Plurality and Digital Democracy

Transcript

Okay, welcome everyone. So, I'm going to kick this off. I'm going to give a very brief introduction and we'll talk for about 40 minutes and then we'll have plenty of time for Q&A. I am very very excited to have Audrey here. Audrey is very much a intellectual hero of mine. I think one of the uh few real thinkers out there. So, it's really an honor. Um Audrey's been doing a lot of work and thinking in the realms of democracy and technology and where those fields intersect. and particularly with uh around an idea called plurality that we'll get into. Um Audrey was the digital minister of Taiwan, one of the youngest ministers in Taiwan's history.

Um and is really a civic hacker at heart. Um one thing I want to touch on is why do we find why is this conversation important at edge? Why do we care about these topics? And we really think about societal change from the point of view of experimentation and systems. And I find that plurality is the most coherent set of ideas that has put um put concepts around societal change towards human flourishing in a way that people can understand. It's kind of a meme that has really started to spread and it's really exciting. But it's not just a set of ideas. It's also a practical set of tools. and we'll talk about Audrey's

experience in Taiwan and some of the actual experiments including most recently in the US which are really exciting. Um, so we'll get into plenty of those as well, but I thought just to kick off, it would be great to hear from you, Audrey. Maybe let's start at the start and help us understand, maybe paint a picture of how you went from a teenage civic hacker to eventually being one of the youngest ministers in Taiwan's history. And so maybe take us to those teenage years. What were you doing? How were you feeling? And help us understand that. Thank you for that. I thought we need to start from the very start. Like when I was born, I was born with a heart defect when I was four.

Doctors told me and my family that this child have a 50% chance to survive until heart surgery. And I eventually got one when I was 12. But for the first 12 years of my life, I fall to sleep every night feeling like flipping a coin. That doesn't land well. I don't wake up. And so I get this very interesting existential opportunity uh that I call you know publish and then perish. So then I learned to you know document what I learned every day on um tape recorders and then floppy disc. I don't know whether you remember those uh but anyway uh and then eventually bulletin board systems and then the internet and always free of copyright in the public domain because if I don't wake up well people can still you know compose upcycle uh

whatever I produced and so that led me to uh essentially civic hacking. I uh attended three kindergartens, six primary schools, one year of middle high before dropping out to do entrepreneurship. uh we built uh the earliest like C2C auction like eBay and things like that in Taiwan got investment from Intel and and and things like that. So did my entrepreneurship stint uh when I was 14 15. Um but it was uh much more interesting uh to engage the open source uh previously fuser compute communities and so I did a lot of work uh with especially the pro community and internet standards communities uh in my kind of teenage to early 20 uh years and that led me to this idea of gz0v or govzero um govzero

is a meme uh gzv.tw TWW started 2012 basically saying for all the governmental websites of Taiwan something.gov.tw if citizen doesn't like that instead of uh protesting uh we can actually demonstrate in a demo sense we can build equivalent thing and something gz0v.tw TWW. So just changing an O to a zero uh gets you into the shadow government. Uh whether it is uh the budgeting site, whether it's the dictionary or really anything, health services, whatever. And the civic hackers can always like mount a shadow government uh website uh always open source free of copyright and then kind of using mass mobilization to force the government to merge us back. Um and

so that worked so well that by 2014 uh we won the trust of many uh civil society organizers. And so when they uh decided to literally occupy our parliament building in 2014 uh to protest uh the trade service deal that would have allowed Huawei and ZTE and other so-called private sector from Beijing uh to our telecommunication or publishing industry and so on. um half a million people showed up on the street and we occupied the permit peacefully. But thanks to people in Gozero, uh we provided live streaming and transcription and all the situational applications that enable this protest to turn into a demonstration. Basically half a million people on the street can show up and just go to any of those open

space. Uh like there's like 20 or so tents and you can choose one site that you care about, environment, human rights, labor right, whatever. And then you uh enter your company name uh and then you find exactly how the trade deal affects you and then uh we live stream the conversations and every day then we do sense making and read to the plenary to literally everyone that this was agreed the previous day and these were the remaining issues to be talked about today and after three weeks of peaceful nonviolent occupy we managed to produce probably the only very coherent proposal from all the occupy movements around this times which the speaker of the parliament then say oh this is a better set of ideas let's just adopted. So it was a occupy that ended up victorious.

And so we kind of invited oursself into the government in 2014. And by the end of that year, I was tapped as the kind of young um what what's called reverse mentor youth uh mentors to the cabinet ministers. In Taiwan, we have this system where people under 35 advice cabinet minister. I was 33 at the time. So we work together to bring this process from the occupied parliament to somewhere online because we want to resolve other contentious issues like Uber or Airbnb or marriage equality or whatever without occupying government buildings again. So we replicate the process online using uh police and other systems we call it v Taiwan. It worked so well that in um 2016 when I become uh 35 and therefore too old to be a reverse

mentor, they promoted me uh into the full minister in that position and then I have my own reverse mentors advising me then. Wow. Well, that is such an incredible story. Just to just to mirror that back to you. So, this was the Sunflower Movement in 2014. There was a very obscure uh obfiscated policy that the government was trying to put into place in terms of a deal a trade deal with China. 500,000 people protested including UN folks occupying the actual cabinet of the government, the parliament, the parliament, sorry.

And then based on the efficacy of that process as you made everything transparent to the people and the government realized that you this was actually a force for good they not only you know followed through with your ideas and um suggestions but they also then hired you directly. They kind of must because the approval rate at the time was 9%. So in a country of 24 million people anything the president says 20 million people saying huh? Uh, and so anything that gets buried at 9% approval rate is worth a try.

So Taiwan is a relatively recent democracy. Is that right? 1996. Yeah, that's the first time we did a presidential vote and I was actually that was actually my first uh website. It's just a poor one of the presidential candidate because my dad was a speaker uh and organizer. uh of course he didn't win but uh it's a variable exercise and in Taiwan actually that means that it's not two kind of people one doing democracy and organizing and another doing technology and innovation for us it's literally the same people the same generation and we imagine democracy like a technology that gets better as more people put time into it feels interesting to me that the democracy of Taiwan is actually kind of more recent than let's say some of the early

technological movements like Netscape and There was even like browsers and and you know early Windows before before Taiwan became a democracy. Do you think that is relevant to the the the efficacy with which you've been able to pick up digital tools and use them as as ways of organizing? Yeah, definitely. Because we don't think of democracy from internet terms, right? If you just vote one referendum topic every four years, there's just one bit literally. Uh so like very low bit rate, very high latency. uh even if you have like a ballot of like 16 people that's like four bits and and so uh we analyzed that and because people already experiment with this online way of organization so we started our conversation with um for

example citizen assemblies around healthcare reform participatory budgeting we learned from Brazil citizens e petetitions and things like that because when we democratize all these become available already kind of on the world web with the browsers help us understand a bit more about the GZV about Gubzero as a as a movement as an idea is if I understand it correctly essentially you were taking data that you had access to and making it much more legible to civil society right it was things that the government was not sharing and you would essentially make dashboards and give much more clarity there is that right yes um and so one very quick example um for example in Taiwan early 2020 uh we had a mask

shortage issue and people queued up buying mask and so one person uh from the govzero collective uh just decided to prototype an app that let people using like usahidi like uh self-reporting uh to report which parts still have mask left uh but because they tap into the same supply that we have been building uh as gov zero with the government on visualizing air quality using you know air boxes PM2.5 sensors that people can put on their balcony and their schools and so on so there existing networks for air pollution, water pollution, noise level and so on.

So because this trusted API endpoints are already there. Um so I at the time already a cabinet minister work with the national health insurance uh to basically work with all the pharmacies and in three days we provide a real time open API uh refresh every 30 seconds uh to do those maps. So instead of people self-reporting now we have a very clear um kind of demand and supply uh real-time data for people to visualize and this has two effect uh one is that the panic uh died down because people know very predictably like when is the next uh batch of mask available and then it unleashed a lot of uh citizen um like civic hacker energy because then people can say oh the distribution was not fair because it was prioritizing uh by

equalizing the distance but in the urban area and Aurora area 1 kilometer means like 1 hour versus like 5 hours or something like that, right? Uh and so we should actually prioritize based on time and so on. So the opposition party instead of just protesting again uh they also have their data analytics team uh they also work with open street map they are kind of forced to provide a demonstration not just a protest and then when the interpolation happened in the parliament uh from the opposition uh to the ruling parties minister the party minister basically just said okay um legislator you obviously know about data analytics and you have a better algorithm let's deploy it next Thursday so it turns conflict into co-creation and so That uh again uh was also how we

visualized uh the vaccine preferences uh and so not only people do not panic about vaccine. We also don't have an antivax um political faction because we turn that energy through open data through dashboard into this my brand is better than your brand uh movements. Um and so there's people who swear that only the Taiwan locally produced medicine that they're only getting that. There's people who swear that that's not good. they're only getting, you know, BNT because TSMC uh sponsored that or Astrazenica or Mona, whatever. I got all four brands just to make a point. Um and so we we show people very clearly in which jurisdiction, in which municipality, which age brackets and this is their preference map. And so when the next batch comes, it just goes

to next age bracket that prefer uh this kind of uh vaccines. And so it turns the very competitive energy, a polarizing energy even into again a co-creation energy. It's like my my soccer team versus your soccer team, but we all love the same sport. It it's it's such a good example. It strikes me that democracy in Taiwan working well and efficiently is extremely important. It's a relatively young democracy. It's a small country. It also has a very strong and motivated antagonist very close to its borders that I believe is relatively regularly trying to poke holes through cyber attacks, through disinformation campaigns, etc. Can you say a little bit

about why democracy working well is so important in Taiwan? Yeah, pressure makes diamonds, right? So uh in Taiwan when we after we survived the 2014 squeeze of the trade agreement uh of course Beijing tried various other ways to influence our polity and they have really only one meta narrative which is democracy never delivers democracy only leads to chaos and that's it right and then every year they try a different variation like in early 2020 the it's the idea that only authoritarianism can lock down properly and therefore fight the new virus. Uh but by the end of 2020, Taiwan only lost seven people to COVID. So obviously democracy does things better. We've actually not locked down any city. Um and so just by this

kind of radical transparency and citizens uh voluntary participation and also what's called humor over rumor uh we we managed to uh overcome uh this kind of polarization attacks. Um the SCZero was also responsible for Kofact, a collaborative grassroots facteing uh process. So instead of a state running anything that uh looks like journalism and things like that, it's a citizens that can annotate them and then again using uh polish-like algorithms where people who disagree mostly with each other if they manage to both agree something is important uh like the left wing or rightwing uh on community notes on X and Facebook and on YouTube if they both upvote something to the upwing and that's the uncommon ground between those

two camps. And so for example early 2020 um there was a strain of meme that says uh N95 the highest grade of mask is the only useful mask because our stars experience there's another strain of meme online that says is ventilation is aerosol any kind of mask actually hurts you uh and I file hurts you the most but before these two kind of polarize into extremes um through you know antisocial social media engagement through enragement uh this kind of uh monitoring device that is built by civil society in other discovered the uncommon ground. So very quickly in 24 hours we pushed out this meme, a very cute Shibainu putting her paw to her mouth saying wear a mask to remind each other to keep your dirty unwashed hand from your own face. So

it's a facial protection device from your own unwashed hand and and that is actually because it's so cute and also it's it's not disputed. uh the science was still out then about mask efficacy but the fact that handwashing works uh I think nobody dispute that we measure tap water usage it did really did increase and everybody laughed about this mean basically depolarized uh themselves and so the importance of this pressure makesment idea is that we need to double down on democracy uh we cannot fight authoritarianism by censoring oursel by um like giving up our top of Asia internet freedom and we can actually get much more trust if we decentralize um democratize defense that is to say the

DIA uh story and uh we make sure that people feel that they can actually contribute to cyber resilience to information resilience to bio resilience as we just talked about to geopolitical resilience and so on and so every time we crowdsource those idea the people feel that the state trust them uh a little bit more some people decide to trust back and so by 2020 uh from 9% in 2014 uh we're back to like 71% approval rate uh by 2020. Oh, that's really fascinating. I think there are so many, you know, I can't help but uh look through the lens of being an American citizen, spending time mostly in the US, seeing or hearing everything you're saying and thinking, wow, it's actually

so encouraging and inspiring to hear that democracy, you know, can work. I feel like right now in the US it's a very polarized time and uh you know in Taiwan you've had to figure out ways to deal with that because you can't let polarization get in the way of you know maintaining safety and you know staying functional especially in the face of an adversary. Um I'm curious if you have thoughts on how this relates to like Taiwan to me feels like a very fertile uh ground for digital democracy, right? It's a relatively small population, high education rates, um internet everywhere um and things like this. Do you feel

like these ideas map onto other places, let's say the US or you know some of the other kind of polarized western societies? Yeah, definitely. And in in a sense um we have seen many of these experiments playing out in smaller polities. For example, in Bowling Green Kentucky, uh they've run exactly the same polish playbook two times now. So if you go to what couldb.com, what could Bowling Green be? um they just concluded their latest round and I think one10enth oneif uh of the wearing county people uh participated in the police conversation uh which basically shows you one statement from your fellow wearing county uh residents um like I don't know I I prefer uh that um Bowling

Green has a distinct identity from the nearby Nashville that is one statement or that we should all be speaking Russian that's another statement a real statement uh we should mandate ketchup uh in drive-thru French fries. That's another statement. Um we should anyway. Um and so uh you you see one statement, you can either upvote or downvote, but there's no reply button, so there's no room for trolls to grow. Um and after you engage a while, you will probably feel that you want to chime in, so you can chime in. And there's a great visualization that shows your avatar among the kind of different groups and also the bridging bonuses given to the ideas that transcends uh those partisan or ideological divides. And so there's a very beautiful um visualization report out of uh Bowling Green that you can

first see the bubble chart of all the large topics like transportation, infrastructure, um culture identity. If you click it, it gets you the subtopics and you click subtopics and it shows this beautiful chart uh where things everybody rejects is on one side like uh we should all be speaking Russian and things that we people all agree on the other side and there's the huge amount here. So basically just looking at this group group Sophie alone um reduces polarization because it's a pro-social media. It is not the antisocial media. Antisocial media amplify the extremes and pro- social media amplify the uncommon ground and so the uncommon ground gets much more viral and much more voice and then people model after those uncommon ground and chime in even

more. And so at the end of the day most of the ideas like I think 85% or more uh are actually um a ground square support and then the mayor and the committee listeners everywhere around Bowling Green can then say okay let's implement this uh these are the low hanging fruits it seems everybody agree and it can actually back propagate to the actual statements to find out the exact words and so the topicalization summarization this feedback and so on of course are all using large language models that they're using it in a non-h hallucinatory Okay. Because nothing is generative, right? Everything is grounded to the specific statements.

Can you talk a little bit about the citizen assemblies that you've done in Taiwan and maybe ground that ground that in an example um that actually played out? Sure. So, I'll use a recent example. We've done many citizen assembly including on the you know singlepayer universal health care that also covers dentistry which is very good. Uh but that that was like 20 years ago. So I use a recent example. Uh so last year uh we witnessed uh on Facebook and YouTube and other places there's a lot more advertisements featuring fake people and like Jensen Huang Nvidia CEO at Taiwanese and like he would say he would say that I want to give back to the country. I'm giving you free crypto and then if you click it uh then Jensen

actually talks to you. It sounds just like him. Uh and uh later on actually looks like him uh but it's a deep fake probably running on Nvidia GPU. uh and it crowds out the actual advertisement from small and medium enterprises and people get scammed sometime a lot right uh but Facebook was you know earning money because the scammer paid more so they have this perverse incentive to not fix this um but the Taiwanese people also summarily rejected an earlier proposal that would give something like digital service act the uh government uh inspection uh rights uh to those large social media people are like no we don't want the state to step in. Okay, so the state cannot step in, but the frauds are real.

So what do you do? Right? So we sent 200,000 SMS text messages to random citizens, random numbers around Taiwan and only asking one simple question. How do you feel about information integrity online in light of defects? So people share their feelings and many people thousands also volunteered to take uh almost an entire day a long half day to deliberate online about this issue to talk through their feelings of each other and then very quickly uh thousands of people sign up. We did a uh sortation uh a stratified random sampling to ensure that we choose 450 people that is statistically a microcosm of Taiwan in terms of the same ratio of uh

occupation, gender, place they live, uh age brackets and so on. And so they deliberated in 45 rooms of 10 people each, but there's no human facilitator. The room itself facilitates. So if somebody is quiet, the room uh nudges them to speak up. If somebody disrupts, the room keeps that to five seconds and the room transcribes again summarizes shows the uncommon grounds uh border voting to move on to the next topic and so on. And so for example, one room will set um let's flip this around. Instead of showing an advertisement until it's flagged as scam, which doesn't work, let's assume every advertisement is scam until it's digitally signed by Nvidia or Jensen Huang saying, "I am Jensen and I

approve of this message." So, anything that doesn't KYC is scam. And another one may be saying, okay, if Facebook keep posting those non KYC uh advertisement, let's find them, but let's not stop at finding them. If somebody gets scammed for $5 million, let's make Facebook pay for that $5 million. Full liability. That's the only way they will comply. And another one may be saying Tik Tok by dance. They don't even have a Taiwan office or legal representative. So if we just punish Facebook, it's actually like, you know, giving uh Tik Tok a lot of leverage because we cannot actually find them.

Are we going to confiscate their assets in Beijing? I don't think so. Um and so what to do? Oh, we should slowly slow down the connection to their servers. So their video loads slower and slower until all their business goes to their competitors. And you would note none of these are content level measures because people in Taiwan don't like content level measures from the state. These are actor level and behavior level measures which people feel as proportionate. And so after the deliberation we weave together 45 rooms again with human facilitators there's no way 45 small group facilitator can just might melt and figure out the overtone window the overlapping consensus but language model do that like in a minute or less and so we can then say to the plenary these are

the compatible proposals from all the 45 frames and so let's vote on it and so at the end of the day we can show all the legislators more than 85% of consensus across all democratic fig groups around those kind of measures. So I said three there's a lot more and then uh that was late last uh March and by last April we check with all the big tech on visibility. Uh last May we uh wrote those uh into law last July it's all passed and so this year if you scroll Facebook or YouTube in Taiwan you don't see any fake ads anymore and we remain the only jurisdiction uh that does this for like non-financial non-electoral just any advertisement ever. Right? So and the point here really is that people sometimes say

deliberative democracy is slower than the pace of technology but if you use AI as you know assistive intelligence you can then augment group intelligence so that it evolves faster than the actually technology harms uh that's happening to the society. So it solves actually a big governance problem which is just this time lag and I think that is why the alignment assemblies uh we call it is now being replicated around the world and also on a global way uh with the collective intelligence project here and also being emulated especially in Japan they call it broad listening is such a such an amazing example of how what you can you know empower people to

have a they actually find a solution without getting into what could be relatively authoritarian um ideas against free speech and actually like maintain freedom of speech whilst achieving the goal. So yeah, a wonderful example. I I'd love to move to plurality now. Um and maybe for the folks who don't know the idea or haven't read the book, could you kind of briefly describe the idea of plurality? Yeah, we already described, right? Plurality is this idea that the societal conflicts, the social differences are like fire on the ground.

Before combustion engine, we we can only, you know, put it out. But if you have a good combustion engine, you should drill down and you can get the energy from the difference from the conflict. And you do two things. First you make it visible so that we know that we have such differences and then you highlight the uncommon ground the things that connects people despite their differences and so it's like combustion engine that uses conflict as fuel and then it fosters two things. First it turns uncommon ground into common knowledge. It builds um epistemic security resilience because then people know it's okay to be different and that we all you know agree on these common

measures. And the second is that it makes sure that the decision-m apparatus is has a very high bandwidth. all the measures that I just mentioned basically means that um the people can feedback especially negative feedback like the government shouldn't do this and shouldn't do that in a cycle that is much faster than four years right with the four years election you get 51% of people insisting doing something and 49% of people feel they have lost utterly and so in four years they do something completely reverse a lot of flip-flopping uh but if you solve this with higher bandwidth it turns out if you pull people individually they're like in imbeies or nimbies they're very extreme but if you pull people in 10

people groups actually everybody is like mimies okay maybe in my backyard if you do this if I do that with caveats and at the end of the day you get very nuanced policy proposals so polarity again is celebrating fostering social differences and then using that as a fuel for collaboration and it produces both this civic muscle societal resilience and it also improved the bandwidth of democracy. You start the book that you wrote with Glenn Wild of called Parality kind of to set the stage and you say that right now we're walking a thinner and thinner tight rope between uh the forces of technology with AI that can enable authoritarianism on one side

and the kind of general algorithmic enragement through engagement domination and doom right to use words. Yes. Exactly. And and and it sounds like so so maybe just touch on why why now is this important and why now is it possible because the technology also can enable civil society, right? It can enable plurality. Yeah, definitely. And I think the way we use AI is important. Um many people feel that those single tongue super intelligence uh take off whatever story um like a very um monotheistic thing uh that uh sees everything does everything eventually leave everybody

behind is not a very convincing story because if you only have the story the only two thing you can do is you know f and doom right you either either say oh accelerate because it will lead us to posthuman or you say stop it stop it pause for 6 months or something. Uh, and it's like very deeply unsatisfying. And so, Plurality tried to offer the third option, uh, which is instead of just doubling down the gas or to put in the brake, uh, how about the steering wheel? How about just turning some of that energy into something that is not a vertical takeoff path, but a horizontal path of again decentralized democratic defense. That is to say, not like mainframes, but like personal computers.

Um, so not like monotheistic imaginations of AI, but rather like animism, like Shinto, like Dowoism, like permaculture, right? You have like 17 or so species that fulfill important ecological functions. And each of them are very small, very explainable, runs on my laptop, uh, very tunable and communities can own them and even if some of them like escape out of control, it's not a big deal, right? So I think this pluralistic alignment idea is now really gaining ground. I was just in Japan. There are new unicorns like Sakana AI and so on are all based on this idea of pluralism a ecosystem of AI lives or agents or co-evolution or digital nature. There's many way to call

this but definitely not a vertical takeoff. You've likened Democracy Thomas to almost an open- source project and that that makes me wonder what does you know what does the pull request equivalent look like? What does upgrading what is V3, V4, V5? How do you think that looks in the years to come? Can we actually upgrade the processes and the infrastructure? Is it this idea of higher bandwidth democracy that you've been talking about? Yeah, definitely. Yes. Uh I think one of the main point that we we say when we do digital democracy is that it's not direct democracy. Direct democracy again is pulling individuals and if you just pull

individuals and give them binding power like referendums, it's actually very very dangerous. It's very flammable. Uh it's not a very good combustion engine. Um but you would note that we always pull people in groups in Polus is the cluster groups of people who feel similarly uh in the alignment assemblies is the room of 10 so on and so forth. And the importance here is because then it builds civic muscle. Um you can easily say oh to be more efficient let's have a chatbot interview each person individually and that's a democracy of chatbots. they can just you know uh have a council among themselves convince each other and then go back uh to their human users and just persuade us or using some neuro attack to brainwash us into into

acceptance will be borked right so so that is one one vision uh in silico democracy um but that's like you know sending your personal robot to the gym and to lift the weights for you and it's very impressive I'm sure it can lift a lot of weight but the day the muscle doesn't grow this way right So to to me democracy is that muscle is that civic muscle and so the digital democracy tools that we make today is more like like exoskeletons that it let us like lift much more. It let us exercise our care on a larger scale uh to like much more people scaling out but also much more deeply scaling in and also uh much easier to replicate uh like with edge cities scaling across and so but it

remains just a exoskeleton um and the care that's exercised is still like between humans and it strengthens human relationship and the fabric of trust which is to me the main uh output of Yeah. Can you um so I'd love to ground this in a couple of examples. I'd love for you to tell the story of the governatorial candidate in Japan who read your book and then was very inspired to run their campaign in a certain way. Yes. So Takahano uh last year uh was probably the only person in Japan that runs with broad listening as his platform. Like nobody know what that even means. It's a foreign word uh in Japanese. They know broadcasting but

what is broad listening? Uh and turns out he read the book plurality and decided to run one month before the gubernatorial election of Tokyo with no party support just 30 something. Um and so and he doesn't have a platform. So because of that he said okay let's crowdsource my platform. Anybody can tag hashtag Tokyo AI on any of the SNS to help to chime in. And he again used pol talk to the city the usual sensemaking tools and so on to crowdsource that platform and for senior people because intergenerational friendliness is big here for senior people uh they can call in uh into a phone number and then a voice clone of Tako Ano would just talk to them and for people can tune in to

his uh YouTube channel uh so that he's he's also a VTuber so his avatar uh like just live streams 24/7 so whenever there's an interaction there's a GitHub pull request literally really poor request and whenever there's a update to his platform then the uh VTuber Ano just announced it uh in real time and so just a few weeks into this he got a very coherent set of platform and actually before the voting day it was independently ranked by think tanks as the top performing uh the best uh platform even better than Governor Kohik um of course Kohik still won that's her term but she was so impressed that she immediately tapped um Tokyo into becoming adviser sort of like my role

when I was 33 uh to uh Tokyo Gvtech uh and so they built together this broad listening device for the Tokyo metropolan area and it got the attention to all the different parties. So when I was in Japan uh a few weeks ago, all the three parties, the two uh opposition and the one ruling all said that we're doing broad listening even better in honor with this with that and it's very good. There's a race to the top. Why not? Such a cool example of how the how we can go from theory and ideas and discussion to, you know, writing a book and then the ideas actually start to spread to, you know, not just Taiwan but also Japan. It sounds like now with Bowling Green, it's

also in the US. I know you're doing some work with um Gavin Newsome in California. Oh yes, I'm going to Sacramento tomorrow. Yeah. Uh and so uh there's a new platform called Engaged California, Engaged. And they uses which is open source sensemaking um project that some people in the DAO space actually used first. Uh and so the idea um at the time uh was that we need to solve uh the same issue about social media governance not about deep fake but rather about the teenage youth uh and because uh California uh really um like they I think they think uh in a very frontier kind of way two years ago because uh the teen teenage people the young children never voted in

the legislature that put in kind of the ban on teenage use of social media and so it's a classic sense of something about them without them and and so the governance team and his first partner Jennifer was like how about let's just involve the the children uh and the children can in a s representative way sets the rules about social media use you know maybe social media is fine if you have a daylight computer right instead of a you know addictive touchcreen and a small screen um so on and so forth and then we consult with their parents and then we can set uh the state rules about social media and on the day of supposed launch I was here um and the wildfire happened so it did not get to launch. Uh on the other hand this quick pivot is kind of good because

wildfire is a acute issue. Uh unlike social media which is chronic issue and acute issue elicits much better grassroots engagement. It gives people in the grassroot organization uh roles much more things to do. And so we got a lot more signups than we anticipated like 5,000 6,000 just in the first week. The click-through rate of the email was like 25% which is like five times higher. Right? So you get an email from from Newsome saying you know chime in about Wi-Fi recovery and prevention in polyon. uh most likely you click through it and then they just published the first uh response, the agenda setting reports and they're like lowhanging fruits that everybody thinks is a

no-brainer uh to do uh like mental health support for recovery centers and things like that. there's a whole list and then there's like four uh main divides so that they can run deliberations on this main device like as we rebuild should we allow even more people in uh the area uh as we make infrastructure should we prioritize public sector or private sector and so on so forth and so I think this is a great example of a like sudden pivot that lends much more legitimacy to the process because of urgency air cover and the politically and then uh in the May budget bill refresh uh Governor Nome basically said okay this pilot is really good so let's allocate a bunch of budget and make a civic infrastructure so that the next time around when we want to

deliver about really anything then the data and uh innovation office in the state uh people uh commit to maintain this open source uh apparatus so that it can be used to deliberate pretty much any level of issues that is so cool how was it in Taiwan how are these tools um kept up to date because you said it Wasn't the state was it then universities? Were they involved? Was it kind of public private partnerships kind of? Right. Exactly. Yeah. In Taiwan, uh we always work with credibly neutral partners and this is very important because if it is seen as a partisan platform, then all hope is lost, right?

Uh and so for example, when we first launched V Taiwan uh with Govzero, we're strongly associated with the national academy. uh and uh the conversation tools many of them uh are based with like student clubs in the national Taiwan university uh like the PTT uh and because NTEU produces all the presidents regardless of their parties so they're kind of okay it's academic freedom is arms links um and so PTT has been running for like 25 years it's all open source no advertisers no no shareholders uh it's just a students club basically uh and so I think the point of working with credibly neutral for people is equally important as uh in Taiwan. We see it in Japan where they work with

like smart news uh which is a news aggregator kind of like Yahoo uh known for like being nonpartisan actually presenting like ground news balanced viewpoints. Um and so yeah, find incredibly neutral partners I think is the first step to unlock and the second thing is to win over the trust of the career bureaucrats and because they need to know that this is here to save their time and also to reduce their risk. They won't accept a time risk trade-off but if there's no trade-off then you expand the parto uh frontier.

So you've you've spent time at Zoo Switzerland. You've now visited Edge. Um you obviously know Vitalic. Well, you were you at Zuzalu or you at least know of the concept. Um I'm curious if you have any thoughts on how these kinds of experimental communities, these network societies can be a force for plurality and how the kinds of experiments that we can run and ideas that we can spread. Yeah. I mean you you all are the research lab for this, right? I I'm just a you know local developer and so the all this new ideas uh like isello and also quadratic voting which we deploy in presidential hackathon uh in 2018 I believe but that's already like two

years three years after QV was first used uh in in around the space uh quadrative funding also was done in Gitcoin and many other places for like four years uh before we adopted it in Taiwan and so and so forth. So there's a very consistent like three year or so lag between the the edge uh and and then um the the mainstream so to speak. Uh and so I think at the edge the the best thing is that you can very quickly cross-pollinate because um like no matter how good we are at installing this into the yearly budget cycle, it is still a yearly budget cycle. And so but if you want to try with I don't know AI politicians on a deep funding deep

governance round uh you can just do this and then a few weeks it's another round right so I think the iteration cycle is fundamentally different in edge uh settings versus in like actual city or governor uh settings with the yearly budget cycle and so I tend to think each edge as kind of one possible frontier one possible horizon almost like of zero like one potential fork and if you fork with the intention to get merged then it can get merged like next year or more likely two years after this. Uh but I think this is like infinite combinations in infinite dimensions of different forks and so the more edges there are the better the bandwidth for exploration.

Yeah, that's very inspiring. Um I want to get to some Q&A but I have one final question. It's actually unrelated to to the topics we've been discussing and I haven't heard you talk about this publicly yet. I'm sure you have, but um it's not it's not that bad. Darn professional. It's not spicy. Um so this week, one of the topics that we have, one of the programs that we have this week at Jazz Moralda is education. And I'm curious for your thoughts given everything, you know, with AI and as your experience as a hacker, your experience, I believe, um leaving school early as a teenager, your thoughts on the future of education, education systems, and anything in that realm.

Yeah. Well, confession time. I was uh on the board of the Taiwanese curriculum and reform uh of 2019. So that was actually my job before joining the cabinet. So in between like V Taiwan uh go zero and join the cabinet. I was in this interesting like lranche point like equally distant from the movement on one side and the government on the other and kind of in a gravity well of neither so that I can relate better. I was kind of not in L1 but like uh here. Uh so anyway like a triangular thing and so when the curriculum reform people uh decided to include like outside voices um instead of going there and sharing my ideas I just typed everything and put it online.

So on say it. You can actually find um more than 2,000 of my meetings uh that I attended in past 10 years with 80 8,000 people. Um and so which is why all the frontier model know my thinking very well because it's public domain. Uh so the the very beginning of that is the public uh curriculum debates. So the question you ask uh were asked around that time and so we I like pahano uh just gets the feedback from the civil society at large and then bring them uh into the curricular reform committee and so in 19 when we passed the new curricular reform um the central idea is this we need to switch from a literacybased paradigm uh where people consume information into a

competencybased paradigm where people co-produce information because the consuming information for road memorization or whatever uh standardized testing allows you to do um are going to be dominated by robots. If you train our children are like robots, they're not going to our robot the robots. The actual robots are actually better at being robots. Uh and so what what's left for human uh to do uh are just those intrinsic motivations and there are three um there's curiosity, collaboration and the common good. Uh so curiosity is the autonomy of uh exploring right and collaboration is uh the ability to work with people that are very different from you and common good is the ability to shape any lose-lose

situation or win-lose situation into win-win situation uh positivism thinking and so all these are the kind of core civic muscles that you must train internally and cannot outsource uh to robots as we just said and so we pivoted to this new curriculum and so the direct result is that people under 18 suddenly find that they can set the national agenda because that's what curriculum tell them to do. They're invited to start very powerful petitions. For example, going to school one hour later because science proves that one more hour sleeps gets you better grades uh than one more hour study. They did get that. And then uh or for example banning Plex straw in our bubble tea takeouts or crowdfunding a menstruation museum to

remove the taboo from the Taiwan society in just two years. And these people then become cabinet level reverse mentor, use advisors, surface role models and not just advising cabinet but leading more youthled changes. And so by 2022, according to ICCS, the international civic and citizenship center study, uh, Taiwanese middle schoolers are now top of the world when it come to civic knowledge, the civic muscle. And so people sometimes say, oh, how about their STEM, right? Are they sacrificing STEM capability? No, we're still um number three to five uh for SDM on the OECD pizza. So obviously it's not a trade-off. You can keep your kind of core competencies and then turn them into civic muscle simply by asking the

young people to assume the mantle to lead the society to set a societal agenda is like pig million effect. If we expect them to be adults, they become adults. Wow. Wrong. Beautiful. I think we're going to get a microphone that can travel around the crowd, but curious if anyone has any questions. Yeah, Neil, maybe for this first one, should we wait? You can go. Okay. Yeah. Neil, do you want to just shout it out? Um, they're somewhat related, but um uh so you mentioned that you you were you were a um one of these reverse mentors um a few years ago. Um yeah so you had mentioned that uh a few years ago you

were one of these reverse mentors which is a really fascinating concept that I think you know we would do well you know import here but uh the question is um you know you alluded to you know when you were doing it a lot of the things that you were engaging with the kind of minister on was things that were like Uber or you know social media what have you things that I think we could broadly categorize as like part of the kind of whatever you want to call it web 2.0 know or kind of the mobile era of the internet and that you were as a member of that generation more adequately capable of informing them about. My question is um now that you are a minister yourself and you have these reverse mentors, what is it that the kind of the Gen Z or the Gen Alpha of

Taiwan like the 25 year olds, 20-year-olds, 15-year-olds, what is top of mind for them? Like what is the what are the big questions about what maybe were for the mobile age to you? They what they're thinking about? I imagine it's a lot of AI, but I'm curious to get a sense of kind of really like on the ground like what are the youth kind of like uh really kind of pointing you in. Um and the slightly related question on top of that is how do you ensure that the youth kind of information that you're getting is well distributed and a good sample of the population? I imagine there's somewhat of an averse adverse selection effect where you know for lack of a better word if there's a way for I know when I was 15 if there was a way for me to get like reverse mentor on my

resume I would totally game the out of that. you know what I mean? And so how do you make sure that you have a distributed set of people as opposed to like you know wannabe road scholars whatever right so I think for agenda setting um statistical representation is overrated you don't need that because it's purely additive and pol has this great uh property that is clone proof so even if you have like two groups that are 5,000 people on this side and just 500 or even 50 on the other side you still have to get this cross-pollination consensus on both sides. So the larger group still have the motivation to convince the smaller group. So even if through this self- selection people who show up to share and so on um people

like more vulnerable or have less time on their hands and so on do not dominate the largest groups who are by the way the 17 year old and the 70 year olds because they have more time on their hands. Um it it's fine uh because then you still have to reach across the aisle uh and convince the other group. But when we're actually making decisions such as with the anti-scam uh example, uh we then use the rigorous polling method stratify random sampling to ensure that we still then have the same uh ratio as any rigorous poll. So it's a combination uh and metaggo has this great kind of picture uh that they're publishing uh a bit later. But um anyway, the point is that you can think of this like many diamonds. Um you can

have a very large kind of discovery phase and then a kind of cocon convergent uh definition phase and then you can just keep running this first diamond in the ideal diagram uh concurrently and also sequentially until such a day that you feel that okay you you understand this issue uh you understand where the uncommon grounds are and then you respond uh with the development and the delivery to use the double diamond vocabulary. So this is the first part and the second part is what does the genzy say to me? Um so I served uh as minister for seven and a half years until last year and starting last year I'm now the cyber ambassador uh cyber means to to steer so like steering technologies um like cybernetics um and so my reverse mentors when I served as cabinet minister like

last year uh one of them are actually in this community uh names mash mashbing uh and so he's uh main thing is to basically uh allow portability uh in all online services uh in Utah actually uh here in the US uh just last month they passed a law very similar to this on-ramp off-ramp idea I use that as example is easier to explain Utah passed the digital choice act that says starting next July if you're a Utah citizen and you have a social media account say on x.com and you want to migrate so to some other social network say blue sky or the fediverse um then x.com must offer two-way uh synchroniz ization so that in your new account you

retain all your followers all your contents all the birectionational conversation so that uh what cory doctoral calls initification cannot happen because then the older uh larger platform do not have the gravity effect and for one person to gradually move out they pay no transaction cost no exit cost there's a credible exit path so um the idea I got from mashm and other reverse mentors is that they would very much like this kind of interoperability and affordability in all the services. For example, we have a decentralized uh government uh wallet uh which is like the mobile driver license here, but the difference is that it doesn't ping back home. Uh and also uh it cannot ping back home and also it's based on verifiable

credentials and so and it's going to be open source so that if you don't trust the government at all but you trust your school or your co-op or whatever uh you can just fork that code and use that wallet instead, but it's still going to interoperate. So this radical interoperability I think is one of the key ideas because the genz have seen uh the result of platform lock in and so anti-lock in I think is a very strong. Um as we get to the next question just a quick note. So on the screen now you'll see an an exciting image. The QR code will lead you to a tool called harmonica. It's a open- source sensemaking tool. you can chat with a chatbot and we will be getting information around a prompt over this coming week and then that will feed into

a polus. So everything that Audre has been talking about, we can do it live. We're doing it this week. Um and then the image is all of us discovering this new tool in a reference to 2001 Space Odyssey. It will be as as transformational as the office. Yeah. Um so the uh the info is all in a channel called plurality in the group chat as well. Um but yeah sorry next question. Yeah and this plus works easily because both are part of the metagav interop project. So there's a very simple export import offering on rent. Yeah.

Um yeah. Yeah. Um yeah. As a mic, what's the what do you think the future looks like for deliberative democracy tools? So, you mentioned a lot of different tools and platforms, poll, smart news, quadratic voting. Um, what else are people working on that's interesting or what do you think is missing? Yeah, I I've been working on frankly uh frankly.org uh which is uh like the 45 rooms I mentioned. We work with Stanford deliberative lab at that time, but that was a proprietary platform. uh and again because of data portability and many other issues many organizer prefer to use uh preferably strong copy left platforms. Uh so Larry Leic uh in Harvard uh purchased uh the Kathon tool

and turned it open source and then I actually contributed the first pull request uh translating it into Mandarin and also to Spanish. I don't even speak Spanish, but thanks to VIPE coding, it works perfectly. And so, so I think that is really the thing because um the interoperability, the off-ramp and the on-ramp, the translation, all those like essential uh connective tissues were not sexy at all. And so these were neglected because if you are a thriving uh open source contributor, you probably want to work on something that gets you much more strict uh credentials, right? Uh but now you don't have to do that. You just tell you know large language models that and it just makes it happen. And so

one thing that really excites me is really this metagav interop which is using vibe coding to make sure that all of this this different parts work together. So this is the first thing and the second thing is also I think the deliberation is uh taking place outside of democratic setting that is to say it doesn't have to anchor to a mayor or a governor or a minister and so on people are more and more seeing that actually in any group uh of five of 10 and so on you can actually augment your group intelligence just as easily with these tools and so this is very good because then those 10 people already commit to each other they don't need to wait for you know another budget cycle and so on.

So what uh really interests me is this like innovation on a very hyper local scale because it's like home cooking if it's all open source and it's all interconnected you can just download that uh as kind of ingredients and even if you don't code you can just talk to your vibe coder and say I want to use it exactly in this way in this particular interface it probably doesn't scale but I don't care just just make it up right and so I think this just make it up in a like non-coder uh space with hyper local smaller communities that's probably the next horizon. Uh, no, I saw your hand up.

This is so inspiring. Thank you so much. So, a lot of people are burnt out from Dows and often times they'll collapse paradoxally from their own success. Like, Vitalic blogs about it, everyone's happy, and then the community drowns as people flood into it. So, my question to you is, do you think Dows could work better if it kept forking at a dumbar sized number? the way for example Syria with Rojava scaled to 3 million people while keeping that base unit of this Dunarian civilization forking at 150 and then having representatives definitely that's the standard like Australian solution right the I I I don't know any other solution I mean maybe use neuroche we can become bors or something and increase the dumbar number but until

such a day that happens uh I I have no other solution than just use subsidiarity as the main idea Yeah. And I think sub security the ostrong uh circles overlapping circles has another very good property uh especially in terms of dows because then it allowed people to belong to several intersecting circles with placebased communities harder to do so because a person can only exist in so many places at the same time. Uh but with purpose-based communities one can exist in like seven 10 whatever it doesn't matter and then that person become the connective tissue. So I think what's important to me uh as a pollinator now I've been traveling around the world the past year

on average I uh change time zone every 4 days uh all my belonging fits into one air cabin back this trip is 55 days in 15 cities uh and so and I don't have jell uh too quick for jell to catch up but I use time shifter is an app that eliminates uh jelly from my life um but the point here is that then I exist exist in this kind of superposition entanglements was like at least 20 different polities and and every time like one polity comes up with a great idea it just cross-pollinates very easily uh and so I think then with a lot of those uh super connectors um you do you do not actually need very large Dows right uh it's more

like flipping the idea of individuality from like individuals form groups no actually we're individuals uh first because we belong in groups and the individuality is more pronounced if you belong to a unique intersection of groups. That is actually a fundamental kind of George Siml insight that underpin plurality and pol and many of those thinkings. So consult the book plurality.net for more. All right, one last one just here. Yeah. Um, sorry. Hi, thank you so much for speaking. I have two kind of vague questions. One is it seems like from like policy obvious tools they're really

good about surfacing public collecting public opinion and doing a value alignment and I'm curious if you have insights through experience on how um the frictions in translating that into actual legislation and kind of how to like iterate and elevate legislation that works. Um and the second one is you know uh how from your experiences seeing this being implemented lots of different settings what do you think it takes to get institutions and political like you said like the bureaucrats to actually adopt this particip um into their institutions versus just treating this as like kind of like an advisory nice to have role but it seems kind of messy and yeah so love to hear your thoughts. Yeah, for for career

bureaucrats to support this, of course, it must be reducing risk and saving time. This is well understood. Um, for politicians to adopt this and make it into actual legislation, you need a shared urgency. Uh, and then you need a air cover like a political will somehow. So for example, uh, in the sunflower movement case, the urgency is obvious, right? It's economic annexation and people already occupy the parliament. the approval rate was already at 9%. So the air cover is this like anything that produce more than 9% legitimacy is better for their uh career. Right? So at the end of 2014, virtually all the mayoral candidates that run with uh open

government at the time, we didn't call it broad listening with open government uh as their platform got elected and sometimes surprising to themselves. Uh and any mayor candidate that didn't mention open government in their platform loed election in the end end of 2014. And so it says a very clear political signal that then the air cover is provided in a kind of very strong fashion and you have to kind of jump on the spend wagon. We're now seeing this happen in Japan. And so uh without a common urgency, it kind of doesn't happen but it's just exercising the muscle of care. Think of it like a kind of care gym, wisdom gym. Uh so that you you start with like very low binding power um issues. you start with uh like

some chronic harm but it doesn't matter this way or the other and almost like trivial uh issues like in Taiwan uh people deliberated about changing our time zone to the same with Japan uh knowing that it probably wouldn't happen 8,000 people petition this way 8,000 people remain in time zone plus 8 plus 8 and we convene both sides we calculate the cost of time zone change we asked them okay if both of you want to make Taiwan seem more unique how about with this budget let's do it more effectively and then people say oh let's um bring risecon to Taiwan let's give open source contributors uh residencies which we did and many other things and this a fraction of cost was actually changing time zone just to make people traveling from Beijing to Taipei to feel

inconvenient but anyway and so this like trivial issues but you have to keep working on it to win the trust of bureaucrats until such a day where a real disaster happens like a you know bio- risk literally COVID And then all these uh infrastructure get turned into national level infrastructure. And so don't worry about um the legitimacy in the scaling up sense just yet. Just keep working on the grassroots community level. And when the urgency comes like a wildfire and you can show up and say, "Hey, we're ready."

Right? I think we'll end it there. Thank you so much, Audrey. Wonderful. Thank you. Great questions. Thank you. [Applause]