Are you comfortable being recorded?
Yes.
Great. Thank you. Okay, thank you so much. We really appreciate this, and I think we're coming back with some more context and questions. Feel free also to make it more of a conversation.
I think maybe the question that I really wanted to ask you was: you use a lot of metaphors in your work. Something that I've been thinking about a lot as we embark on this project is how metaphors can help us see things, but they can also prevent us from seeing things.
A lot of the metaphors that I've seen in concerns about information integrity have been cyber warfare metaphors, military metaphors, combatting...
Whereas I always say “epidemic.”
Yes. And you really draw from public health frameworks. I'm curious if you could walk us through that choice and what you see as the value, as well as other metaphors, other ways of thinking that you suggest we look into.
Well, I mean, the phrase “going viral” already stems from epidemiology, right? Viral is first and foremost a public health word. So I think it's quite useful if we decouple the toxicity, the bad effect that information manipulation has on people, from the virality, the cultural transmission, and things like that.
It is only when we do that can we start looking at the digital spaces as conducive to a more healthy respiratory arrangement or places that are almost designed to get everybody sick. That is not found in the cyber warfare metaphor, because in cyber warfare metaphor, there's really no deterrence.
It’s all about gaining ground, influence operations, deception, and gaining control of systems. But Facebook and other platforms, Twitter and so on, are not passive battlegrounds. They are themselves nexus of pro-social or anti-social dynamics. Using cyber warfare is, I think, to reduce the agency and the active role that these platforms and their algorithms can meaningfully change the nature of the conversation.
Whereas using a public health metaphor nowadays, people all know that a well-ventilated place is very different from a non-well-ventilated place in terms of COVID prevention and so on.
But with a virus, there's no such thing as a good virus.
There are, of course...
Can you give an example?
Do you mean good biological things? For example, we have this idea of symbiosis where people coexist with the microbiome system, which is not a virus, of course.
Okay.
But it is something that is transmitted in a culture for its probiotic properties. By extension, one can also say that inoculation or vaccination is a good virus that the state really wants to get so that you develop antibodies against the worst virus. That was actually the original idea of inoculation before modern vaccines and so on.
So, I mean, it's good if it's preventative against something that's worse, or if it's good if we can tolerate a side effect, but making sure that we can use the probiotic symbiotic effect. So, yes, I would say there are good viruses if it prevents me from getting infected by a bad virus.
And technically speaking, there are good viruses that just didn't incorporate into our DNA, and they're just now part of us.
Right, exactly.
It's built in, it's like evolution.
Okay, let me give you where I'm trying to go with that.
Biology. [laughter]
Well, it's important when you use metaphors, the deeper you understand the metaphor, the more you can play with it. What I'm thinking about is if you work in a social movement space or you work in organizing, right?
Community organizing, often you're trying to get what you see as positive messages to spread. And so sometimes when I look at the bridging algorithm space, or when I look at some of these efforts to make our digital conversation spaces more robust to the spread of problematic messages, I really worry about the ability of community organizers and social movements to do that work.
And why is that? Because what the bridging algorithm does is it makes something more prominent. If it's more bridging, it's not making virality go away, it's just making sure that we only produce viral vaccines or viral cures rather than viral hazards.
I think I worry about it because sometimes those messages come from very small groups, right? And groups that are very marginalized. And so whenever you really foster bridging, you can almost end up with the most milquetoast position.
I see what you mean, yes.
Yeah. So it's hard to fundamentally shift a conversation or to throw a really wild new idea into information.
That stems from the upper bound of chasms in our imagination that bridges can cross. If we are now reasonably sure that there is no chasm too wide to build bridges, that argument would not make sense because it just takes longer to build a bridge. But we can always build a bridge, right? Even across a tunnel, whatever.
I think it depends on the competence of everyday citizens and practitioners when it comes to bridge building. And whenever the social divide feels insurmountable or impossible, I would say that it's because our competencies need development, not because there's something fundamentally against this kind of diversity.
And you think that's competencies like social competencies and interpersonal competencies, digital affordances, and digital competencies.
It's all that. As I mentioned, as long as the spaces are generative in the sense that it fosters newer bridge-making skills to be demonstrated, then it's a kind of learning space.
Because by definition, bridges are composable. If you bridge community A to B and B to C, it increases the possibility that A and C will also be bridged.
So it's like a true Haskell program, right? [laughter]
Exactly. Yes. It's conflict-free, replicable data types all the time. I can do natural transformation, but let's not go there. Category theory, right? Exactly. So the homographic movements that the bridge makers do, as long as it's in a space that increases cultural transmissibility, the space itself doesn't need to figure out how to cross the chasm. It just gives rise to more and more modular and composable bridge-making narratives.
I think we sometimes feel a little bit hopeless because the spaces online are not conducive to that composition. Rather, they flatten things, decontextualize things, so that the only things that go across are massively simplified sound bites or dunks, retweets with comments, and things like that.
And when only those things cross the community chasms, these increase the hate of both communities. And so it fosters a kind of connection, but of a negative valence. So that's, I think, the main problem of today's intersocial social media. If we change the algorithm so that only pro-social valence gets viral, I think they are naturally composable.
Do you have any online spaces that you really like?
Many? [laughs]
Are there some that you think do this well?
Well, if you look at how Gov Zero is organized around Slack or IRC or Telegram or whatever, most of the online part is just to capture what's happened in the physical or synchronous spaces, like with VR or Google Meet, so that the newcomers do not need to go through hundreds and hundreds of pages, but rather can join in a productive way. So the online space is just there to make entering the next synchronous meeting or hackathon easier for a newcomer.
It's more like a summarizer than anything. It's not there to replace the synchronous co-production. So something that is anchored in iterative, continuous social co-production, something that is definitely a social object that everybody can just patch, just fix and fork, and so on. I think that is a good space.
And of course, Microsoft now sponsors GitHub, which is another such place. And in both cases, the social co-production and the code co-production, the importance is on lowering the threshold of initial entry and also making each synchronous meeting or hackathon or meetup or whatever composable. And again, it goes back to the initial two pillars we talked about.
Are you still contributing on GitHub?
Of course, yes. If you check my GitHub profile, it's all green.
Love the good green streak on GitHub. Wow.
Not bad for an active minister.
I wonder if there are other active ministers with that kind of green streak on GitHub. So just one more question about GitHub and then I know Robert has 27 questions. Our colleague Denae Ford worked really hard on some really cool stuff on GitHub around a community maintainer toolkit. Recognizing that open source is wonderful in theory, in practice, you can be trapped forever in maintenance. And what it means to support people in maintenance, and what it means to bring in new people and give them things like green streaks for community building, might be kind of a challenge.
And I'm just curious. Yeah, because we're absolutely going to go home and chat with Denae and run these ideas by her, I'm curious if you've noticed that, if you have any thoughts in that space that we could share back.
Of course, the ways to publicly give kudos to someone, to attribute credit in a way that celebrates a community instead of just a hashtag somewhere and so on, all these are important. And I think the great thing about the open source movement is that from the very beginning, it's built upon people who understand how difficult it is to convey positive valence over written media.
With well-known personalities like Linus Torvalds, who publicly repented so that he now only says "hug you" instead of something else. The care people put into online communication is indicative of how easy it is to troll a mailing list, to troll IRC, and so on.
Because of this evolutionary pressure, the open-source community that did not have robust community resilience ended up getting forked. By classic evolution, now we see the outcome of that selection. All the larger communities in the open-source space have some sort of communal resilience.
You know the phrase "Wikipedia without Wikipedians doesn't work"?
Of course.
So it's almost like social norms.
Yeah, Wikipedia is just an excuse for editors and curators and nerds to meet each other.
Yeah, yeah. I have many more questions, but I feel like we should probably focus in a little bit. Robert, do you want to ask yours?
Increase the ventilation?
Yes.
You mentioned, in one of your essays, you said, "We did pandemics without lockdown, we can do disinformation without takedowns."
Yes, infodemic without takedowns.
I wonder, like, I hear people talk about freedom of speech versus freedom of reach. Is that kind of a viewpoint that you...
Yeah. So during the three years of the pandemic, you could always freely move from Taipei to Tainan. There was no domestic movement restriction. On the other hand, that's because we had one of the strongest border quarantines.
So it's not "no freedom without borders." That doesn't work. The very fact that we can maintain freedom of movement, freedom to play baseball, or whatever during the three years of the pandemic is because we have a very strong border quarantine policy.
In the infodemic, it is equally important to distinguish foreign sources from domestic sources. We usually don't do attributions beyond saying it's foreign, mostly because, well, physically we can see something transmitted through the subsea cables to Taiwan. So we know it's foreign, but how many hops they jump through, we don't know. So it's hard to attribute to a particular regime or jurisdiction.
On the other hand, having this distinction is important because then the freedom of movement or speech or assembly and so on is extended only to residents and citizens, like people who have a claim or have a stake in our polity. We're not blindly saying that everybody else needs to have a vote. This is suicidal to say that. And this is very Ostrom, right? A clear community boundary that is safeguarded recursively across many levels. This is the precondition to this: without lockdown, without takedown, always with a big caveat of except at a border.
But if somebody in the media... You're suggesting like an information quarantine policy almost, right?
But we don't have a great firewall.
But if there's cooperation, yeah.
Yes. So for example, why the Facebook civil integrity team can implement this kind of advertisement transparency with the incentive of not getting fined by collateral liability is that we have a very robust way to prove that you're a local business or you're a local citizen and so on.
We have the capability of not just a national ID card for universal healthcare, but also the TW Fido or the citizen's digital certificate that is also given to residents. So if you're a resident, you also have two chips, one for your residency card and one for your health card. Even if you're abroad, it is not difficult to prove that you are a Taiwanese resident or a citizen.
So it then becomes easy by simply requiring that you need a local SMS number, basically, which requires this kind of two photo ID to basically identify that you're a resident. It's not without vulnerabilities. For example, a few months back, it was discovered that Hong Kong telecom abused their roaming arrangements or whatever. So you can manufacture SIM cards or eSIM cards from thin air or whatever, but those are exceptions. Once it's discovered, and again, it's for a scam ring. So it's always first discovered there. Then we, of course, clamp down and so on.
So now we can, with reasonable accuracy, say that if you authenticate using SMS or TW FidO or something, we can tell an actor that is domestic versus it is foreign. And so we don't need to do that on the delivery part. It's attributable to the actor.
So, yes, okay. So when somebody gets it, like, I got an eSIM at the airport, and when you do that, you check their identity, you make sure that they are a resident or qualified. And then is that also unique? Each person only gets one.
Yes.
Okay. And so that gives you this robust entry point to the information system and a way of checking.
Yes. And that's how our contact tracing works.
Yeah.
That's really the secret of the Taiwanese contact tracing system, is that it only needs an SMS and nothing else.
And then are there privacy concerns with that? Do you then end up with every...
Our contact tracing is zero-knowledge in the sense of, it's sometimes called oblivious storage.
The way it works is that if you go to a venue like a restaurant during the pandemic, they will print a 15-digit random number embedded in a QR code, but you can also manually send the digit to a well-known number, 1922. You just scan that, and it's an sms: something that appears on your SMS app of choice, and then it says the contact tracing code, 15 digits, this information is for pandemic control only and not to be used for any other purpose. I personally wrote that.
If you press send and show the screen to the restaurant worker, they only learn the 15-digit number, which they already know anyway. They don't have your phone number, but they know you have sent it on this timestamp. But they already know the time anyway, so they don't learn anything about you except for nothing.
The COVID status?
They only know you have sent a message to check in, but they don't know anything about you that they don't already know. So it's zero knowledge on the venue. This message is sent to your telecom company and not to the state.
The telecom company already knows your phone number anyway. They actually already know that you're in the vicinity because of geolocation cells. And they don't know what those 15 digits stand for. So again, they don't learn anything about...
They don't know that you went to that restaurant.
They don't know. They just keep this 15-digit random number for 28 days before deleting them. And we did an interpretation with the Ministry of Justice so that these are not subject to wiretapping because these are not communications. This is just a pandemic measure. It's a check-in. So unlike, I will not name names, but many other nearby jurisdictions...
This is very different from my experience in Shanghai.
Exactly. Right. So different from pretty much any other East Asian jurisdiction. The law enforcement, if they're investigating a criminal case or whatever, they gain no knowledge.
Yeah.
And so we get to keep this voluntary. At no point was it mandatory. You can always just manually write your contact and put it into a slip of paper that you get into a physical box and so on, which is good for people with no phones or people with seeing difficulties or whatever. So it was always opt-in. But people opt-in only because they know that the state and telecom don't learn anything that they don't already know.
And when there is a community outbreak, of course, the local contact tracer will go to the venue, they will get the 15-digit code, they will send a command to the five major telecoms, and they will recursively send exposure notifications to people who checked in. And at that point, anyone can go to a well-known website and see in the past 28 days which contact tracers in which city have requested access to their check-in.
So it's mutually accountable. So we have transparency, accountability, and voluntary participation. And that is why this system held until the first Omicron, which is unheard of. Most other places fell after the Delta wave.
Did you have enough participation even though it was an opt-in system?
Yes.
It's surprising to me, given the behavioral psychology evidence, that opt-in versus opt-out are fundamentally different thresholds. Do you have any sense of why participation was so high?
Because the venues don't want the hassle and the privacy breach repercussions. If they don't post a 15-digit random code, which, by the way, are free software, online generators everywhere to generate a 15-digit random code is not rocket science.
If they don't do that, then all the guests will write their phone number on a piece of paper and put it in. They have to destroy it after 28 days. And if they don't take care, the people queuing down the line will see the phone number of the people queuing before them.
Which is what we had in Shanghai before the health code came. A lot of sexual harassment.
Right? Exactly. If I really like the person queuing before me, now I know their number, and no venue would like that in their name. They become liable for everything and gain nothing.
So the voluntary participation by the vendors, I mean, when we decide that, we only say it's mandatory for MRT and high-speed rails and train stations, which are state-controlled or state standard because rolling a 15-digit number is so easy. So I think the first week, 2 million different vendors just generated their own code.
Wow.
And to register into the database of receiving exposure notifications, all it takes is their SMS number. So if I'm a small vendor, I don't need to file paperwork or something, I just send an SMS to verify that it's actually my shop.
So it's analogous to if you're a small business payment platform.
Exactly.
You don't want to lower people's credit card...
Exactly. It's a strict liability for you.
You want to get paid and not have to worry about... You don't want to know anything that goes...
Right. And so as long as there's like 80% of vendors in a night market adopting that code, it creates immense social pressure on the remaining 20%. So within a month or so, everybody started printing those one and two SMS characters.
So when you have the system set up and you have this quite robust use of SMS as a privacy-protecting way to verify that somebody is a resident or a citizen, why don't the platforms use that?
But they do. If you want to create new Gmail accounts or Facebook accounts or whatever, at some point, like on the second account, it starts asking for that.
And on PTT, like the local Reddit equivalent, you need to pay the SMS by manually texting. And go back to PTT because it's non-profit, right? So it doesn't have the money to pay Twilio. So you have to pay that in by SMS.
So, you know, something that I've heard that I haven't verified is that, for example, with the Spamouflage case, right, when many accounts were taken off of the different platforms, there were not Spamouflage actors on PTT. I'm curious if you think that was a factor.
Yeah. So PTT was a victim of fake accounts for a very long time. But the great thing about PTT is that it's open source. You can find its source code on GitHub, and people who want to improve PTT can just send pull requests. It's been in the social sector for 25 years now.
So obviously, people very quickly converged on SMS and in an active mode, on the passive receiving mode as a way to verify. So yes, I think at peak infodemic 2018, PTT was one of the first to switch to freezing all new account creations. They purged the inauthentic coordinated accounts. And then for every new one, you now need to do this reverse SMS authentication.
So they know, the freezing, you can now make new accounts on PTT again, right?
But only with a reverse SMS.
So the freezing period, that was sort of a period where they were trying to really clean up their ecosystem before. That's really helpful.
But Facebook doesn't have... I mean, you can have a two-factor authentication, but there's not like a little flag that says, this person has a Taiwanese phone, right? There's not...
No. If there are disputes, like if you're flagged as a robot or something, then you're prompted to prove. And if you want to run advertisements, that's just a freedom of reach. You have to authenticate that you're a local resident, at least, if not a citizen. To run advertisements, you have to have an invoice number, I think.
Okay, that's very interesting.
Invoice, like the tax invoice?
Yeah. So if you go to the ad library, it will actually show the unified registration number. So you need to have a registered company number, which means you're at least a resident.
We don't have this in the United States. It's quite easy to get a SIM card in the United States. Am I wrong?
SIM card?
Yeah. I don't think you have to register or prove residency or show a passport. Yeah.
I've never thought about it.
I just got my SIM card last year in Michigan. I don't remember them checking my passport.
I'm just wondering if this is right. Like, we've been studying misinformation and the role of generative AI in disinformation, and so far most people are like, haven't really seen much.
There are issues here, there is a lot going on, but this sort of massive increase in scale, massive change in quality. Most of the organizations that we're talking with say, I'm seeing a few here and there, but nothing massive. And I'm curious if this is the bottleneck.
Just to your earlier point. I think Twitter probably said that they're using credit card numbers for the same purpose. You're asked to pay a minuscule amount just to prove you're a human. And if one single credit card number created 5000 different accounts, they will know.
And if somebody manufactured 5000 credit card numbers, well, the banks will know. So it's piggybacking on the bank system like we did piggyback on the telecom system ultimately for the same purpose. And I think because social security numbers are not used that way in the US, just like our health card number is not used this way in Taiwan.
So I think credit card number is by far more preferred. Especially you can do debit cards.
Yeah, I don't know if you can.
When you get a credit card number, then there's...
Some identity verification.
Social security number.
I suppose you could try to pay for a SIM card with cash in America, but I've never tried. It'd be interesting to find if that's possible. You can get burner phones...
Yeah, I mean, the problem is, right, we have like a non-negligible fraction that is not banked. We have an unbanked population.
Exactly.
And we don't want to disenfranchise them.
Exactly. And with SMS and photo ID and with Taiwan being famously known for unlimited 4G bandwidth for $15 a month, everybody has that. I think this is more universal. But of course, the bank system is stronger.
But I think we push for the more universal system simply because people who are residents, people who are unbanked, people who are under 18 are equally active in the Taiwanese democracy. So it makes no sense to exclude it just because they're a resident or they haven't turned 18.
Okay, so that's a key difference between using credit cards versus SIM cards in a context where bandwidth is considered a human right and you've made active efforts to spread, is this strong equity concern that the credit card approach...
Yes, and I would say yes. There's a strong factor in being clone-proof-ish in the social media platforms.
Yeah.
I wonder if, kind of going back a few steps to your focusing on infodemic as opposed to cybersecurity. I guess one thing that I struggle with, with that metaphor is that a virus is not sentient.
That's a worm, right?
It's not malicious. I guess you can argue, like, it adapts its techniques based on evolution, but it's not consciously seeking to...
To do things other than survive?
Other than survive. So does the metaphor struggle a little bit there in terms of when you have a dedicated, well-resourced actor with strong intent?
With a P3 lab. I'm sorry, not going there. But actually, that's where it intersects.
Right.
If you can arbitrarily increase virality on something toxic that's not viral, or if you can arbitrarily increase toxicity on something that's already viral, if you engineer that, then it's still a virus, but it's now also a weapon. I'm not going conspiracy theory.
So in our discussion so far and in the interviews I've read and seen of you, very deliberately, you only say foreign. You never specify...
Which jurisdiction, that's right.
Our regional neighbor. Is there a particular reason why?
Well, because we cannot attribute conclusively just based on IP address or metadata. If you just look at an IP address, it's almost always from some botnet. And for what we know, those botnets are for rent. Anyone can rent them.
Right.
So, like the origin of the kind of state-based actor, we can make educated guesses. But even Microsoft just says it's Volt Typhoon or something, right?
Even cybersecurity might...
I don't know, I would say...
... much is de-emphasized relative to just
...
Yeah, I think Microsoft stops at Volt Typhoon or some other typhoon, shows characteristics that it may be funded by a certain nation state, but they never say that it is of that nation state in origin, which is different. Like, this may be a PR company or whatever.
Yeah.
And now there are nation states that actively collaborate on these things.
That's right.
Including national neighbors.
Yeah. I think Westphalian taxonomy is the wrong thing for this kind of fight, right? Westphalian taxonomy, meaning like the nation state.
Right. The notion of a nation state is wrong because it's like trying to trace COVID to a nation state origin, and it may actually come from a certain nation state origin, but it doesn't matter when it's out.
Mmhmm.
So whether it passes to Italy and then to someplace and back to Taiwan, or whether it goes straight from Wuhan to Taiwan, it almost doesn't matter. When you work in pandemic control, you just want to safeguard your order, regardless of which place the airline is coming from. We realized that very quickly in the pandemic case.
But isn't that -- I mean, that's in part because other nations weren't inspecting their borders. If there were some that were doing a really good job, then you could coordinate just...
Yeah, we had travel bubbles, Taiwan-Palau, Taiwan-New Zealand, or something like that. But because these were few and far between, we basically said that, no, the quarantine is just there.
And, I mean, you made a really good point. It's conceivable that maybe, you know, Iceland, New Zealand, Taiwan or Palau or whatever can be a major group of a shared border.
Of a digital, you know, these are the Facebook accounts that get blue checks in this, right?
Exactly. No, this is really good. And I would still just say that it just extends the notion of domestic. So the domestic foreign distinction is still there. It's not like domestic foreign and allies. When you're doing travel bubbles, you're essentially considering them domestic. Taiwan is many islands, actually, so we already have some of them.
Whose job is it to say, you know, when we're online, privacy is important, but it's important to me that this person be real, that they be located in the place that they say that they're located, and that they be unique. Right. Whose job is it to make that happen?
It's everybody's job. Your job and mine.
Yeah.
So if it becomes difficult to be authentic, one needs to empower each individual to choose whatever way they're most comfortable with to stay authentic.
That's almost like a trick question. Because this is like asking a central arbiter where power needs to concentrate now that we have shapeshifters, but that itself assumes a dial, how much trust you implicitly trust that arbiter.
But from the contact tracing SMS case that we just brought up, it's evenly split between the telecoms, the venues, the individuals, the people who make those SMS scanning apps, including LINE and so on. And when it's everybody's job, there's real check and balance, so that when somebody wants to abuse that, everybody else can make a feasible alternative to bypass that and render that attempt at capturing obsolete.
If you start with saying that we need to, and I'm sorry for saying that, trust karma for everything, the health code for everything, then it overly concentrates power so that it then becomes the nexus in which people want to stop protests, people who want to stop citizen assemblies and so on, they naturally hack into that.
So infamously, I can't remember which province, which city in China, but there was essentially a bank run, so people were protesting and we can't get our money out of the bank, and all of a sudden, the people who were assembling all got red codes. They couldn't go anywhere. So the health code was just famously abused.
It creates a nexus of power with no check and balance.
Right? That's the Hunan case, yes.
Can I take this to algorithms?
Sure.
Something that we've been talking a lot about is participatory inputs to algorithms, right? You can think about an algorithm like a lot of contemporary platform algorithms, as a nexus of power without checks and balances.
Of course.
But something that I have been delighted by as I get into this work has been the use of MyGoPen and TFC in the algorithm. Right. That if they produce a fact check, we don't know exactly...
[inaudible]
And so I'm curious just for your thoughts on kind of this, like how else we can incorporate lots of forms of contributions into algorithms and whether this provides us a mechanism for decentralizing some of that algorithmic power.
Yeah, so as you mentioned, the fact that Facebook decides to outsource whatever with the IFCN members shows that there are nexus powers with credible neutrality that can act not just as advocacy, but as co-producers, as symbiotic producers to the information ecosystem.
I think one of the contributions that Cofacts people make is that even within people who make like browser extensions, and I will use one very concrete example like Gogolook, which is a local startup, it has this app called Message Checker, and when you install it, I think it's Android only for now, it scans all the push notifications. So it doesn't care whether you use Facebook, LINE, or Instagram, because it works entirely locally, like a spam filter does.
Interesting.
So the more you move to the edge, to the downstream, to the deployment site, the more freedom you have to work with multiple information sources. So instead of waiting for Facebook to change its algorithm, you can just like actually advertisement blockers is the canonical example, you can just customize your experience by working with things like Message Checker.
So you're moving the algorithms to the edge.
Yes.
Wild.
But that's how spam fighting and advertisement blocking work.
Yeah. Okay. You'd have to require the platform, right? There's been some new legal work around requiring the platforms to create both just uncurated timeline in addition to the curated timeline, right? Where it would just be by time. And of course, the problem is it's terrible.
It's not an effective dopamine-like slot machine.
Yeah. Okay, well, that aside, it also is not an effective source of information.
Yeah.
But it's a great input to an edge algorithm.
That's how podcasts work, right? So they use just this ancient format of ATM or RSS, but all the innovation happened on the edge on the podcast readers and players.
If you had a magic wand and you could do anything you wanted in this space.
Oh, super intelligence alignment. [laughter]
Really?
Yes.
Can you say more?
Yeah. So my main research interest now is to automate as much as possible for a community of people who just have layperson's interest in an agentic system, to steer this agentic system the same way as actually how political leadership works, because we don't know the details of all the expertise.
I know something about machine learning, but I don't know everything about the digital ministry. But I can make spaces where people articulate their goals, sometimes conflicting ones. I'm pretty good at conflict resolution or conflict co-creation. So I don't need to understand the details of any of the expert arguments. I only need to know facilitation. And once it's facilitated so that people can all live with this shared rough consensus, that's our policy. This is how we make multistakeholder policy.
So imagine that each participant is a superintelligence, something that is higher in cognitive bandwidth than human beings, and then human can just play the role of essentially a jury or an arbiter and so on, and say that, okay, based on this back and forth debate or whatever, I say that we now agree on something that is still within the continuous norm of civilization, and you may now proceed on these constraints.
And as the intelligence artificial ones are becoming more and more symbiotic with humans, like cyborgs and so on, I think this is one way out of the alignment problem, that we just incrementally increase the ability for people who are good at facilitation, but not as good as the AIs are in cognitive capabilities to shape this kind of debates or forums or whatever for continuous alignment. So if I have a magic wand, I'll just accelerate that research.
Do you want to ask a question, Robert?
I want to build off this.
I think in one of the things that I read, you used an analogy of diversity. I think the gist was that maybe the attorneys being like the AI and...
Yes, they debate each other.
They debate each other, and they're doing. And they're making very sophisticated arguments, and
then the judge does not need to understand.
They don't have to know the details. Most judges don't.
For example, the patent case, I guess.
Yeah, exactly.
So they don't need to understand the details of the technology and the patent.
Yeah, because the flaws in the argument will be pointed out by the opposing side.
Right. So what is the role of the judge in this situation?
A judge or a jury is to make sure that instead of just the technical conversations, the result of this judgment will meaningfully continue the blended volitions of civilization so far, right?
So instead of having one disruptive case that sounds good and sounds reasonable and sounds rational, that disrupts the societal fabric and forces people to invent 16 new words for aunts and uncles, a constitutional judge will say that, okay, so based on your debate and narrative and so on, here is our interpretation that somewhat changes the societal norm, but in a smooth, continuous fashion that takes into account all the arguments about human rights, equality, and whatever, right, but without going all the way of saying the winner takes all.
And now we need to invent 16 new words for aunts and uncles. So a judge is at this point more of a kind of buffering role, of ensuring the epistemic continuity, of ensuring the plurality is still here, instead of going all the way to singularity, so that the society still makes sense at the end of each, this kind of arbitration, so that it doesn't converge to a point where there is no meaningful diversity after that point.
I'm trying to understand. So the attorneys, in this case, the AIs, you rely on them to their superintelligence, to be able to check each other in terms of the quality of the argument.
Yes.
And then the judge...
Would then take the agreed-upon common understandings or form of, as well as the points of dispute. One side would say that society needs to be organized this way, and the other would say society needs to be organized in that way in response to the transformative technologies.
Many a time without this kind of structured debate or co-creation, what we often hear from acceleration people is that, for example, somebody who is pro-progress and puts a minimum stake on continued human participation, would say that something that optimizes the utility of this or that function on the progress side is worth the sacrifice of human participation, or meaningful participation, or the decentralization of power or checks and balance, or whatever. Or that another may make the case that if you do not impose the draconian measure, a limitation on compute or whatever, then the risk will outweigh all the progress gains or whatever.
And so you need to implement this particular security measure, and so on. Without this structured co-creation or overlapping consensus, one would expect the site that developed the most persuasive AI systems to basically win an argument and constrain the future toward their particular viewpoint.
But the point I'm making is that we don't need to accept any particular progress-safety trade-offs, or progress-participation trade-offs, or participation-safety trade-offs. We can make a space so that only the ones that take care of those overlapping concerns and also open up more spaces for nuanced overlapping in the future.
So instead of discounting the future, we put a premium on future possibilities. And the judge will always prefer those solutions, those arbitration results, that open up for more possible overlapping between those three different value concerns. That's my main argument.
I'm sorry. Enumerate the three value concerns again?
Yeah. So that's democratic participation, like not leaving people behind. That's progress, which is efficiency, utility, automation, all that. Prosperity, all that. And safety, which is against existential agentic risk or whatever.
So the safety, the progress, and the participation arguments are sometimes posed as making trade-offs. But once you start getting into the trade-off mindset, you're foreclosing possible futures where all three are taken care of simultaneously.
My main argument is that if you do not take care of all three simultaneously, then you do not get to a sustainable future. Your light term becomes narrower and narrower every time there's a stronger breakthrough in artificial intelligence.
When you say participation... I think about participation, given my background, as two things, or in two ways. Number one, is this kind of like almost like a moral value that for something to be participatory means for it to be inclusive or to bring...
Nothing about us without us?
Exactly, yes. And then the other thing that I think about, and I do not think of these, they're under the same English word in my brain, but I don't think of them as the same thing, is the sort of collective intelligence that the more information you bring in, the better the outcomes are.
Yes.
Is that, when you say participation, are you thinking, I guess, inclusion and diversity of thought, or are there other things?
Yeah, these are the two main allied lines.
Right.
So to not leave people behind is to ensure their continuous safety. So this is the participation-safety edge. And to make a co-creative solution that solves wicked problems that cannot be solved independently. That is the participation-progress edge.
And what's the progress-safety edge?
The progress-safety edge is to concentrate power.
That's a strong statement.
Right? The progress-safety edge, really, the industrial-military complex is all about being accountable to each other by reducing the nexus of power, the variety of nexus power, so that two-body systems are more stable than three-body and all that.
And so to concentrate meaningfully as hierarchies or as whatever other topologies is to ensure continued progress and safety, because they can more meaningfully make confidence-building measures toward each other when the number of players is small. But to the explosion of emerging new voices and people who are epistemically involved.
Is there also a naive, positive framing of the progress-safety edge? One thing that I've heard in several cases, the research, as I understand it, is quite mixed on this, but sometimes aligned AI or AI with good guardrails is actually a lot better to use, a lot more usable for the end user than AI without guardrails.
I think the word guardrail is perfect to describe it. So instead of guide rails, which is more participatory, guard rails are there because the powers know better than individual citizens.
Interesting. We've talked about co-creation as a piece of participation and inclusion as a piece of participation. But I'm still struggling with the piece of, like when you have the broad participation, for example, in a democracy, you can still end up with really bad outcomes for marginalized and minoritized groups, right?
And that's where you get things like bills of rights, human rights frameworks, or equity orientation that says okay, like 90% of people think that this is okay, but for this 10% chunk, it's really, really important. And so we're not going to accept it, right? And so I'm curious how you, I mean, this is just like the majority problem, right?
I think it's just that the bitrate is too low. When you collect preferences, if you leave everything to a referendum proposition, it's literally one bit per proposition. If it's among eight parties, you don't often get eight, four parties, then it's just two bits per one.
So if you are forced to make your preferences known and aggregated with this not even eight-bit quantization, then you're guaranteed to get something that's very skewed. And which is why increasing the bit rate of democracy is as important as reducing the latency.
We've mostly talked about reducing latency. When we talk about real-time contact tracing, pandemic, infodemic, it's about responding here and now instead of four years later. But equally important is the increased bitrate. So people can propose nuanced co-creations, like repurposing the Taipei Main Station seating, a floor, and so on. Imagine if you're forced to do that only by voting among four different candidates. You cannot express that.
So reducing the latency of democracy is about changing the pace at which information turns into action, right?
Right. And people would watch on a national broadcast as a journalist asks all the questions in real-time and so on. So it shortens what's usually regulatory speaking is usually a 60-day cycle into 24 hours. So that's reducing latency.
Okay. And then changing the bitrate is about...
Expressing preferences in a way that is not individualized and that is structured to allow more dimensions.
Okay.
So like all our ideas of Polis, because you get to type that statement instead of choosing among five survey questions, it automatically increases the dimension every time somebody types a new statement.
Okay, can I ask you just about the latency piece? Another book that's on my to-read list is Recoding America by Jennifer Pahlka, and she writes about how, at least in the States, we have a sort of like, we almost front-load democracy. We make these grand strategic approaches and then just assume implementation will be taken care of. And it sounds like your framing and approach is really
about constant iteration and implementation.
Yes. So we're taking all these tools, collective intelligence, and all that, and using them in the public sector. My ministry has 600 full-time staff. Any 30 of them can sit on a petition on a local discourse instance. And we always live stream every month everything that passes, like 30 likes, and if it has 100 likes, and it's a discourse forum. So of course, with more structured and unstructured inputs, I am basically pre-committed to implementing everything that 100 of our 600 people agree on. There's also negative votes, so like net 100 votes and so on. This is not like the top-down industrial power you see in Yes, Minister or other movies. This is an internal collective intelligence system. And we do that now also with Talk to the City. We're just deploying it internally so that you can even just hold book reading clubs or public hearings or multi-stakeholder conversation, focus groups, whatever, and then just throw everything into Talk to the City for that kind of visualization. So I think bureaucracy itself, because the staff, all of them, are very good at multimodal communication. It's part of their entry exam, right? So they're by definition capable of high-bandwidth input into policy conversations. So you just need to install such apparatus in place in order to structurally respond to it every week, every couple of weeks, and finally every month in a live stream.
Is what you're referring to the Join platform?
No, that's for the citizens. So this is alluded to as internal Join. This is what's called MyModa, which is the Moda internal instance of co-creation.
And when you do the collective intelligence project work using Polis, VTaiwan, Remesh, do people have to verify using their SMS?
They actually do if it's on the Join platform. One of the registration requirements is to verify your SMS. And this is also because there's participatory budgeting on the Join platform, so you can only participate in your resident municipality or city. And for that, we need your SMS code.
Interesting. Cool. As an urban planner, I love participatory budgeting and I love strategies that enable people to have more of a voice in their locality, more of a voice in the lines of governance that affect them most. Really deeply interesting. Are there other places that you think are operating on this level?
Well, we learned all these from other places. The pro and con two columns with their reply button to each. We learned from Better Reykjavik from Iceland. The PB dashboard and everything design, we learned from Barcelona and Madrid, which is why we keep saying jurisdictions, so that I can just say Barcelona. And then for many of the pioneering uses of discourse, like rough consensus, consent instead of consensus, we learned from Loomio, which is New Zealand. We're just one tentacle in this collective intelligence lab. Recently, Sitra from Finland deployed Polis nationally in their national conversations, so we're also learning from that. They made Polis GDPR compliant, which we're looking forward to merging.
They did? That's cool. Anything good from Estonia?
Yeah, well, our T-Road system is an homage to the X-Road system of Estonia. I think Estonia shines because of its commitment to public code. It can afford to do what anywhere else would be called too centralized deployments of electronic voting or whatever, by the virtue of them being public code.
So not just open source in code for programs, but also code of conduct, code of regulations and so on. That shows how the public sector is integrating people's contributions, including bug bounties, of course, but also meaningful extensions and so on to their public code. The X-Road has now evolved into the Nordic Institute, the NIIS.
So that Finland and Iceland, I think, are co-maintaining with Estonia this fundamental system for data exchange, the data fabric layers. So we look to Estonia for this infrastructure of stuff.
One more question and then I'll hand it over to you. Or to you, Vickie, the talent. If you have bug bounties, you have all of these open source systems, all of this is made better when residents can co-produce this tooling. But in order for that to be possible, you need your students to take computer science, right?
Maybe just Scratch. I wouldn't call Scratch computer science.
Yeah, and maybe with AI, the barrier to entry is lower. But certainly in my own experience, I found that when you bring in students from diverse places around the world, students, computer scientists with all kinds of different backgrounds, what you get is tooling and approaches that are more usable to the communities that they come from frequently, right?
And so I'm curious, if you think about building the capacity of people here to fully participate in this work, how you think about that?
Yeah. Which is why we insist on competency instead of literacy in education. In the curriculum, starting 2019, the basic education, we change media literacy, digital literacy, data literacy, whatever, to media competence, data competence, and digital competence.
Because literacy is when you act as an individual consumer, but competence is when you work as a community co-producer. So not only it is about creating social impact, but it is also about the mindset of common good. Like, a student would only get high marks in competency for media or digital or data if instead of comprehending the material or whatever, if they translated it in a way that makes a positive social impact.
In fact, one of the most viral petitions we get from the Join platform to ban plastic straws from bubble tea takeouts is a civics class assignment to a senior high school. The civics teacher basically said that there's a Join platform, find something, make it go viral, start a movement, without specifying which or how.
So this is a classic competency curriculum, and then the student, being 16 or 17 years old, gets a lot of empathy from 16 and 17-year-old people who care about sustainable environments and very quickly forced a shift in how bubble tea and straws interact with each other. So I think these kinds of examples let people who are 16 or 17 know that they are actually leaders, agenda setters in democracy.
Instead of not yet mature enough to vote, instead of apprentices, they get to be like, almost mentors to show a possible new direction outside of the existing power balance, that we can actually have a new design that takes care of the decomposability of straws or whatever.
Fascinating. Thank you, Robert. Vickie, do you have questions?
Yeah, the literacy versus competency conversation is great.
I have one.
Yes.
How would you suggest I get my mother involved? Like, my mother is 64 this year. She is comfortable with her smartphone and her tablet, but is terrified of a computer and a keyboard. So how does someone like her meaningfully participate in our digital democracy?
Well, one thing is to introduce her to Cofacts.
Yes, I've done that already during this research.
That's one of the entry points because senior people also want to be the provider of truth to their family. Nobody wants to be a provider of rumor and discord.
So just by signing her up with not necessarily Cofacts, it could be Gogolook or Trend Micro or whatever bots that are downstream to Cofacts, they become active participants. They can share the real-time clarifications, they can ask for clarifications. And many of the downstream bots also protect against scams and phishing and all that.
So I think, yeah, this is about appropriate technology, meeting them where they are instead of asking them to use a mouse or a keyboard if they like, beautiful line charts to share with people, or quick videos and so on. We need to crowdsource using that.
And that's, I think, what the Cofacts playbook of being almost a hub, a repertoire. Instead of just one way to interact with Cofacts, you can interact with it with any number of downstreams. I think that's where it's really strong because then it allows people to cater to different senior people's communication styles.
Thank you.
Is there anything else we should be asking that we haven't asked?
Well, many, yeah. So let's see. I think we started talking about “helping the helpers” in a strategic, not tactical way.
We talked about competencies in education, about identifying a solid social sector, about bridge-making as something that we push the platforms to actively do instead of passively react and so on. One of the work we are doing is to change the way the state funds this kind of work, because these kinds of works are not strictly speaking charity. They are co-production.
TFCC produces useful work that can be actually sold to Facebook, Facebook can buy it. And on the other hand, we always worry about what's called a mission drift. The more useful your co-production becomes, the more likely that somebody with a profit motive will want to maximize shareholder value and then take over it.
So for the state, it poses a conundrum. If you put a lot of public sector money into these places, sometimes you're actually asking it to privatize because the shareholder gets guaranteed profit. So it can get a lot of social impact investment, but then it also pushes it to be not entirely controlled by a mission. Maybe they restructure into a capped profit.
As just a casual example, I don't know where that comes from. [laughter]
Yeah, totally arbitrary. [laughter] But the general direction of this kind of transformative technology is always like that because it's useful to its original purpose, but it's also useful for so much more. And that so much more, especially if there is a network effect, will always create an outsized influence. So there are many possible ways to navigate this.
In Taiwan, we have mission-locked companies — essentially a charity owning a company — which is becoming a more used instrument now because we have special voting shares that are owned by a mission-oriented charity that can always veto. But we saw how well that played out for OpenAI.
So it's not a cure-all, but there are also ways to structure the government funding so that it flows. It's called pay-for-result. The US also has a procurement, not procurement, a funding bill for this. And in web3 it's called retroactive public goods funding, so that you're giving funding not as investments, but always as awards. And the award is pre-committed, but it's only awarded to the positive social impact that the project has caused as an ecosystem.
So instead of choosing one team over another, you would foster meaningful synergistic co-creation between these, like GitCoin has some of that too, with quadratic funding.
So I think there is something to be said about comprehensively taking care of this kind of positive network effect, easy to drift in mission, transformative open-source work, and classify it as not traditional digital public infrastructure, but almost as a civic infrastructure, and change our funding algorithms correspondingly.
Can I give you a different metaphor? It's not fully formed yet, but it's something that comes up for me over and over when I think about this space, which is in the climate change and ecology literature, the idea of ecosystem services, which is where you look at all of the value that's created when a farmer adds a hedgerow for bee biodiversity or something like this.
Various researchers have gone and shown that what's called ecosystem services, classically not considered on balance sheets or budgets, it's not funded, it's not resourced, but it creates billions of dollars of value. And so there's been a real move to make that measurable, to show its value and then to both fund it but also to tax for it. So I'm curious if that's a useful metaphor.
Yes. Just as we talked about how participatory budgets increase the bitrate of democracy by extending one or two bits into any bits of mutually supporting projects, now we're seeing, of course now there's dollar accounting and carbon accounting, so at least two.
Exactly.
Well, there's nothing preventing the same infrastructure for carbon accounting to also account for social impact. And so multivalent accounting scoped to an ecosystem or a bioregion is also very useful if one can build regenerative economic models so that people can thrive on non-fiat exchangeable money systems.
That is partly what I was implying when I said that we need to relook into our funding infrastructure because the Presidential Hackathon already did some of that. The system works by quadratically voting for the top 20 projects out of 200 or so that work on one or more SDG systems. But because of QV, we can find synergies between multiple teams and the top five teams domestically each year receive no funding, there's no prize money.
It's just a presidential guarantee that whatever they demonstrated to work on the municipal level or district level will be incorporated into national policy the next year. So this is another accounting. This is the policy influence accounting. That is given as the prize to social innovators. That fixes a lot of perverse incentives because they cannot gain any money even if they put in a lot of work into it. So it only encourages the kind of people who build cross-sectoral solutions to maximize social impact.
Fascinating. Are there other people you think we should talk to while we're here?
Let's see. For specifically information or contextual integrity, you've talked to the fact-checking organizations. Have you talked to the private sector people about this, the Trend Micro and Gogolook people?
We're planning to talk to Gogolook. I wasn't planning to talk to Trend Micro, but maybe we should.
Yeah, I would encourage you to do that because Trend Micro is a good example of a company with a strict cybersecurity mindset and then moved to counter spam and scam and then moved to tackle information integrity. They carry this continuum from safety to participation.
The way that it gets encouraged toward this movement around 2018 is in itself a very good story. It's a good showcase of the State Department actually doing a lot of good because the bootstrapping for that was funded by the Global Engagement Center, part of the US diplomatic service. This is a general pattern of people not going out of the way to build a social enterprise but finding a particular product or service within a for-profit company to be aligned with the social sector.
Because we use an umbrella approach for the social sector, we're not excluding them just because they're a public-listed company. One division or one department within that company now becomes almost like a charity. But they started as being a division within a for-profit company.
Interesting. Helpful, thank you.
Sure.
I think that's all of our questions for now. Robert, are you good?
Yes.
Okay. Thank you so much for your time. I'm going to turn off the recording.
