You make really transcripts?

Yes.

I thought it was just like an audio file, and that sometimes you make a transcript?

Well, it’s up to you. We can also do a video recording. But since you are producing a written report, usually we…Like if you do text, we do text. If you do video, we do video.

It’s the norm that we’re establishing. It’s called “radical transparency.” Let’s get started.

The idea is that if something is published…?

People have the whole context.

Yeah, yeah. They can’t say, “Hey, Minister Tang said that,” and you could still quote back to…

Always within context, yes, so people can very easily see whichever journalist, like Wall Street Journal, NPR, Deutsche Welle, Schweizer Radio und Fernsehen, and what they asked and what I answered within the context.

If you search for it, you can find this page, which is one single quote, but you can always see it within context ‑‑ how it is produced.

Is that checked a lot?

Yes, it is. It also enabled the journalistic community to have access to the internal meeting, because I do that not only with journalists and visitors, but for all the internal meeting that I chair as well.

It make investigative journalism easier than tabloids, because you have the raw material. You don’t have to get the scoop, so to speak, so you can work on your perspective.

My parents were both journalists.

I think it’s not only a lot of work but way more difficult for you, no?

No.

Because you always have to be careful not to say anything off the record?

No, if we, for example, go to a photo session, as we did before, because we’re not producing anything quotable, so we may chat there, but because it doesn’t result in anything binding…The idea is anything that might be binding, either in a lobbying sense or in a internal meeting sense, must be on the record.

There is no off the record, actually.

Right. If I say something off the record, that’s in my personal time, and it must not carry any binding power.

Sometimes to protect sources or to protect partners…

Yes. We do have people who prefer to be anonymized in this system, and that is entirely fine. We even have…

How do you note it, then?

We even had a journalist that requested that all their questions are removed, and only keeping my answers.

Why is that?

Because they considers their questions company property. That’s fine too. It’s kind of flexible. For example, there was a visit by a think tank, and many of them prefer to speak in Chatham House rule.

When they do that, all of them are marked as “Audience Member.” Actually, this is not the exact words they say. It’s what I paraphrase. I double‑check with them. I paraphrase their words, and then that become their words with that attribution.

Sometime they only agree to ask questions in a off‑the‑record setting, so I just pause the recording. [laughs] They ask the question. I continue recording, paraphrasing their question, and so on. That’s the maximum I would do.

Do you notice an effect on the…

Quality of visit? Yes.

And on the public discussion about your work?

Yeah, definitely. There’s two main changes. The first is that people, when they lobby, under this radical transparency, they tend to make arguments that are for the global goals, that are for the public good.

Even if they are arguing for a private company ‑‑ for example, Uber ‑‑ when they David Plouffe lobbied for Uber, but under this condition ‑‑ all his arguments is about climate change mitigation and so on, and so it must be a public benefit.

He would refrain from saying anything that sounds like it’s only good for the company and me, and not the public. That’s the first thing. It’s much more prosocial and pro‑public.

The second is that I don’t have the possibility of being quoted out of context. A lot of spins in politics is about building a straw man argument, a partial quote, and then attacking that.

Because context travels very easily. Really, the whole context is just one search engine search away. In that context, I’ve found that almost no editor would misrepresent my words or do a straw man construction based on part of my work.

Yeah. Of course, I hope to ask a lot of questions about disinformation. Does it also in that sense help to limit disinformation?

Of course, because it increase trust.

If you have someone who’s your friend who agrees to meet you every Wednesday ‑‑ as I do in a public space…There’s even no walls anymore. You can just walk straight into the Social Innovation Lab, and I’m there every Wednesday ‑‑ then if you hear some rumor or some disinformation, you will check with your friend.

If your friend only answers your email once every two months, and only on the part they choose to answer and ignore everything else, then of course you will tend to believe rumors about your friend. The basic trust, I think, is one of the key things to counter, to disarm disinformation.

The other thing is humor. We do that a lot too.

The other thing is humor?

Yeah, humor. Whenever there is disinformation that tries to provoke outrage, there’s a system in Taiwan ‑‑ you’ve heard about it perhaps ‑‑ that we roll out within two hours, but now usually within an hour, memes, things that go viral on the basis of humor.

For example, there was a rumor that said perming your hair would be subject to one‑million‑dollar fine. Within one hour, our prime minister wrote out this card saying, “It’s not true. I may be bald now, but I would not punish people who look like my youth.

“What we have done is introduce a labeling requirement for hair products starting in 2021. Premier as he looks now, since if you keep perming your hair many time a week, it will not damage your pocket. It will damage your hair. When serious, you may look up. You look like me.”

It’ s humor, because he makes fun of himself, not other people. This goes viral, much more viral than the original disinformation. It serves as a inoculation.

It also goes much more viral than just…

Just this fine print. Of course.

Every…

Ministry, yeah.

…disinformation about a minister or ministry has to be…

Clarified this way? Only if it has the potential of going viral. If it’s just five people sharing it publicly ‑‑ it doesn’t go viral ‑‑ then no problem about it.

Who decides?

People who flag things as disinformation. Online, on Facebook, many multinational platforms that has a built‑in…just like email, you can flag something as spam, inline, you can flag something as disinformation. So far, there’s this many flag as of today.

Today, this is disinformation?

This is total reports.

Oh my God.

Then this is unique messages. Each message has on average about four or five people flagging it. These are the ones that have received clarification, and those clarification have been sent back to the people who flag it.

Those numbers are terrible.

Large numbers.

Yeah.

Yes.

Is that increasing?

This is accumulated.

Ah, OK. I thought it’s today.

No, this is as of today. This is not just today. [laughs] If it’s today, it’s too much. [laughs]

As of today from when they started doing that.

I think July. You can see the trending ones, like this is the top 10. Currently, almost a thousand people flagged this one. This is saying that a doctor purposely said that one should not eat sashimi because there’s a new, whatever, worm, discovered. Whatever.

It’s fact‑checked by the Taiwan FactCheck Center as wrong. Actually, that particular hospital has it on its web page a clarification ‑‑ said that “Our doctor never said that.”

It still is very popular because people care about their families’ food safety. That’s the top social issue in Taiwan.

I think after 2016 elections in the US and after last year’s elections in Taiwan, there’s a lot of attention for disinformation.

Certainly.

There was also a report of the V‑Dem Institute in Sweden saying that Taiwan is number‑one when it comes to foreign dissemination.

According to the experts they have interviewed, yes.

How serious do you feel the problem is? In Taiwan, I mean.

According to the Taiwan Fact Check Center, there’s many different types. One is one that try to undermine the public trust in journalistic institutions. This is perhaps, I would say, one of the most serious kind because the journalism training is the main thing that disambiguates misinformation, as in misinformed, from intentional disinformation.

Anyone who have a media competency, received basic journalism training, when looking at misinformation can shield oneself from this disinformation effect and really check the sources and so on and turn the parts that’s misinformed into this kind of attribution report and turn the parts that are actually true into something that people can understand. That’s core journalism work.

If disinformation try to undermine trust in journalistic institutions, then that, to me, is a affront to this whole journalistic integrity system. Then it will reinforce the echo chamber effect because then it discredits the journalism in general.

If you look at the TFCC, many of their focus is on this kind. For example, the latest one, the 234, that says this photographer use a long‑focus 70 to 200 mm to take a picture of a certain parade.

In the Kaohsiung parade.

Yeah, the Kaohsiung parade. Someone tried to frame it as Photoshopped because the lights are too close to each other, ignoring that this is what that lens is designed to do. [laughs] Basically, it’s trying to discredit this whole photo, as well as the journalistic institution behind it.

The Taiwan Fact Check Center has to work in a very rapid fashion to re‑establish why does this, if you have a certain lens, you will naturally look like that. It also compared to Google Street Map and so on to ensure the accuracy of this picture, even going into the forensics field.

This is important to restore people’s understanding that there any journalistic output went through this kind of fact checking. It’s just the institutional fact checkers won’t share their work so publicly, but a journalistic standard says that something like this must be done internally when you’re reporting anything.

This is the main challenge. It’s still ongoing. There’s unending ways trying to frame the journalistic output as not trustworthy, but the TFCC is working very diligently to restore this kind of journalistic integrity.

It’s attacking integrity of news that is actually real news. It’s at the same time spreading fake news to the point that people say that the outcome of last year’s elections might be actually influenced by it, might be different than if there would not have been so much disinformation.

I would say certainly people at that time didn’t have a easy way to check whether something is disinformation, as in intentional, harmful untruth, or if it is a real journalistic output that just contains some content that they happen to disagree with, which is normal. There was no easy access to the Taiwan Fact Check Center.

Now, as of this election, not only the TFCC website is much more accessible, but the partnership with LINE, with Facebook, and so on is firmly in place. When people see a trending frame like this, they can very easily check. Once TFCC fact‑check this, actually this number will stop increase that quickly because people won’t see this on their news feed.

This is number of clicks?

Number of shares.

Shares, yeah.

Previously, when TFCC finish their fact‑checking, it’s usually in the thousands. Nowadays, they fact‑check when it’s in the hundreds or even around a hundred. It’s trending, but it’s not yet reached that many people.

Once they finish fact‑checking, it influence Facebook’s algorithm so that this will stop being preferred to share for people to see on their news feed. It’s not a take‑down. If you go to that post, it’s still there. It’s just by default ranked lower.

I think the last year’s elections were some kind of waking‑up moment for the Taiwanese.

That’s one part of it, but I would say not only election, but also referenda. Last year’s referenda, I would say, is the main surface on which disinformation spread, less so elections.

There’s a lot of people who say that current presidential candidate Han Kuo‑yu got elected as Mayor of Kaohsiung thanks to the disinformation.

I wouldn’t personally say that. I would say that there are a lot of referenda at the time that are basing their referenda campaign work on partial or plain incorrect information. People often discredits journalists when they make a clarification about the referenda topic.

Because there were just so many referenda topics, there’s bound to be one or two that really radicalizes people on that particular one or two they care about. I would say people are much more polarized when they step into the voting booth because, I would say, mainly of the referenda.

Whether that mood affects their election voting behavior, that’s a possibility, but I wouldn’t that quickly jump to this causational link.

Do you actually know what the main sources of the disinformation is?

Yeah, sometimes they are very loud about it. Like the fact check number 204 that says there is a trending rumor that says, “The rioters in Hong Kong pays students up to $20 million to murder police.” That’s a very serious accusation. Complete with pictures and all that and with purported recruitment posters and all that.

Of course, it’s very quickly traced to Zhongyang Zhengfawei [Mandarin], which is the official Weibo account of the central political and law unit of the PRC. Because of that, it’s easy to see that they have shared this first here.

Then it became packaged, their main message become packaged, complete with so‑called evidences and spread on the social network here. Then the fact checkers then look into it and saw first they use the Cantonese words incorrectly.

The Telegram group actually used Hanyu Pinyin. It’s very difficult to imagine why people protesting in Hong Kong would take the effort to have a Telegram in Hanyu Pinyin. If you do join does Telegram groups, you see mostly porn in it. [laughs]

It doesn’t quite look like [laughs] a recruitment poster. The Zhongyang Zhengfawei picture is actually taken from Reuters. The Reuters said nothing about them being paid or anything like that. Zhengfawei took this photo and added a different frame to it.

They’re actually not hiding that.

They’re not hiding. The Zhongyang Zhengfawei said, “30,000. This is what they paid these 13‑year‑old teenage protesters to murder and to cause riot.” This helped them to, they say, as reported that they got a new iPhone based out of it. If you look at the original story from Reuters, it said nothing like that. They are very loud about it.

How do you know it’s from Zhongyang Zhengfawei?

You can check their Weibo, actually.

It was first spread on there.

Yeah, it was first spread on there.

Here.

Yeah.

Wow.

It’s pretty clear‑cut. [laughs]

Experts indeed often talk about Chinese cyber warfare aimed at Taiwan. Then we imagine that this is happening quite hidden.

In a covert fashion.

In covert fashion.

I wouldn’t say so… I would say a lot of narratives is done in a very overt fashion too. It’s just repackaged and localized into something that they think maybe the local people here, after seeing this drawing, will more readily believe. The narratives actually came from the propaganda unit in a public fashion.

The repackaging and the localizing, where is that happening?

It’s just a nature of the social media. It’s like evolution. If you have a memetic variation, that cause it to spread more. You see it more. People will remix it more and so on. That’s part of the social network. The core message came sometimes very overtly here.

It’s not the only example. During the Hong Kong protest, there’s many examples. During the WHA that Taiwan could not attend even as a observer to the World Health Assembly, there was a similar narrative that said that Taiwan is well‑represented because the Beijing delegate will take care, in a efficient manner, anything that the Taiwan people need to know.

They are a good representative to the Taiwan delegation and so on. Of course, that’s not true, but that’s their official line. That become then repackaged and shared on social media and saying that Taiwan people don’t need to worry. If there’s any health hazard, people in Beijing will take care of it very efficiently and so on.

It’s actually happening in a more organic way than it’s often portrayed, as in Chinese content farms covertly making up things and then sending them to Taiwanese Facebook groups that they maybe have bought and then spreading it in Taiwan. That seems like a very intended strategy and very covertly, but you’re telling it’s more…

That may used to be the case, but most of the large platforms, Twitter, Facebook, Google, and LINE, took countermeasures. Now it’s much more difficult to create fake accounts and fake profiles and share content farms without being caught by the algorithm, so they largely have to shift tactics.

Nowadays, if you want to create tons of fake accounts, you may have to resort to machine learning to synthesize the photos. Even if you do that, Facebook recently just announced that they uncovered a set of fake accounts using machine learning to synthesize the profile.

They’re playing catching up but playing it in a very efficient manner. Now, on the most popular social networks, most of the old tactics of about two years or three years ago no longer work.

The fake news is still spreading, but in another way?

They’re basically forced to be much more overt.

And count more on the whole…

Narrative.

…public.

That’s right. Still they have to undermine first people’s media competency. That is to say the capability to perform journalistic work by the people themselves, as well as relying sometimes on the institutional journalism to check the work. These narratives rely on people distrusting journalistic outputs and themselves not willing to perform journalistic work.

You’ve been talking about the fact checkers. I know there’s also a law that will be presented to the Legislative Yuan for the third time on 31st of December.

It’s raised by the legislators. It’s not by the cabinet. We didn’t discuss it in the cabinet meeting.

Do you think that law will be helpful?

I wasn’t really involved with any discussion on that particular act. It’s entirely a parliamentary work. If you want to look at something resembling the cabinet position, the mac.gov.tw, the top press release not too long ago is their position on the anti‑infiltration act.

Of course, they have done some work on this act. If it passes, they will help enforcing it, I’m sure. They do have a position. I do not supervise the MAC, so I don’t have a position.

It seems like fact‑checking is, of course, very useful, but it’s already after the fact. Let’s say after the fake news has been created.

Of course. Otherwise, it’s called censorship.

It would be interesting to find out who has created the fake news.

The attribution work.

Exactly.

That’s what the TFCC is doing. They trace the messages down to its raw form. Like before it got remixed, the raw form was posted here or on Zhongyang Zhengfawei or somewhere.

The problem is if that is happening often in mainland China, there’s no way to touch the source.

As in, you can’t easily interview Zhongyang Zhengfawei?

Yeah.

[laughs]

You can’t charge them here.

No, but sometimes the attribution by itself is good enough. If people understand this is from the state propaganda account, literally ‑‑ with all due respect, that is a state propaganda account ‑‑ then people would understand how it’s not something organic.

It is not something that people who participate in Hong Kong protest reported. It is something that’s created by the PRC. Sometimes that’s enough.

Is that often the case? Because you give this example now, is it…

That’s sometimes the case. I would say in the really serious ones, the ones that would target at undermining the journalistic integrity, you can often trace to exact point because that’s very intentional.

In many cases, of course, it’s not quite intentional. It’s just a rumor that mutates. You can’t quite trace the source, but then that tend to be less political in nature. Like things pertaining to health, like what fruit if microwaved would cause what disease, or whatever. It’s hard to trace the source because it’s like urban myth.

That, actually, if you look at TFCC, that’s the majority. For things that really attack the journalistic integrity, more often than not they found the source of it because it takes something intentional.

More often than not, that source is found on Chinese side, or is it more often domestic?

It depends. That Hong Kong one is from the PRC. For example, this one, when they trace the accounts, most of them are domestic.

This one, again, is a attack on the media. It says there is a video of a protest in the front of presidential building, but all the institutional media “turned a blind eye on it.” Trying to undermine integrity. Actually, that protest was in 2016, quite a long time ago. [laughs]

TFCC traces sources, but most of them are domestic. This video itself is from one of the main campaigning coalitions against marriage equality back in the referenda. The source was from there, but there’s no attribution to the PRC. It’s likely that they supported this protest because of referenda.

That protest was initially, indeed, saying that the new curriculum, because it include marriage equality text, need to be blocked. That was the original context. Whether this re‑purpose is done with the blessing of the original coalition against marriage equality or not, that’s debatable.

At least, the original source of this frame and this video, neither come from PRC, but it is a intentional attack on institutional media.

That’s actually what I’m a bit wondering about. Often when you read media reports on disinformation in Taiwan, it’s often presented as a phenomenon that mainly has its source in China. When I talked with some experts here, I was surprised to hear that they feel like it’s happening on both sides, both government and opposition, and maybe in more or less equal amounts.

I was surprised to hear that they feel like it’s happening on both sides, both government and opposition, and maybe in more or less equal amounts.

I would say it’s not in the best interest of government ministers to undermine trust in institutional media or in democracy itself. Information that cause public harm in terms of the long‑term trustworthiness of institutional democracy or institutional media, that’s certainly not in the best interest of the cabinet.

For many politicians who are more senior or they have a longer involvement in the Taiwan’s democratization, they would not like to hurt that either, whether they’re in opposition or whether they’re in the ruling party. Of course, it aligns with the PRC’s interest if people here don’t trust democracy nor the media.

There are also people in Taiwan who are much more, I would say, authoritarian‑leaning. They would also like people to trust less the institutional democracy and media. There are certainly both sides quite prevalent also. They sometimes reinforce each other. I wouldn’t say that one certainly cause another.

You’ve mentioned the fact‑checkers. You’ve mentioned your own…

Clarification attempts.

Yeah, exactly, and your emphasis on transparency as a way…

Yeah, open government and all that.

Indeed, right now, in the election campaign, we do see less obvious fake news or big fake news that has gone viral. Could you say that…

It’s having some effect, certainly, and thanks in no small part to the crowd. Like this one is a crowdsourced attempt to fact‑check each and every word publicly uttered by the three presidential candidates

During their platform speeches, during their interviews, during their public conversations, these are the almost 1,000, maybe more than 1,000 now, people who type these videos into transcript, check the transcripts, separate the transcripts into individual statements, and flag the ones that they feel are wrong.

Then the partnering media would then send their fact checkers after those statements. They partner also with like public TV and so on for distribution. Many a time, right after a presidential debate, you can very quickly in real‑time see one of the partnering media said that, “They made these claims. These are wrong. These are right.”

This, of course, has a effect saying that if they would like to make public statements, they will be more careful because they know that they will be fact‑checked, sometimes in real‑time.

What is the number that we see?

Things that are fact‑checked as wrong.

There’s also the projects that promote digital democracy, or there’s vTaiwan.

Yeah, for the involvement. vTaiwan is one, but there’s also Join, which is a public platform for all ministries. Not just all ministries, but literally every single ministry, all 32 of them. Actually, 30 here.

All 30 ministries, they share their public budget and the KPIs and whatever they have done in the past quarter. People can very easily ask questions and receive a prompt answer, at least quarterly but sometimes faster, from the career public service.

Once you have this, it’s actually very difficult to make counterfactual statement because everybody is just one question away from receiving what’s happening, from the career public service which is neutral. Politically, I mean. Not affiliated to parties. This makes all the presidential candidates, and indeed all the campaigners, think twice if they want to say something strictly counterfactual.

On the one hand, it’s very impressive, all these efforts. It’s clear that it helps verify a lot of information. On the other hand, it was only yesterday or the day before that that picture that you showed before of the parade in Kaohsiung, it was spread by Mr. Han himself, or he made a statement on it.

I wasn’t aware of that.

[laughs]

You mean this one.

Yeah.

I wasn’t aware that he made any comment.

It just seems like it’s maybe not making a huge impression on him.

At least we know that it influence the algorithm so that this spreads less vigorously, less than one‑fifth of its original intended population if TFCC had not stepped in. We know that quantitatively.

Of course, less than one‑fifth, then there’s still some virality to this message, but then the clarifications hopefully spreads faster than that. That’s it. We’re a liberal democracy. We can’t make something disappear even though it’s false.

With all these projects and platforms to counter disinformation, do you actually think that Taiwan could be an example for other countries?

Especially if you want to keep liberal democracy. There’s plenty of jurisdictions nearby, according to the CIVICUS Monitor, that took this as a excuse to have the ministers words somehow even more powerful than a journalist’s words to infringe on journalistic freedom and on the civic space in general.

It’s a good excuse. Like because disinformation is too much, so we have to authorize certain administrative powers outside of the judicial. The judicial is too slow. That’s usually the argument.

Taiwan managed to do all this without having a journalist’s words somehow have a less impact than a minister’s words. We’re still saying that each journalist is as good as a minister’s words, that we need to be on par in our dialogue. According to CIVICUS Monitor, in the entire Indo‑Pacific, all the way to Africa, only Taiwan and New Zealand are keeping this line.

Could you maybe rephrase that? I’m not entirely sure what you meant by that.

Sure. The CIVICUS Monitor is a tracking condition website for the entire world according to how much the freedom to assemble, freedom to speech, of expression, of the press, and how much they’ve been declining in the past few years.

Disinformation is one excuse. It’s not the only excuse. One excuse or another, the civic space is shrinking across the world, according to them. By now, in our part of the world, only Taiwan. Well…

[laughs]

Only Taiwan and New Zealand are ranked as open, as in we didn’t infringe on any of the major freedoms of the people. Of course, this part is faring much better than we do. [laughs] I would say we would serve as a example to our nearby jurisdictions, that they don’t have to go all the way to closed, that they can still learn from Taiwan. They do.

Like, for example?

For example, there was a Cofacts project, right here. That was the original prototype by the civil society for anyone who look at something on LINE system, which is encrypted, and forward that to a bot.

That bot builds a dashboard of the most popular trending rumors and the most popular trending responses. That could be crowd‑sourced. That’s a prototype to the production‑level LINE dashboard. They still list Cofact as one of the four partners, the TFCC being another one. This basically inspired LINE to develop something like this. It’s like a civil society prototype.

If you take Cofacts and remove the S and go to cofact.org, you see the Thai version. This is entirely open source. Any civil society can very easily download and customize it for your local population.

So there’s hope. The elections, the democracy is not entirely threatened. There are ways to…

To establish a new norm that makes people still vigilant but not too depressed about democracy.

There’s elections on 11th of January. What if the result is that the KMT can govern? Do you have partners within that party? Do they also favor…

I first worked with the KMT cabinet back in late 2014. At that time, the Vice Premier is the current KMT Vice President candidate, Simon Chang. We work on the Join platform together, too, when he was the Premier, during the hand‑off.

I’m pretty sure that many of Han’s ideas…For example, he wants to change the petition platform’s rules. Instead of 5,000 people as the threshold for ministerial answer, he want to lower it to 3,000. That obviously came from the work that we did with Simon back in 2015.

I said publicly that actually we helped the Tainan City run a local‑level petition system. The participation officer network, we also helped them to establish. If they want to run this in Kaohsiung, like today, we’re also happy to help. Open government is one of the directions that all the candidates only reinforces. They can’t take anything back because it’s a new social norm.

No matter who wins the elections…

They will continue the open government work.

Very interesting to see. You would like to remain in politics and continue this work, or is it…

I’m always doing the same work, anyway.

Doesn’t matter in which capacity.

Well I was a reverse mentor to Jaclyn Tsai, a member of the Ma Ying‑jeou cabinet. Now, of course, I was promoted from intern to a full minister, I guess. [laughs] It doesn’t really matter to me because the work I do is mostly for between elections, like everyday participatory democracy through petitions, participatory budget, all sorts of electronic participation, presidential hackathon.

All these are things that are done between elections. I work mostly on the participative part of democracy. That part can be contributed across sectors. I’m ostensibly working with the public sector, but I’m not working for the public sector. I’m still a bridge between the public sector and the social and economic sectors.

Thanks a lot.

We’re good?

Yeah, we’re good. [laughs]