings. The Pol.is process is graphically displayed on the screen, so a person can move their avatar closer to the statements they agree with and further away from other less-resonating statements.
Critically, if you don’t feel like you agree or disagree with the displayed statement, there is not a reply button that would allow trolls to disrupt the process of consensus formation. In each case, what we have seen is that the vast majority of participants roughly agree with one of the statements and that people share more or less the same positions on a host of matters. This shows up on the screen by a dense congregation of agreement on one statement, with a few scattered disagreements at the margins. This rough consensus then shapes the policy response. Yes, ideological issues are always there, but not much time is spent on those, and they are passed over where views coincide.
This same pattern, which reveals that most people actually agree with others in their community, has also appeared when we’ve used the Pol.is process outside Taiwan. In Bowling Green, Kentucky, we asked whether the arts are important in science, technology and math education. All agreed that the creativity entailed in artistic creation is as important in creating technology, for example, as in the arts proper. This became the basis for a policy to support arts funding.
This overarching reality of people being able to arrive at a common approach to most problems simply doesn’t get much attention in either mainstream media or social media. Yet, consensus is the measure of success in political life. Once common norms and values have been established, innovative solutions can follow.
Gardels: You said that participation only works if there’s a real impact on power. Let’s take the Uber example. What was the rough consensus, and how did the government follow up with legislation?
Tang: The unlikely consensus was that, whether you hail a ride from an app or not, taxis don’t need to be any particular color, like yellow, and that, cab or Uber, they need to compete with each other on fair grounds.
And so that meant all had to have the same insurance and registration. That is to say, all cars are free to operate with innovations such as surge pricing. But they can’t undercut existing meters because a meter is the basis of fairness. Apps and dispatch systems are fine. As long as they’re registered, their insurance is taken care of and the competition is fair, people are happy.
With this consensus, these rules were put into legislation. Uber threatened to quit for a couple of months, but because the consensus was so strong, in the end, they stayed put. Uber is now just another part of the local fleet, along with the other taxi companies and co-ops, which are also allowed to do surge pricing.
This adaptation of a company to the consensus reached through deliberation was also the case when we negotiated with Facebook about transparency in advertisements for political issues during the election. We see the same pattern time and again on various issues because, once registered through transparent processes of deliberation, the social norm is so strong that they don’t have much of a choice.
Gardels: Was the legislature bound to take up the citizen recommendations, or did they just do it because it was a consensus?
Tang: In Taiwan, we have a system where the presidential administration prepares the draft of laws for the legislature. And the administration operates in a largely non-partisan space. There are more independent cabinet members, including me, than those from any particular party. We are accountable not just to our constituents, but to future generations. So, we always take into account environmental impact and maintain a longer-range perspective.
In parliament, of course, the party dynamic is at play. There are four major parties. But when we show that we have widely consulted stakeholders and defined common values, then parliament has the basis to work from our draft to make it more coherent with other legislation. Our main contribution is that we have proven through our consultations that we have discovered and defined a level of support to qualify a measure as a social norm.
The parliament is not bound to pass legislation exactly as we proposed it, but the administration is bound to propose a response to the public.
Gardels: Does this process work as well for controversial issues, such as same-sex marriage for example?
Tang: Yes, definitely. Each ministry has what we call “participation officers.” These people are dedicated to connecting with citizens who are engaging online with controversial issues, sometimes via a hashtag. The point is not to convince people one way or another, but to ensure that a rough consensus reached in the deliberative sessions is shared widely among all those concerned about a given issue so that they can take that back to their respective communities to build support for a solution. At the end of the day, that is what is needed for the results of each and every citizens’ assembly or sortition exercise to spread. If it doesn’t spread, it doesn’t reach the level of a social norm and make an impact on society as a whole.
The role of participation officers in the same-sex marriage case was to make sure that those who are for marriage equality and those who are into more traditional family values were invited to consider the overlapping points of consensus discovered in the deliberative sessions.
The point of consensus that emerged was that the same rights enjoyed by those in heterosexual marriages should be extended to individuals, including single women who use artificial reproduction. The driving point agreed to was that we need to prepare society so that the children of any arrangement have the same opportunities and freedom from discrimination.
By changing the civil code to reflect this, we in effect ensured that a same-sex married couple and their children would have the same rights and duties as traditional families — but determined through legislative bylaws instead of magically becoming law through heterosexual marriage.
This solution accounted for the concerns of both sides, agreed on a common value and came up with a way to make it work practically. This is a good illustration of how deliberation is not just about compromise — it is also about innovation.