Hello.

Hello. Can you hear me?

Yes, I can hear you now. Can you hear me all right?

Yes, I can hear you all right.

Fantastic. Well, thank you so much for doing this.

It’s my pleasure.

I can leap straight into it so as not to take too much of your time. You had, from me, the explanation about who we are and what the party is. The context in which this will be shown is, of course, our WE Party Conference. It’s ahead of the debate around whether data-driven technology can help create equality, or is part of the problem, or both things. That’s the context.

We’ll just get a segment of this conversation for broadcast. We’ll show you what it is, obviously, before we do.

That sounds great.

I’ll have the entire recording that you’re already recording.

Yes.

I’m free to publish it afterwards, after you had your...

Exactly.

...display. Time-wise, I have two hours, so we can talk about pretty much anything.

We’re intending to have a much shorter excerpt, but I’m happy to talk to you for longer. There’s a lot I think I can learn from you. Thank you very much. I’ll begin as for this segment now.

Minister, thank you very much for finding the time to talk to us today. I’m very excited to talk to you because there are very few politicians who I would say are engaged in transformative politics. There are fewer still who are finding ways to harness technology to that transformation.

I wanted to talk to you in the beginning about the Sunflower Movement. There wasn’t as much publicity around that in the UK as you might have imagined. Would you mind starting by explaining the Sunflower Movement and your involvement in it? It was really something quite amazing.

Certainly. The Sunflower Movement was a Occupy movement for 22 days in March to April 2014. During those 22 days, students and activists occupied the parliament. They did this for demonstration, but not demonstration in a protesting sense, but rather demonstration in the demo kind of sense.

The reason of the occupation was that the MPs at the time were refusing to deliberate a trade service agreement with Beijing. Because of that, the students did the deliberation for the MPs because they went on strike. They occupied the parliament and did the deliberation with the people.

At the height of the Occupy, there was about half a million of people on the street and even more people online watching the live stream, the real-time transcripts, the various discussion boards, and all the different products of this massive Occupy movement.

Two things distinguish this Occupy movement from other Occupy movements around the world. The first one I think is very important, is that during the 22 days, mostly non-violent, every day people converge a little bit more on what they think about the trade service agreement.

This means that more than 20 NGOs, concerning labor, concerning equality of the genders, concerning, for example, environmental protection and so on, each had their own kind of corner around the occupied place.

They were all linked with a kind of nervous system that channels all the consensus made, all the points made, all the people’s feelings into a shared document that is shared by all the more than 20 NGOs in the Occupy. The end result is that like cross-pollinating ideas, people would just go from one NGO’s booth to another. Over the course of a day, maybe they would visit five different corners.

At the end of the day, people did a recap, a synthetic document. By the beginning of the next day, people just talk about the things that they did not yet have consensus. That’s the first distinguishing factor in that it converges rather than diverges. The second factor I think is...

Sorry to interrupt.

No, it’s good.

In Taiwan, there had been a dearth of democracy. You had a government that was not listening to the people. Instead, what you got through the Occupy movement was not only people coming in to make their views heard, but you found a way of having a discussion that created greater understanding around the issue and greater consensus around the issue.

You moved away from polarizing politics.

That’s exactly right.

And you were already using technology for this as well as live debate.

That’s right. The second distinguishing factor is that we allowed people who participate in any corner in the street to view a real-time live streaming of what’s happening inside occupied parliament.

There’s also a live chat channel where people just type in what exactly they were hearing from the 20 or more debate stations so that people can just at a glance see what the topics are being debated as well as getting the consensus that’s being made from this debate on the street.

This is a ICT technology, but it is for enabling people to listen to hundreds of thousands of people rather than speaking to hundreds of thousands of people. We often see social media or even television or radio used in a broadcasting manner, that is to say, it enables people to obey, follow the party line, or to somehow get the talking points from one or two people with stellar.

This is the inverse of that. This technology enables tens of thousands of people to listen to one another. It enabled the people who did occupy in the Occupy parliament to listen to those people and get them their voice as a single voice. This is a deployment of what we call assistive civic technology that led people as points being heard in a way that is accountable.

It’s moving away from echo chambers and towards a way of facilitating debate and a way of creating information. Sitting here in England where we’re now living with the results of a referendum, which in some way looks like people power, but of course was based on only partial and very polarized information, it’s a very, very interesting piece of history and with an incredibly interesting outcome.

What happened after all of these debates?

Afterwards, I think at the 21st day or so, the general occupiers as were participants converged on the set of recommendations toward the trade service agreement. The head of the parliament and head of the MPs accepted every single one of them and so the occupy was a victory.

The idea, very simply put, is that people want the treatment with Beijing to be treated as any other foreign treaty to be deliberated by the MPs in exactly the same way.

They also called for a national forum on making a real-time responsive deliberation system modeled after the ICT system deployed during the occupy so that people can look at any part of the budget, look at any part of regulation, do e-petitions and so on so that they don’t have to occupy again if some other controversial topic comes up.

During the National Forum, it was decided that we did a national e-participation platform called the Join platform. Of the 23 million people in Taiwan, about five million at the moment is using the Join platform.

Five million people using it on a basis that helps form policy very directly.

On a active basis. That’s right. About one-quarter of people who can use Internet. Of course, that number can still grow, but still, I think we’re performing much better than other countries.

What was your role in this? You came into this as somebody with Silicon Valley experience, which is also fairly unusual for a politician.

Yeah. At the time of the Occupy, I was a independent consultant to a company called Socialtext, which is called a Wiki company.

We make wikis, we make microblogging, we make a lot of so-called social media tools but for Fortune 500 companies as well as large non-profits in order to enhance their internal communication and make sure that the knowledge is captured within the organization rather than inside specific silos.

Basically, we do something that’s like Facebook plus Twitter plus Wikipedia, but with the aim of getting people’s things done in the flow of work, rather than just keeping them on the website and watching as many advertisements as possible. It’s the same set of technology, social technology, but with a very different goal in mind. It is deployed to people’s business hours rather than off hours.

That’s my background. I’ve been working with Socialtext since 2008, so quite a long time. At the same time, I’m also a consultant and independent contractor in Apple’s Siri team. I’m also working on machine learning, on cross-language understanding, on semantic understanding, and things like that.

I bring my contribution to the Occupy mostly by setting up the communication network and making sure that fact spreads faster than rumors. This is easier said than done because rumors have a way to provoke outrage. Outrage makes people share messages even before they fact-check them.

The only way we can stop it that makes facts spread faster than rumors is to make facts fun, is to make facts something that’s interactive, something that you can just type in your company, registration number, or your trade you’re doing and you see a very interactive graphics that shows exactly how the trade service agreement affects you. Of course, all these are done by a wider civic tech community.

My main role is just to maintain the portal, the information portal of the Occupy movement, and making sure that all the different endeavors that makes it possible to have informed discussion is aware of each other’s existence, as well as maintaining the backend infrastructure against the cybersecurity attacks and things like that.

I don’t know of any other part of the Occupy movement that’s actually ended up in government anywhere in the world. You actually went from this uprising, this popular uprising, to actually being in government and putting these things into practice from the top down, even though what you’re talking about is something with much more participation and civic engagement.

It is somewhat comparable with the 15-M movement in Spain, where many Occupiers at the time was also becoming like Madrid mayors and also people in various important cities.

You are now Digital Minister. As I say, you’re bringing all of this to bear as not just something that is part of a protest, but something that is everyday politics.

That’s right. We see our demonstration as a demo. [laughs] Now I am deploying demo into production, as we talk about this, yes.

[laughs] When you were in Silicon Valley, you also saw at close hand some of the difficulties with that culture. One of the things that people worry about technology is that it’s been made by a small and not-very-diverse group of people with certain ideas about what it can do.

Do you worry about this as well? Do you think that technology can be the answer to the concerns around technology?

It is a concern. At the beginning, when we were doing the mobile version of the Socialtext, what we call Signals, which is very much like Twitter, we didn’t anticipate that it kind of has a effect on the habit of people.

Back in 2010 or so, it’s like if a person only installs one messaging system on their phones, it actually increased their productivity. If a person installs three or four instant message systems, it’s like a cocktail effect on the mind because the mind would be constantly context-switching.

It actually, from a mental health perspective, puts people in a always adrenaline rush or a fear-of-missing-out state that is not conductive to the kind of deep listening or deliberation that we were just talking about.

We were part of the problem, I guess [laughs] , but we’re, I think, also aware that it is possible to have technologies that reduce the demand on attention and indeed the inherent bias that certain technology has introduced.

In early 2000s, I also participated in the so-called spam wars. At that point, people thought that email is helpless and email may soon go away because it costs nothing to send junk mail. It wastes everybody’s time. It also degrades the trust that people put on each other’s messages around the Internet.

The solution to the spam problem was not a single, top-down action, a law, or an act. Neither was it a single technological change. Neither was it at a protocol change. Neither was it a single intervention by any civil society. Rather, it is all the different points that I talked about coordinating in a way that is accessible to all the stakeholders involved.

Every single action increased the cost by spammers just a little bit and reduced their expected reward just a little bit. Taken together over maybe three or four years, we delivered a coordinated action that made it much more manageable. Now people don’t complain much about spams anymore.

I think this is one of the blueprints that I’m using in order to look at the manufactured addiction problem, the inherent bias problem, machine learning, the people’s distrust in general.

For example, algorithmic decided outcomes and things like that, all of it is much easier to solve if we get all the stakeholders on the same table, in a continuous relationship, and mapping out exactly what everybody’s stakes is and keeps it transparent and accountable, just as we did in the Internet society and related organizations during the spam war days.

I think there are a set of technologies, civic technologies, that can help solving these problems, but I don’t deny that there are also actors that would like to maintain the monopoly on precision persuasion for a lack of better term. This is an ongoing dialogue.

Also, that if you are building these kinds of things, it’s very important, presumably, to have real diversity in terms of who is inputting to these technologies. You talk about bias. Presumably, you would see a danger if there are too few women, as we know there are, in STEM, for example, that this can replicate biases then in terms of what is produced.

When I founded my first startup in 1996 in Taiwan, actually people majoring in computer science or working on IT and so on, the gender ratio is very healthy. It’s close to one on one.

When I got online, I discovered that the free software community, the maker community, and so on, many women are forced to use male-sounding nicknames, not because they identify as transgender [laughs] , but rather, they did this to avoid harassment.

It’s apparently a very important and significant issue on the Western, English-speaking world. That took me completely by surprise because Taiwan is not like that. [laughs] I think this highlights the importance of participation. In Taiwan, we have different problems. For example, the indigenous people, they don’t participate enough in the design of everyday technology that affects them.

As are the other ethnicities because Taiwan is like 98 percent ethnic Han. The other ethnicities’ voices don’t often get heard simply because of their language or their lived experience differences. One of the main points in diversity is not just getting sufficient number of people, although that helps, like our spokesperson now is a indigenous women.

Our President is also partly indigenous, she’s a woman and not anyone’s daughter and wife. We think this is very important. Rather than just diversity, I think real inclusion means that all the different people participating in the end result of the designs, such as, for example, in our K-12 education, we emphasize that people use technology to work with children. They must prefer open source technology.

That is to say that students have a say in where the technology is doing, like the access to machine learning and computation resources. There must be no difference between the city and indigenous or rural areas, things like that, and also broadband as a human right.

What I’m saying is that it is very important to have diversity in the community that makes things. Even more important is to have full inclusion in a set of users, in a set of people who use these technologies so they can fully inform where the technology is doing and in so doing democratize the production of technology itself.

Thank you. That’s an incredibly useful way of looking at it. I’m interested if you have any particular thoughts about how we as a small party operating in a system that is in many ways stacked against us might use technology ourselves or think about using technology ourselves.

I’ll give you one example. There was a lot of concern here around the Cambridge Analytica controversy. What that actually showed, although this again got very little coverage, was not just about that one incident. In fact, there are many people who legitimately provide such services, where they get a great deal of big data around the electorate. They analyze it, and they are able to zero in on what that means.

The bigger parties that can afford to pay for that therefore have an advantage when it comes to, particularl y in a first-past-the-post system, in terms of targeting their voters.

One of the things we have to think about is how we can be clever and do things that try and mitigate some of these in-built disadvantages under which we are acting. Do you have any particular tips for us about what that might look like?

In the g0v movement, which is not just about supporting Sunflower Occupies, but also about making more attractive and fun open alternatives to the government websites, for all the government websites that the g0v people don’t think are useful or attractive enough, they end in G-O-V-T-W, those government websites.

Like the legislative is L-Y-G-O-V-T-W. The g0v people will just make a new domain name, ly.g0v.tw. Basically, by changing an O to a zero, you get a shadow government that is more attractive and more participatory.

One of the recent interventions that g0v community did, which get a lot of press attention, is the Councilors Voting Guide, which includes the Mayors Voting Guide. We have a local election coming up in about 90 days from now.

The Councilors Voting Guide is designed to maximize the people’s informed information before going into the voting booth. Not only does it include all the voting records and all the political career over a candidate’s career, what they voted, what they advocated, and what are their disclosures, spending, and things like that.

Also, it innovates by, if you go to your precinct or your region, it lists all the councilors in a random order and with random color. What this means is that all the smaller parties’ candidates and all the independent candidates get as much the same coverage as the large parties in the voter’s guide.

They also crowdsource for newcomers their platforms in the form of a small, short YouTube video. They also allow people to sign in in their social media profiles and vote for the people they want. They even have some grants for people who receive the most number of people’s likes and things like that.

In all these cases, they act with what we call the ACE principle. A means actionable. It means that if you support a small party’s candidate, there’s something that you can do with five seconds’ time. There’s something you can do with your five minutes’ time or something you can do with five hours’ time. That’s the actionable part.

It’s connected, meaning that whenever you do this, it raises your relative status among your peers [laughs] so that people would feel proud to endorse a candidate, to ask a candidate relevant question or to make a summary of a candidate’s position and so on. That makes its social so that people see on their social media profiles all the time there’s an independent candidate.

For example, there’s a so-called obasan coalition. Obasan is a Japanese law word, literally means elder women. [laughs] There’s a loose coalition of elder women councilor candidates who are all running for the first time that maximize the use of social media through this use of the councilors voting guide.

They’re not just doing it on the Internet, but rather it crowdsources the agenda that people care about on the Internet and hold face-to-face deliberations based on the topics that people on the Internet feel as important.

That bring us to the third, which is extensible. Extensible means like the MeToo hashtag. Nobody controls that hashtag. Everybody is allowed to add to it without asking permission.

This permission less innovation also lies behind the g0v philosophy because the entire code, the entire data set, everything is under what we call Creative Commons Zero, which means no rights reserved. Everybody is allowed to take it to wildly different directions and to mix, remixes and re-remixes and so on.

We’re already seeing a lot of local campaigns that use this as a canvassing tool. That develops a much more targeted way based the collected the information on the councilors voting guide, but that’s pretty particular to Taiwan. I’m not sure whether how much that helps you, [laughs] but this is how we’re doing it here.

No, I think there are some wonderful ideas in there. How feasible they are, I don’t know just because different legislative environment and different cost base for these things. I absolutely believe that we should be exploring all of this and seeing what we can do, because it’s not as if technology is going to go away.

That’s the other thing. It’s what we see coming towards us. What would you see as the technologies that are heading our way that might make a difference as well?

In the social innovation scene here in Taiwan, we’re seeing a lot of work around mutual distributed ledgers. I try to avoid the blockchain word because it is just one of the many technologies that can...

[laughs] I was...

Yeah. [laughs] ...that can deliver a mutual distributed ledger.

Yeah, I was laughing.

[laughs] Yeah, right. The newer generation distributed ledgers, they’re not even using chains anymore. They’re using a cyclic graph and so on. Without getting too technical, I think what’s important is that it’s a ledger that people can add to but not delete, that people can audit but they cannot censor.

That is an important part of it. It’s not the ICO part of it, although I’m sure that people are interested in that as well. [laughs] For my purposes, what this means is a relatively cheap way to build accountability, to build something that people can reasonably sure that will not be changed, to be censored or modified.

We’re already seeing it in use. For example, in campaign donations and even in disaster relief donations. For example, a lot of Taiwan people donated to the Nepali flooding disaster. People want to know that their donation is being used in a conscious and accountable way across many different actors in many jurisdictions.

Previously, to buy this kind of accountability, it’s very expensive because you have to hire...I don’t know, KPMG or equivalent [laughs] accountants in all the different jurisdictions to make sure that the numbers add up. It’s pretty expert language and people cannot easily verify it by themselves.

They’re still doing that, but they’re now also using distributed ledgers to make sure that even if the crowdsourcing or crowdfundings go away, they can still reconstruct the entire accountability trail from the Ethereum public distributed ledger alone.

We’re also seeing a lot of people doing public discourse this way because they know it will not be censored by editors. They will not cave to pressure by powerful sovereign entities if they try to censor them, even attempt of censorship will be very apparent on the distributed ledger and is often a lost cause.

It’s also important to get people’s voice heard in a way that preserves integrity. I think accountability and integrity are the two not often advised, but I think it is actually a primary value that distributed Ledgers offer us today much more than its financial potentials.

One of the things that strikes me about a lot of conversation about what distributed ledger technology, which yes, people here mostly know is blockchain would create it is that people don’t always think about the different impacts on gender and the different ways this would affect us.

For example, there are some women and certainly people from minority populations, for example, who would be very worried by something that was irrevocable and which identified them and held information about them irrevocably because they have spent a lot of time actually trying not to be seen or not to be defined in a particular way.

How should we approach these different strains, if you like, in evolving technology? You can see the beauty of something that can’t be censored and can’t be changed in some cases, but then it may have other unintended consequences.

That’s right. The two use cases that I mentioned, one for the use of public charity donations, and the second for people’s public discourse. Those two are definitely in the public sphere and not at all in a private, French, or whatever family setting.

I would not advice the use of mutual distributed ledgers at this point in time for any actions that you just mentioned that has a privacy part in it. I understand that there’s many mathematicians working on privacy-preserving, so-called zero-knowledge mathematics, but they are far from mature at this point.

If people are intending to use distributed ledgers in a way that interacts with private data, sensitive data, data with limited distribution, and things like that, there are other cryptographic tools such as end-to-end messaging and publicly auditable, forward secrecy-preserving chat tools.

For example, personally, I use Wire, but Signal also has a lot of users, and things like that that are much more useful than distributed ledgers for this setting. One part of literacy or one part of awareness is to make sure that people understand the material of technology and the property that they uphold.

Code in this case is like law and not like jurisdiction law. It’s like physical law. Each different technology imbues with itself a different set of physical law that makes things easy, that makes things possible, that makes things impossible.

One of the most important thing when we did our K-12 curriculum is learning with the children how they want technology to behave, and then make technologies or allocate technologies that respond to their expectations about a social setting, and not the other way around.

Many buzzword makers would like it to be other way around, but I think that is actually detrimental to the autonomy of people with various different ideas and different social expectations.

That seems to me a very important principle for anybody who is trying to make policy around these issues, to understand that aspect. Can I ask you about, I think it’s called pol.is?

That’s right.

We were talking before about the ways in which you were able to find consensus from very strongly polarized viewpoints. Is pol.is something we as a party could be using? What would it look like if you applied it to some very polarized debates that there are within feminism, for example?

Pol.is is great if you use it as a agenda-setting tool. By that, I mean that it surfaces what people’s common values are, just by their differences. It enables people to find possible solutions that follows these common values. What it’s not so great is to work out the details of those ideas. For that, you need other tools. The great thing about pol.is and other technologies that...

I’m going to interrupt briefly. pol.is, for people who haven’t read or heard about it, it’s a platform, yes?

Right. I’ll explain it very quickly.

Pol.is is like a open questionnaire. When you go to pol.is, you’re seeing one of the few, what we call seed questions that ask simple yes or no questions about how you feel about one particular issue.

The first time we used pol.is in Taiwan government, we talked about private ride sharing in the form of UberX. For example, the first time when one goes to pol.is, one can see a yes or no question, like, "I think private passengers still need to have protection from accidents by commercial insurance providers." You can click yes or no on that particular sentiment.

As you do so, there is a face of the crowd underneath this yes or no question that shows the clusters, the people who think similarly, about things. You can see all your Facebook and Twitter friends if you sign in. If you don’t sign in, you see random famous people and how they locate within the different clusters.

Perhaps there’s people who care about innovation. There’s people who care about safety. There’s people who care about insurance. There’s people talks about taxation, and so on. They will form different clusters. As you answer yes or no questions, your avatar will move toward the cluster that most resemble your ideas. The beauty of pol.is, there are two distinguishing factors.

First, there is no reply button. You can just press yes or no, like up-vote and down-vote on other people’s sentiments, but you can never reply to them. Trolling, ad hominem attacks, and so on has no place on pol.is. If you can get 5,000 people voting exactly the same way, there is still one dot in the two-dimensional map. It doesn’t pay to troll in a pol.is environment. That’s the first thing.

The second thing is that after you answer a few yes or no questions, you can contribute your own feelings. The more resonance you have for your own feelings, the higher the score is. It still engage people in a competitive way, but people compete to win resonance, people across the aisle.

By competing to win people over to their viewpoint, it’s almost like a game.

That’s right. It is actually a game. To make it more visual, here is how it looks like. I hope you can see the screen.

Yes.

On the top, you can see one yes or no question. Below, you have an avatar moving as you answer, and you can see the cluster of people. After a few weeks, we always see something like this, where people identify the divisive things that they agree to disagree, but people spend far more time on consensus statements that they want to win over people from other aisles and other clusters.

Whenever we see a shape like this, which is always actually, we always know that people want to spend much more time to work out the details of what they feel as important as everybody. It also lets people know that although there are a few, like five, divisive questions that tears people apart, no matter which groups you are, there is like 99 percent of people, 98 percent of people who share their common values after all.

On normal social media, it flip around. People spend most of their time on arguing their differences while spending very little time arguing about their consensus.

Is that platform something that anyone can use?

Yeah. It is entirely open-source. We have a instance running here in the Taiwan national government. You don’t have to be a Taiwanese citizen. You can still use our instance. It’s easy to set up your own if you know technological people who can set up a machine.

As we’ve been speaking, I’ve been imagining the very few people that we actually have in our team with their heads in their hands, because they know I’m going to come away from this conversation going, "We should do pol.is. We should do pol.is."

That should be, as you say, the websites that look like government websites, but actually have the information in them that the government websites don’t. All those things requires a lot of work. My question to you is, how is this actually done practically? Who is doing the work here?

We actually start normally with a easier version, which is not pol.is. It’s called Slido. Slido is something that I use for all my public lectures. I just came out from a conference called Tai Chi, which is a Taiwan computer-human interface conference.

The Slido idea is similar to pol.is. You go to Slido.com. You enter a number or a code, and you get to start asking questions. Unlike pol.is, there is no clustering. There is no moving around, but you can still like each other’s questions, and you can make things that people like the best float to the top.

Similarly, there is no reply button, and similarly, the only way to get something floating to the top is not attacking the current top question or top idea, but actually proposing something that resonates with more people.

Slido is best used in a town hall setting with 200 or 2,000 people in the same room synchronously, while pol.is is best used over several weeks’ time, so that people have more time to come up with nuanced statements. We often mix the two.

For example, we will have a kickoff meeting where the people in the civil service, people who care about this, the activists, the various stakeholder groups, and so on, we first do a kickoff meeting where we talk about how to define the topic of this conversation.

For example, just the name private ride sharing while charging people for it took us like three meetings to arrive to this definition. For example, we had another important conversation on pol.is where we again took two or three pre-meetings where we talk about non-consensual distribution of intimate images.

Like everywhere, it was hotly debated, because people want each stakeholder group to feel as comfortable as possible with the definition of the name. For this kind of meeting, we don’t use pol.is.

We use regular teleconference or face-to-face discussion with deliberation facilitators and with Slido so that people can feel comfortable raising their points in those face-to-face meetings without the people who hoard the microphone taking all the time about it.

Then after a few such meetings where we then converge on the well-defined topic of what we can talk about, then each stakeholder goes back to their group, and we give them the same URL, the same web address or pol.is in the same time, so that people can share at the same minute.

It’s important because if people come to pol.is and they see like they are alone in one corner, while 90 percent of people was dominated by some other faction, they will be turned off and they will not share a pol.is conversation. It’s important that a balanced number of stakeholders get the pol.is conversation at the same time.

We find ourselves saying after, I don’t know, a month or so, any popular opinion, any resonating feeling that surface in the pol.is that’s convinced a super majority of people, meaning that in all the different clusters, it convinces the majority.

The more resonating it is, the easier for us to use this as the agenda for our next multi-stakeholder meeting, which is usually live streamed, and which the facilitator just checks the consensus from the people one by one, with the stakeholder saying the people have spoken, so what do you feel about it? Is this feasible? Is this possible? If it is, what actions are we committing to deliver on those shared values?

This is a three-part thing. The first one is the stakeholders building trust deciding on exactly what topic to talk about. The longer asynchronous pol.is stage, which usually lasts for three or four weeks until we get a set of consensus, then the same stakeholders, or even more stakeholders because people become aware of it as a result, go back to the same room, live stream or at least take a recording or transcript.

Everybody responds just to the points on the pol.is that’s everybody’s consensus. Then we bind ourselves into coordinated action. This is how often it’s done in the vTaiwan method.

Then at the end of all of that, you might have a vote on something in the way that you would expect in a democracy, presumably. You go through this process of deliberation and moving people using both direct and representative democracy?

Surprisingly, only about 20 percent of vTaiwan cases led into a law change, in which case, of course, the parliament need to do their conversation, and they need to vote on it. They mostly know that it’s already the people’s consensus. Other than one case, the other cases, the MPs just accepted the consensus pol.is.

On the other hand, the other 80 percent, which is regulatory change, policy change, or even behavior change, it doesn’t need a full conversation by the parliament. Just people committing to the actions, deliver, and work through those actions. They hold themselves accountable to it, and that’s it.

For many things like, for example, cyber bullying, what we did was not make a new law, but rather making sure that each ministry and each department deliver their responsibility in doing their part of work against cyber bullying. It doesn’t always lead to a vote. Maybe just one in five cases lead to a vote.

You said you’re using it on a consultation around the non-consensual sharing of intimate images.

Intimate images, that’s right.

What we might refer to as revenge porn, for example.

Yeah, that’s right. That was the initial name actually, but then people pointed out it’s often not porn. Calling it porn actually obscures the original image collection, which was to show intimacy and not at all for arousal, or not always for arousal. It is mostly to show intimacy. People eventually changed the name to NCII. I’m neutral on this matter, as I must. This is just how it happened in Taiwan.

NCII is very interesting also because people started talking on it as a primarily affecting women thing, but then we discovered that there’s also NCII cases around LGBT groups. There’s also NCII things in basically anywhere that has unequal potentially oppressive power structure. NCII is used as one of the power vehicles.

While women are, of course, one of these groups, they’re not the only group. This is the idea of intersectionality. People are vulnerable on various different parts. Any humiliation or any power struggle in any of those intersectional parts can reinforce each other’s power.

When we want to talk about this, saying this is wrong, and we don’t have to resort to calling it pornography, or calling it indecent image, or things like that, because people need to have control of their own intimate images, no matter which gender or which social status they are.

This is a more inclusive way of having a dialogue, because stakeholders just discovered this conversation and then we learn about new stories that we did not anticipate that they are also victims of NCII.

That’s completely fascinating. I am very aware that I’m talking to you for much longer than I intended to.

No, it’s fine.

There are so many things here that I want to know about. I mentioned before that you came to Silicon Valley ridiculously young, didn’t you? Did I read somewhere that you...?

19 is not ridiculously young. I stayed in Germany for a year when I was 11, and then one year in The Valley when I was 18 or 19. I’m mostly based in Taiwan, actually.

I may be misinformed here. I thought you had founded a startup very young.

I did, when I was 15, actually. That was my first startup, then a series of startups.

I see, but not...

It got invested eventually by Intel and was one of the dot com stories in Taiwan. At that time, I think that everything is very international, and we don’t call ourselves a Silicon Valley company. We’re still, at that time, mostly based in Asia, but it’s my first startup when I was 15.

I’ve been wondering. I’m very glad that you’re in Taiwan and doing what you’re doing, but there was part of me that was wishing that you were in America maybe finding ways to revitalize the democracy there right now.

We hold workshops in New York City. Many people in NYC and also people from 18F, from the federal government. We also talk about people from the US Digital Service and the usual suspects. I think there is still a lot of people doing useful work around civic participation open government, especially on a state and city level.

To think about it, Taiwan is just a larger city. The north-most city, Taipei, and the south, in Kaohsiung, taking high-speed rails is just an hour and a half. While we are 23 million people, admittedly, it’s on a relatively small geographic island, which is why we can say broadband is a human right and so on.

It also means that the technologies we develop are mostly scaled to this geographic scale and to maybe five million people, give or take, which in the United States is just maybe half a New York State or something like that.

I think it makes sense to start our experiments, start our coordination and workshops and so on, on this self-ruling, to a degree, cities or states in this kind of size.

Sorry. I just lost the sound for a second there.

I’m recording my side of the sound, so we can always stitch everything together.

That’s good. Just at the end, you froze for a second. I’ll finish with a question that we’re going to go on and debate at this Women’s Equality Party Conference. That’s really whether data-driven technology can actually help to resolve inequality. Do you have views on that issue?

Yes, strong views, I would say.

Good.

I think data agency, data as a relationship and not as an asset, is something that even with the GDPR, many policymakers still have not internalized this view on data. I’m not just quoting GDPR.

In Taiwan, the privacy protection act, the PIPA, says the same thing, it basically says if a organization or institution holds my data, it begins a relationship where I can always ask what’s happening with this data. I can update it.

If they want it to be used in a way that goes beyond the original collection purposes, I need to be given a chance to be informed and even update about it. The alternative is just to have a pale shadow of a data-driven simulacrum of me four years ago, a small part of my behavior that would be extrapolated, often wrongly, about my current status.

If we see data as an asset and not a relationship, we will end up with the data that are there but are basically reinforcing biases, not just from our past selves, but also from past social conditions.

This is very important to see that only a living relationship between the data so-called producers, so-called data processors and so-called data users or consumers, they need to trust each other through a accountability framework that enables constant interrogation, constant relationship between all those different people involved.

Only at that time can we get the agency of the people back to the people so-called producing the data for collection. I think if the people producing data or people collecting data are generally aware of this, then they see their contributions as in the commons.

For example, in Taiwan, we are now working with the Mozilla Foundation on a project called Common Voice, which is basically us reading aloud random fragments of public domain text.

Basically, informing the machine learning algorithm so they can recognize the different accent, the different ways people use language, even ethnic minorities, indigenous people, and things like that instead of forcing everybody to speak in a accent that most resembles whatever [laughs] the original voice actors that the machine learning companies contracted with.

This commons that resembles the way people actually speak are entirely done by voluntary contributions and with the people knowing that they can also use the voice data in this commons for whatever purpose they like. They can also update it and reflect what they want things to go to, again democratization of this technology.

If this is in the commons, if this is managed by a social enterprise or a cooperative that everybody can openly join, then this is seen as something that shares everybody’s stewardship responsibilities, but also rights.

If, on the other hand, the collection itself is opaque, if, on the other hand, the collection itself makes biased assumptions that people are not aware of, not even the data scientists themselves are aware of, then we get a situation where we get a lot of food. We feed it into a machine learning algorithm, but there’s no nutrition labels.

We don’t know that whether it has skipped over those and one thing or another. Then the data agency will be much harder to be built as a kind of ad hoc or post-tactile way because the mathematics is just not there yet.

It need to be privacy-enhancing by design, and if people choose to contribute to the commons, this is by their voluntary action, and again, the stewardship need to be designed in the beginning, not as a tacked-on law or regulation, because that doesn’t work, frankly speaking.

Thank you, that’s such an excellent answer. Great way to set the scene for the debate, and incredibly useful insights into what we might be doing. Thank you again so much. Thanks for speaking to me.

Thank you. I’ll send you my end of voice recordings, and you’ll send me the video recordings on your end, so we’ll have a very high-quality stitched version.

Yes.

OK. Thank you so much.

[laughs] That’s right. Thank you. Bye.