2019-08-03 Interview with Facebook Civic Integrity

Let’s get started.

Max, want to kick this off?

Sure. We are here to understand what we have heard a lot, about Taiwan issues, the very interesting challenges that our platform have enabled.

Abbas and team, Thomas and the other team, they were here last week, since last week, last Monday. We have conducted the elite interviews, like 15 people.

Focus group as well as individually?

Yes, also focus groups.

That’s great.

You’re the last interview, the last stop.

OK.

We really want to hear from you about how you feel about the challenging threats that Facebook enables here, coming from a user experience perspective. Abbas has a lot of questions for you. Actually, we are trying to learn from you.

There’s a couple of things before we start. The first is that it’s helpful if you can be as specific as possible -- if you give us specific cases or experiences with regards to social media in the civic sphere. Second of all is, do you have any questions for us?

Would you like to briefly introduce yourself and the role that you play in the civic integrity team?

Of course. My name is Abbas and I’m a UX researcher on the civic integrity team. Our role is to go out, conduct research of various kinds in order to inform product-to-partner policy and to partner with operations teams as well.

That sounds great.

I’m Tom. I’m also a UX researcher on the civic integrity team. We have very similar roles. I have a slight more quantitative bent, but otherwise, same, same.

Cool. The first hires of our office are interaction designers and service designers. Right now, we have 21 interns, all working in UX and are looking systemically to improve the government services as well.

I’ve been kind of indoctrinated in design thinking...

Double diamond, user journey, all that.

Diverge/converge.

Yes, exactly. So for your questions, I’m sure that I will be as specific as possible, as I understand it’s part of the ethnographic... hanging out.

Ethnographic hanging out. Exactly.

My first question for you is about what you see as the role of social media in the civic sphere as the 2020 elections approach. How would you characterize the role of social media?

Social media in general, or Facebook in particular?

For this first one, we’re interested in general.

One of the largest social media in Taiwan is the PTT, which is like Reddit, except it’s an interface inspired by BBS from dial-up era. The greatest thing about PTT is that it’s entirely open-source, with its own governance mechanism.

Before Facebook, LINE, and Googled signed the counter-disinformation best practice, many approaches were already prototyped by PTT developers.

Because it’s open source, the governance is in the social sector, and people can trust that. They will not be bought by Tencent [laughs] because they’re in academia, hosted by the National Taiwan University.

When I think of social media, also because I was involved in BBS since ’93, it kind of is a prototype of my understanding of social media’s role.

The PTT is interesting because, first, it is a public forum. Most of the public boards are written in the Discourse style. Instead of multi-level threaded conversation that distracts people’s attention, it has a very simple up-vote, down-vote, with comment idea that can very quickly converge on similar values despite different positions.

I would say that it’s designed with convergence in mind. Although Facebook did make some improvements, for example, changing the comment box into rounded corners -- I think that helps [laughs] -- by and large, it is about keeping people’s time on Facebook instead of about converging into similar values.

The more converging a social media is, the more contribution it will have on the upcoming election. It allows people, although supporting different parties or different platforms, to reassure that we’re of the same polity after all. Instead of being shepherded into filter bubbles of echo chambers, it can serve as a true public forum without people silencing each other’s voice or things like that.

I would say that a lot of PTT’s design criteria is because it’s a public forum and because it doesn’t have a purpose other than a social purpose. It’s a single-purpose, not-for-profit social media.

You mention this concern, if I’m correct in what I heard, about people’s voices being silenced. I’m curious to hear from you about what your concerns are or where the source of the silencing might be coming from, in your view, on social media.

From its interface affordance point of view. In Twitter, it’s now very easy to mute people, almost too easy. On Facebook, likewise, actually. Although I think you show something like you can distance from that user instead of blocking them, still the block affordance is very easy to reach. Indeed, very few people actually retract from blocking.

Over and again, we see discussions that ends in unfriending and/or blocking as a result of a conversation. By definition, people with the most time wins a conversation by default. Because the system doesn’t have this idea of a rough consensus, or like Discourse or other specific forum ideas, it’s always easy to do necromancy on a discussion thread, that is to say, revive a discussion that is long-concluded.

All this makes it easier for people to identify other people with a negative projection, see people as the distractors from civil discourse, and choose to then flag, distance, or block other people. All those affordances I’m sure are great for mental health, but then it automatically make a polity more fragmented as part of interface design.

You’ve talked a bit about affordances and some issues you perceive with those. If we think about actors and the role they might play in silencing voices or marginalizing voices, what do you think the role of that is with regards to individuals or groups in the political discourse online in Taiwan that might be contributing to this problem that you’re perceiving as well?

I don’t think it’s a Taiwan-specific issue though. In Taiwan, maybe Facebook penetration is very high and people tend to confuse friends-and-family conversation with political conversation. In many other jurisdictions, we see people using Facebook for one and Twitter for the other, or some other thing. In Taiwan, it’s all mixed.

Part of our culture is that you can call a random person on the street aunt or uncle, and people won’t [laughs] find you strange. It’s just part of the large-family thing. I guess it’s appropriate that people just randomly treat Facebook "friends" as "families."

That then makes the conversations very easy to leak out of the -- to reuse a Google word that’s no longer relevant now -- circles of the original designated audience. For example, somebody may post something on a friend-only or a private group.

The private group, the private is a lock icon. It means that a access-controlled group. That same icon is also used for secure end-to-end communication in both browsers and your end-to-end encrypted chat. It creates confusion of people looking at a lock icon and assuming that this is a protected conversation.

Actually, not only it could be data mined, but it’s very easy for people to forward a private group conversation. You can always screenshot, but it’s very easy to make it leak to the public. People’s expectation of a end-to-end conversation is violated regularly.

I actually provided a suggestion that maybe you reserve the lock icon into true end-to-end encryption so that Facebook can securely say, "We don’t even know of the conversation details." Currently, it is hidden behind a hostile design that people have to explicitly choose end-to-end encryption instead of, like on WhatsApp, where it is the default.

There’s multiple interface pathways that leads people to amplify the supposed to be friend-and-family conversation into the public sphere. The public sphere then does a critical examination on it, but it’s almost always out of context because people don’t have access of the original group where the conversation happens.

This kind of out-of-context leads then to tribalism and to people fragmenting their audiences into different worldviews. I’m not saying that it’s all about interface design. I’m just saying that because you can change interface design [laughs] that are the pertinent parts that I observe here.

I found your last statement very interesting about information out of context contributing to a problem like tribalism.

To polarization.

Let’s talk a bit more about this out-of-context problem. How do you see it manifest on our platform specifically? What kinds of consequences do you think out-of-context content is having, if any, on the political discourse here?

A lot of political discourse here is centered on screenshots. It could be a private group on Facebook. It could also be a conversation on a closed LINE group, which is a similar setting socially speaking. There was a large thing that a few months ago ran about a airline worker in the private chat group talking about, as part of the strike, how they want to "punish" the captains that did not support the strike.

It makes total sense in their private context, but if taken out of context, it could be interpreted as that she’s about to poison the captain. I’m sure that she didn’t mean to poison the captain. [laughs] She just said this is someone that we would add stuff or things like that. Within context, it’s a in-joke. Out of context, it sounds like a threat to public safety.

I don’t have much specific incidents, but if you look at the rumors database, Cofacts and so on, you can easily find many such instances. I would say that a lot of political discourse is, indeed, centering around out-of-context.

Previously, it’s about politicians’ speech taken out of context in popular media. Now we have seen that, even in private groups, people are very keen to take screenshots or relay it out of context and then do a public tribunal on it.

That leads me to a related question about looking at misinformation. What capacity do you think the average Internet user in Taiwan has to perceive misinformation?

Well, I don’t think there’s an average Internet user. It’s like there’s no average citizen.

I can rephrase the question then. What capacity or what capabilities do you think are available? Which ones should be available to help understand and reckon with misinformation?

From the state side, what we are providing is a real-time clarification. Within one hour, we produce, always with at least two pictures, less than 20 words as title, less than 200 words -- by words, I mean Chinese characters -- as content. This means that it fits on one smartphone screen and, therefore, would not be taken out of context, even if you do a screenshot.

That is our primary way of responding. We invested a lot of people, and I’m grateful that Facebook also provided training, especially on our diplomatic foreign service personnel, about how to frame their messages better. There’s an outdated example that’s not new anymore. It’s a month or two old.

This is a typical clarification provided by the state, this time by our premier. When we detect there’s a rumor that says perming your hair multiple times will be subject to a NT$1 million fine, which is false, it not only says that it is false, but with a young, premier’s photo -- two pictures, you see -- saying, "I may be bald now, but I would not punish people with hair."

The fine print says, "What we have introduced is a labeling requirement for hair-product makers in 2021." I didn’t translate that, but it says a now-bald premier saying, "However, if you perm your hair multiple times in a week, it will damage your hair, and you may end up looking like me."

It’s good humor. He makes fun of himself, not other people. Quantifiably, we know that this reach more people than the original rumor because it’s funny. It went viral. People who read this message will not share the original rumor anymore. It’s a good inoculation in terms of public health.

A good thing about this, it’s because the state is not at privilege here. Any other party, any other community, can also respond within one hour and get the same kind of legitimacy. We’re grateful that the fact-checkers in the civil society -- there’s maybe four or five of them -- once they receive clarification from the ministries, they give us fair coverage in their fact-checking network.

Again, it’s not because we’re the state. It’s just we respond quickly enough. I think this affordance of, for example, LINE now offering a very easy flag as spam, it’s not absorbed internally, but rather dispersed to, I think, four fact-checkers in Taiwan that everybody can independently do fact-checking.

It’s very useful because, otherwise, people are always going to say, "Why do you choose this fact-checker over the other?" I’m sure that you face this all over Asia, [laughs] all the jurisdictions, fact-checkers that are more sympathetic to the administration, fact-checker that are not sympathetic at all with the administration.

Instead of having to choose, LINE just said, "Anyone who sign up as fact-checker gets this distribution of those trending rumors and do whatever you want about it." While the IFC and design of the Facebook-accredited fact-check network is useful, I think this more crowd-sourced fact-checking scene is also useful.

You get strength in numbers. You also get the civil society that specialize in, for example, healthcare or medicine can just fact-check their narrow domain without having to care about everything else that is unrelated to medicine. I’m not saying that you should lower your fact-checking standard. I’m saying this two-tier collaboration is actually very useful.

Related to that, I was curious to hear about your views regarding media literacy and what types of mechanisms you think are most effective to achieve increasing media literacy.

I think a popular TV series is very useful. It’s called 我們與惡的距離, "The World Between Us." It’s now airing on not only HBO Asia, I think all over the world now. It’s funded by a public budget as part of our infrastructure plan. It’s done by the public TV here. It has 95 percent on IMDB, so really good TV series.

It teaches media literacy. After people view it, it talks about how media framed things. I think they use both YouTube and Facebook in the opening of each episode, showing how the social media justice can actually hamper the real dialog of investigative journalism and things like that.

Everybody, after watching this film, gains more insight into media literacy than listening to 20 hours of lectures. This kind of public TV is just the beginning. We’re aiming to produce similar public discourse TV series for all the controversial social issues.

This is something that the state can do and, indeed, can sponsor. I’m sure that the social media people can also work with other people. The public TV is actually part of a network that does public television work. For example, the CTS is now airing the most popular clarifications from the IFCN-accredited Taiwan fact-checking center.

Each clarification actually becomes content, and you can make it viral by introducing the funny part of it and things like that, like framing it into a detective film, a whodunit episode, or things like that. Just making it fun instead of a pure, "You say that. I say that," thing, which, according to research that is now common sense, only reinforces people’s stereotypes.

It could be packaged in a funny way so that people participate in the investigation and therefore learns media literacy. Working with community colleges...

In K to 12, starting this August, all the schools can now design their own part of curriculum as must opt-ins and opt-outs. We have schools now specializing on philosophy, e-sport, which is not part of curriculum, and [laughs] all things like that, and teaching media literacy -- how to be a responsible, live-streaming YouTuber -- because people are interested in e-sport, and they have to learn that as part of their curriculum.

I would encourage you to look into that and maybe work with partners in the K to 12 curriculum. Every school is now under tremendous pressure to differentiate their curriculum versus other schools’ curriculum. Media literacy may just be the ticket.

You mentioned talking about social issues. One thing that we’re curious to hear from you about is what you think about people facing harassment for sharing their views on social issues and what you think should be done about that in order to mitigate that harassment.

There’s harassment meaning a personal attack, and there’s arguments, which is like, "I just don’t agree with your opinion." Having a more clear-cut designation of the nature of the debate really helps. In the system that I participate in designing, all the public-binding discussion forums are without reply button.

I think reply button is the affordance that makes people confuse personal versus opinion arguments. If I am replying to your argument, I’m both replying to your personally and also your statement, verbally. For all the system that I use for public discourse -- for example, there’s a open-source system called pol.is -- you can also see statements from your fellow citizen.

You can also comment, but you can only upvote or downvote, that is to say, agree or disagree with their statements. Once you agree or disagree, it just flashes another statement from a fellow citizen. There’s a principal component analysis that outlines the most divisive and the most cohere parts of the argument.

People can see that, despite the different positions, it’s still your friends and family on the other side. Always, we end with this picture in our final report. As you can see -- this is from Bowling Green, USA, in Kentucky -- only five things divide people.

If you look at the local media or social media, you think those five things are all it is. People spend 95 percent of calories just talking about these five things. People forget that most of the people agree with most of their neighbors on most of the things most of the time.

In Bowling Green, Kentucky, everybody agrees that the K to 12 need to include arts. There’s too few arts. In the STEM, it should be STEAM. People think that there should be more choice in broadband providers, things like that. Just this picture itself is very convincing.

When understanding that we’re a polity after all, people can focus more of their energy on how to implement this on the second diamond instead of wasting their time on divergence on the first diamond. People can say, "How might we do something together?" where this "we" is a polity instead of just their tribe.

Most of the design, for example, the e-petition site that we borrowed from Reykjavík, from Iceland, has a pro and con in two colors. For each e-petition, once it gets 5,000 signature, we have to respond, but people can always, instead of joining the petition, post their counter-arguments on the right-hand side or their supportive arguments on the left-hand side.

They can upvote or downvote, but they cannot reply across the aisle. There’s no way to do personal attack so people don’t do personal attack. Because you can upvote as well as downvote, but we rank it by the number of votes that each one receives in absolute numbers, if you downvote the most, it’s usually mobilizing from the other column. That makes that argument flow to the top actually.

It’s these details of interface design. We don’t show the proportion to comment bar on the pro versus contra. We also show them as exactly the 50 percent part, so that people would not mobilize with trolling. All this is just skills at troll control, making sure that if you mobilize 5,000 people, that does exactly the same thing. Its effect is nothing.

In pol.is, if you get 5,000 people voting exactly the same way, it’s only to be one dot. In the principal component analysis, it has no effect at all. Whereas, in a reply-happy environment, if you get 5,000 people individually replying, then you just marginalize everybody else.

I am not saying that Facebook should take away the reply button. It’s maybe too drastic, [laughs] but it may help to make it easier to do single level. It may be easier, in the default filtering view, to not highlight that much the second-level replies.

It may be helpful for people who do the replies in a systemic way, reply to the statement instead of to the person by, for example, only showing notification when it tags a person specifically instead of just by me replying to you and somebody else reply also to you. That really shouldn’t trigger a notification on my part, and things like that.

I’m sure that there are parts of attention management that you can adjust. My experience is that if it’s a reply-free situation, then people learn to respond to the statements, not to people.

Definitely have more to ask, but I want to be mindful of time and allow Tom to follow up on questions he has.

When you do see harassment on the platform, is it around particular divisive issues? You shared that wonderful graph of the divisive issues that soak up a lot of our time. What are the divisive issues that dominate here?

That depends on the topic of discussion. When we had a public conversation with the AIT, that’s the US de facto embassy, asking how to make more visible in the world, the divisive issue that would trigger harassment if it’s a reply-enabled situation is the statement saying, "Every time PRC closes a international door for Taiwan, the US has the obligation to open one door for Taiwan someplace else."

As you can see, [laughs] half of the people absolutely love this statement. Half of the people really doesn’t like this statement. How to define our relationship with PRC and whether this is part of a US triangle or whether it’s a bilateral thing is the main political divide in Taiwanese communities.

If we had allowed reply, easily hundreds of harassments would very quickly show up. Again, because we don’t have replies, you can see it’s the only thing that divides people. There’s nothing else. [laughs] People generally agree on most of the things.

We just got the AIT people, the MOFA people to work on the parts that pass what we call group informed consensus. Both the grouping and well as the group informed consensus are very useful here to quantitatively show people that people have more in common than we thought about.

If people get into the rabbit hole of just expanding that one divisive issue, then they have this false perception that these are people who are anonymous enemies that are even not part of the polity. That leads to both fragmentation by unfriending and blocking, as well as personal harassment which are not really personal. It’s just reacting to that one controversial statement.

You mentioned false perceptions. Are there any other major false perceptions you think dominate the Taiwanese polity other than the fact that there’s more agreement than is apparent?

This kind of false tribalism worldwide, not just Taiwan, is the main false perception. I wouldn’t say it’s caused by social media because we see that in radio and television era as well. It’s certainly amplified by social media. Previously, people may think in a binary us-versus-them. Now, with social media, there’s 500 us-versus-them axes. We can choose any one. [laughs]

I think this kind of micro-profiling, plus precision-targeted advertisement does make people feel more lonely than before. People would feel that they’re in the minority, that most people don’t think the way that we think. The majority of people think they’re in a minority. [laughs]

Then people are keen to post more polarized because they feel that they’re disempowered. This kind of disempowerment is also a feeling, paradoxically, caused by social media. It’s so fragmented. Every tribe is just very minor among the entire polity.

I wanted to ask a question about the issue of fake accounts or fake amplification. How do you evaluate these two issues? What is your perspective on addressing these issues?

The second one is fake...?

Amplification.

Amplification, so people robotically does the same sharing or things like that.

Sure or reacting.

Now that we have GPT-2 and transformer technology, what used to require a lot of crafts work is now entirely automatable. It’s like the old CAPTCHAs, I guess. It used to be a Mechanical Turk to defeat CAPTCHAs. Now it’s harder to design CAPTCHAs that doesn’t prefer AIs. We’re in a different era.

Whatever we learned in a previous era where CAPTCHAs still work doesn’t apply now to this current, where machines easily defeat people in most cognition tasks that are simplified. A few things, for the e-participation platform, we insist on having two factors. That is SMS to a real SIM card plus a social media or email login.

It makes it very hard for people to get 5,000 SIM cards. I’m sure it will trigger income money laundering or something like that. We are relatively sure that people who are on our e-participation platform are real people. Then we allow pseudonyms because many of them are advocating for things that could be seen as whistleblowing a little bit and not popular.

Without people revealing whether they work at some public sector or things like that, it’s only when their idea finally become policy and they want to meet eye-to-eye. Then we discover their identity. It’s entirely optional. We have both pseudonyms but also SMS-based authentication. That’s a good compromise for non-legally binding things like e-petitions.

For referendums, of course, we insist on a PKI card with proper e-signature capabilities, which at the moment maybe one in four people have one. The ones with NFC capabilities, maybe less than one in seven people have the newer PKI cards.

Our main goal starting next year is to get the roll-out of such NFC enabled PKI cards to as much people as possible so that people can even join referendum signatures by using their phone and do a NFC authentication and entering passphrase on their mobile PKI card.

That is a infrastructure that is responsibility of the state to provide for legally-binding referendums. That’s the identity thing. For the amplification, it’s very simple. If you vote but not review the result of voting or if you use visualization but, as showed in Polis, it measures the diversity of statements not the number of people.

You can see that even with, not even 200 people, this faction has more area than this one. If you have 5,000 people all voting the same way, the numbers may change but area will not. It doesn’t have any effect on the perceived sound or amplification. I guess it’s part of interface design.

In the Presidential Hackathon, we also used voting. You have to join through a e-participation to vote. First of all, we used quadratic voting, which is a new system designed by Glen Weyl and friends at RadicalxChange. The QV says you’d look at 100 projects. You have 99 points each and you can vote 99 votes to 99 projects or you can vote into one single project.

Two votes is going to cost 4 points and four votes, 16 points. Even if you vote everything into one case, that’s going to be only 9 votes and you still have 18 left. You can find something else to vote 4 votes that’s going to cost 16 and you still have two points and you can find some other two.

Many people discovered that these two have synergy. Maybe nine votes is too much. I’ll check something and vote accordingly. Mathematically, this means that each new vote, the marginal cost is the same as its marginal utility.

People will vote according to their true preference. While they’re voting, we’re not revealing in a scoreboard of each project. We get a variable balance outcome out of these kind of voting by, first, not revealing during the voting period and, second, using a quadratic voting to make sure that people don’t vote blindly but rather consider the synergies of their votes.

All of this can be boiled down to mechanism design. If you do the mechanism design well enough, then you can prevent much of the fake amplification issue.

When we think about fake amplification or fake accounts, when we think about harassment, when you think about this whole set of activities insofar as these set of activities might be used to affect how the public sees the credibility of government, do you feel that that is currently taking place from domestic actors, from external actors? What do you think should be done about that?

Internet knows no borders. What we have done during the election is revealing where the funding came from. The real amplification method on Facebook is cash. [laughs] It’s not fake accounts. Fake accounts gets detected in no time. If you have sufficient amount of dollars you can precisely target whomever. That’s your business model. [laughs]

What we have asked is essentially asking Facebook to treat these as campaign donations during elections. I’m very happy that I’ve heard positive commitments from Facebook so that we can reveal. In our campaign donation law, it’s extremely transparent.

You can get a spreadsheet of individual records of donation. Every media can independently analyze these. If any political campaign donation came from a foreign source that is to say someone with no voting right, it is subject to a very large fine. Only people who vote can do campaign donation.

What we’re asking during election is that Facebook review both the precision targeting criteria, as well as who is ultimately the provider of the funding, and that we can use the same laws as anti-money-laundering to then ask those operators who their funding came from. Then recursively until we know who ultimately is paying for that.

If they are all citizens, then it’s reveal campaign donation. If they are non-citizens, then we find the closest hop, I don’t know, NT$50 million or something like that because they are meddling in a election.

Honest advertisement is consensus across jurisdictions. I’m very happy that Facebook is willing to try in the next election for this kind of transparency. In election, it will help a lot.

You’d mentioned that divisiveness can often be amplified where there’s more consensus in reality. Are there particular kinds of rhetoric or language that concerns you that you see increasing or on the rise with Taiwanese users online that contributes to this perception of polarization or being more divided than they are in fact?

That’s a great question. A lot of it are still along party lines. Especially, the next election is a single winner in each district or presidential. We’re not going to elect two presidents, right? [laughs] It’s a very zero some thing. Usually, in this kind of elections, most of the rhetoric will be along party lines.

People who are affiliated with one party will be motivated, incentivized to paint people from the other party as essentially outside of the polity. This is particularly easy in Taiwan because of that single divisive issue with PRC.

That’s gradually changing. If anything, it’s not on the rise. After Hong Kong, people are generally not saying that anyone likes one country, two systems anymore. It used to be a real side that people can take. I don’t think anyone is taking that side now.

Even the leading KMT, Mayor Han Kuo-yu, his main campaign policy thinker, ex-Premier Simon Chang, is now publicly saying the ’92 consensus is out of date and we shouldn’t hold to it.

That rhetoric is on the decline. It’s not on the rise. I don’t see any divisiveness on the rise at the moment. If you asked me a few months ago, I would say it could tie into the referendum. Because we witnessed that if we tie referendum into the voting day, then everybody votes in the referendum with a very binary thinking, which isnot good for the deliberation of referendum agenda.

Now we’re on alternating years. It’s an election year, then a referendum year, election year, referendum year, so they won’t polarize each other anymore.

Referendum seems to be along referendum lines as we saw last election/referendum where we have mayoral candidates proposing referendums. That obviously ties the two together.

We are probably not seeing that in the next election. I’m cautiously optimistic is what I’m saying.

To bring it back to the clarification cards that you showed before, did those get manipulated?

As in photoshopped?

Yes. Do you see photoshopped versions of those clarification cards circulating?

We do but they so far are not making a lot of impact mostly because the ministerial memetic engineering is so good now that...For example, the ocean council being a very new agency, a new council that’s just set up this year.

They posted -- there was supposedly a typhoon that narrowly missed Taiwan -- that if you do dangerous activities when there is a typhoon warning already posted, we may not save you. "We May Not Save You" is a title of a popular TV series. I don’t know how long ago.

In any case, that one went absolutely viral. That Facebook page from almost nobody following grow to be, I don’t know, 100,000 now. It’s by one viral post. They’re now very good at this thing. It’s very hard for a photoshopped version to get that kind of virality.

I wanted to make sure Max has a few minutes as well.

Max is always around.

I’m always around. Tell us when you have time to...

Feel free. I have nothing after this but you two have a plane to catch. It’s you who have the deadline.

Sorry. I’m still taking notes from the last one.

It’s fine. You’ll have a transcript... but I understand that note-taking aids long-term memory. [laughs]

Exactly. You were talking about how the supporters of one party will be motivated or incentivized to perhaps characterize people from the other side negatively.

No. Excluding them from the polity, portraying them as non-citizens.

Right. In some circumstances, it can be difficult to know whether these claims are true or false. Is there any way that the parties themselves are policing these sort of claims?

I wouldn’t think it’s a party-platform thing. In Taiwan, we have this, I don’t know how to translate 側翼, something like a radical wing. Flank?

Flank.

It refers to popular social media presence, like Facebook pages that everybody knows supports a particular party or political candidate. However, that candidate always disclaims any responsibility or authority over that particular Facebook page.

That, I would say, is the majority of where those exclusionary rhetoric came from, especially the top three parties. They would not do something as blatant as portraying half of the population as non-citizens. It’s not to their benefit to do so. Their flanks are very happy to do so.

If worse come to worse, they can disclaim responsibility. Those flanks are the most vocal ones during the election. There was a foreign policy article that says a few of those flanks were operated by two employees at Tencent.

I’m sure that they did any number of VPN to manipulate a Facebook page. I’m sure that you haven’t been operating there. Sometimes with foreign help. The candidate in question always disclaim any responsibility or indeed any link between their own campaign and those flanks, which is why this radical transparency on political advertisement is so important.

Then it let us know whether they’re actually funded by some extra-judicial force or whether they’re funded by the candidate themself. This is a very important thing to know. We didn’t have that data in the previous election.

Great.

I’m sorted for questions that I was interested in asking.

Awesome. We’re good.

That’s great.

Thank you so much for being game and engaging us on these questions. Really appreciate it.

No problem.