What are the enabling conditions for institutionalizing digital democracy?

I think there essentially need to be two thresholds, one after another. The first is a widespread sense of urgency about something the British call a “wicked problem”—like there’s not a single actor that can solve it in a Pareto improvement kind of way, but rather, it requires everybody to sense-make together.

The trade deal with Beijing in 2014 was certainly of this shape. The first generation of AI manipulation algorithms in terms of Uber, social media, and so on was certainly like that in 2015; the pandemic in 2020, of course, was also of this shape. The reason why is that the career public service usually sees anything that is not urgent, like chronic issues, as something within their domain—they’re the competent authority. However, when an emerging situation arises that demands a faster turnaround than the existing cycles would they consider this kind of lay governance as an accelerator to sensemaking. So, the sheer sense of urgency is the first trigger.

The second threshold is that there needs to be “air cover.” Either the governor, the mayor, the city council—somebody needs to pre-commit to close the loop. It needs to be a pre-commitment; that is to say, no matter what result the people want, leadership should at least be accountable in a sense of design thinking. The commonly discovered urgent priorities and the collectively defined uncommon ground (when it comes to value agreement) demand a pre-commitment to answer through policy development and delivery that specifically addresses the people’s consensus. Again, that doesn’t always happen during times of urgency. It’s very easy to pass the buck, kick the can down the road, or things like that, and basically say, “Oh, we are just following the playbook.” So it requires somebody as a face to provide the air cover for the polity.

In California, we’ve been working on the second enabling condition, which was a law passed by their legislature demanding the state government prepare social media codes for young people under 18. This came from a widespread understanding that kids are kind of being trapped in the social media landscape. While they didn’t want to go as far as Australia (which discussed banning social media altogether), they did want something to be done. Governor Gavin Newsom and first partner Jennifer Siebel Newsom, who very much care about the mental health of young people.

However, on the day of the intended launch, the LA wildfire happened. I was in California, so we did not launch the initiative then. On the other hand, the wildfire provided an even more urgent meta-context, which was not just how to build back better when it comes to policy, but also how to prevent future wildfires. So now, the platform launched with wildfire as the main topic and, of course, got something like 6,000 or more registrations and active participation. The uncommon ground was quickly discovered, and so on and so forth. It’s a success story so far, but I think that is because the first enabling condition (urgency) was met beyond the initial expectation. There’s no pro-wildfire party in California. So that is just to address your question about enabling conditions.

Yeah, so interesting. And that urgency can’t be manufactured, can it?

You could highlight the most urgent aspect.

Because I’m thinking about the climate emergency, which, you know, you could argue is not getting the…

It’s like chronic?

Yeah, it’s chronic, it’s, you know, existential, and yet we haven’t really got the action that we need, or it’s only in pockets.

Yes. So, by highlighting an urgent aspect—for example, for AI governance, which, as we just learned, is all-encompassing—if you ask people, “What do you feel about AI governance?” you’re not going to get very useful results because it’s just too vague. Instead, last January or a few months before that, we had this first ideathon where we asked people to portray their ideal outcome in 2040 in an AI coexistence future. Once people went through that exercise, we then turned around and asked, “What are your priorities when it comes to the immediate concerns you have that block your imagined future from happening?” People then became much more focused, and most people coalesced around online fraud, deepfakes, counterfeit fake persons, manipulation of the public, and epistemic security as the main urgency.

So it’s not that we manufactured that urgency, but rather we put people in the mood of figuring out what the short-term urgency is—the thing that if we don’t solve, we stand no chance of solving everything else after that. Kind of like a blocker in the path. Then, using that, last March, we sent 200,000 random SMS texts just asking, “How do you feel about information integrity online? What should we do together?” That led to a very successful KYC (Know Your Customer) co-created law last May. So, yeah, you can take a chronic issue, do another even bigger diamond process before the first diamond, and just ask people, “What’s your preferred ideal scenario? What’s the ideal climate future?” Then you turn around and ask, “What is the one short-term thing that would block this future from happening?” And then you concentrate on that.

I guess what I’m hearing is that there are different scales of democratic innovation. There are democratic innovative activities, which you can really see in that California case. I guess what we’re kind of playing with is the idea of how you turn that into systemic change and a real, sustained innovation in our democracy. How do you get the whole system change rather than just doing add-ons? Because, since we last met, with our change in government, we have done so much more participatory work (less on the digital side), but there is a huge appetite compared to a year ago, which is really great. There’s a part of us which could go, “We’re very pleased with what’s happening, this is great, let’s all go home.” At the same time, democracy is on fire around the world—definitely not giving this enough political cover. And it’s not… I guess my ambition is that that political cover comes in the form of a story about where democracy can go with people—like an evolving system, yeah, rather than an add-on.

Um, so one clarification though: Engaged California is a general-purpose platform that any ministry within the state government, as well as potentially city-level or maybe even district-level governments in the future, can reuse to introduce their own topics. There are also conversations about enabling, like in Taiwan, a citizen e-petition, so that people who collect, say, 10,000 signatures can kind of force a consultation on the Engaged California platform. So the enabling conditions are for the legitimacy of the platform over career public service. But once the public service accepts it as a risk-reduction tool—and increasingly with AI, also a time-saving tool—then they are more willing to put issues of lesser urgency on the platform, just because the bootstrapping cost has been absorbed by the urgent first case.

That’s a brilliant phrase, “risk-reducing and time-saving,” because so much of the pushback we get here is “people are problematic.” We’re doing a whole load of work around the planning rules at the moment, where we’re making the case: go early, go representative, and you essentially engage the MIMBYs rather than the NIMBYs and the YIMBYs.

Yes, MIMBYs (Maybe In My Back Yard). I love that.

And in doing that, you do all the things that…

Yeah, because in Taiwan’s experience, like 90% of people are ‘maybes’.

Yeah, exactly. And they are here, but you hear them 10% of the time.

Yeah, definitely. So I totally agree. And I think the risk-reducing part is now widely understood because if you keep backpedaling, you don’t go anywhere. The risk of vetocracy is well understood. Although the time-saving part, I think, is not well understood. I always stress this, and it certainly resonated with the Newsoms, but also resonates with Koike-san, the governor of Tokyo. She’s in her third term now. In her gubernatorial re-election campaign, she had a challenger named Takahiro Anno, a 33-year-old machine learning expert. One month before the voting day, he read Plurality.net, a book I co-wrote, and decided to run. He decided to run and basically say, “Okay, this book is my campaign.”

So he used sensemaking tools introduced in the book, like Talk to the City and Polis to crowdsource his campaign. He live-streamed on YouTube 24/7. You could even call into a number to talk to his voice clone and ask questions, or maybe hashtag #TokyoAI to post your suggestion to his platform. This uncommon ground built using these broad listening technologies then became his updated platform on GitHub, and his avatar just kept narrating these updates. His platform became the top independently ranked platform. Koike-san was impressed enough to tap Anno-san’s expertise to join GovTech Tokyo to enable this broad listening as a Tokyo city metropolitan infrastructure—the same idea as Engaged California.

The point here is that they also realized it’s now much easier to close the loop. If you, inasmuch as you had an online conversation or typed in a few words and suggestions, and then the Tokyo Metropolitan Government decides to implement some of them—previously, all they could do was send a newsletter to everyone who participated saying, “Thank you for your contribution. We implemented these five things.” But now it can be extremely personalized, like “because of these words you said, the influence function of your utterance…” and “it’s you here in this spectrum of people represented,” and “that’s how you literally moved the needle,” and “now we’re doing this because of you.”

Closing the loop—not just the sense-making/listening part, but the closing the loop part—used to be so time-consuming. It was impossible to scale, even harder to scale than the listening part. The listening part, at least you could use some sense-making coding and human ‘small language models.’ But closing the loop, like the government literally calling you and thanking you for a specific contribution, simply was not possible. But now it’s been perfected and is almost like off-the-shelf software. So that’s what really drew them in, because if you’re running for governor or parliamentarian, this is how you win hearts and minds from your constituency.

It’s interesting. When you say that about closing the loop, it’s so exciting to hear that kind of application of technology. But hearing you say that kind of direct, “you said, we did”—there’s something about that which could kind of go full circle and become a populist tool. How do you make sense of that? What about when it’s not delivering exactly what you wanted, but it’s found the bridge between what you wanted and what someone you thought was your opponent wanted?

Yeah, the surprising validators. Yes. And I think that is really the point. With language models, one can now identify, in a huge amount of text—a rant that is a personal attack, very toxic or whatever—the few words in it that can be construed as constructive and bridge-making. I always make sure to profusely thank them, sometimes publicly, for this very specific part. This teaches them that the other parts, designed to induce a sense of status harm or polarization, simply have no effect. Only the bridge-making part has an effect.

A couple months ago, I had a hour-long conversation with Laura Loomer, a prominent influencer of the MAGA right. She and her 60,000 live-stream viewers on X and Rumble. We very quickly identified the bridging sentiment as countering authoritarianism without sacrificing free speech and finding this uncommon ground. It’s also quite attractive to her because she was also wary of techno-feudalism. By discovering that uncommon ground early on, I think this surprising validator enabled people to see that there really is something in common, and they become very much pro-Taiwan at the end of that hour.

So, to your question, I think there’s a model around uncommon ground, around bridging, and there’s also another model around the loud extremes. It really depends on whether the space is a broadcasting medium or a broad listening medium. Just as we, in groups of six or ten, would not be very toxic because it simply wouldn’t resonate with the rest of the people. But if you put the exact same people individually in the antisocial corner of social media just to win dunks and retweets, they become much more toxic because there’s no resonance check in place. This resonance check, what I call conversation networks, is what keeps the populist tendency bridging instead of polarizing.

And it’s the difference between optimizing for consensus rather than extremes?

Not necessarily consensus; it could be just bridging.

Just bridging.

Yeah, I think there’s a fine difference here. Consensus sounds like people have to sign their name on it, everybody agrees, like a joint protocol or something. But bridging is more about consent, not consensus. It’s about, “Okay, we can live with it,” right? It’s not that I will sign my name on it, but I won’t sign my name against it. Finding that overlap is much easier than finding consensus where everybody has to agree. So in the sense of the ‘maybes’ (MIMBYs), it’s more like finding where the Overton window looks like this, rather than “we all agree to do precisely this.” Which is why I say it’s in the first diamond—just the shared values—instead of the second diamond, which is a specific implementation.

So tell me a bit more about how you understand populism in the context of this way of working. Is there a kind of populism that this enables?

Yes. I understand populism as a way to bypass the vertical institutions of trust—whether it’s media, university, or the career bureaucracy—and go directly to the horizontal mode of trust, which is people essentially resonating with each other’s vibes. I hope that jives well with the British understanding of populism.

In a sense, populism, escaping vertical trust, is a direct consequence of people feeling closer together on digital platforms than through any intermediaries. Any intermediary, whether it’s BBC or CNN, that tries to do interpretive work necessarily relies on the broadcasting capability of vertical institutions that reach millions. Populism is less about that; it is more about finding a very short, resonating message that lets millions of people kind of horizontally disperse it as floating signifiers.

I’m just processing. The big debate in this country at the moment is all around populist tactics.

Yes.

And how, I think, the political system is looking around the world, seeing incumbents not winning, and feeling like they need to be disruptors—even in government, to be seen to be disrupting the system and not part of it. And in that, I guess some of the discussion at the moment is whether it’s possible to be a disruptor in government without ending up adopting some of the negative tactics as well.

So just moving in different kinds of political positions. Rather than… one of the challenges I hope to address with this project, one of the questions is: how can you make this idea of a more collaborative democracy, of a more sense-making, bridge-building, inclusive democracy… how can we make that a disruptive idea? Because I think it is interpreted as something kind of nice and soft and inherently kind of woolly and liberal, in a way that is counter to that disruptive narrative.

Yeah, and I think that is partly because people associate deliberation, with good reason, with taking the time needed. People instinctively think that destruction is easy and coming to consensus is hard, so it takes more time. In Japan, when we say “broad listening,” we always couple it with the idea of “agile governance,” which challenges this and says, actually, broadening now with AI sense-making is even quicker than parliamentary debate. So by portraying this as a higher bandwidth, lower latency democracy, it becomes… you can reuse the DOGE argument. It makes the government more efficient and also more effective.

Yeah. That’s so interesting. And it speaks in a very visceral way to MPs here who just feel overwhelmed. You know, I’ve met so many new MPs who were elected last year who are just so overwhelmed by the scale of correspondence, by the scale of legislative bandwidth they need to have. That’s a very powerful sell.

Yeah, I was interviewed by Rory Stewart, with Alistair Campbell. Before the interview, I read his longish op-ed about his experience in government and why he ultimately decided to step out. I think bandwidth saturation, although he did not use that term, was the main theme of that op-ed. So yeah, anything to reduce the cognitive burden, I think, by definition is good. Whereas Obama, when he was first president, recruited a team of human ‘small language models’—staff—to sort through all the letters to the president, choosing like five representative samples, like a medley, for him to read every day. You can easily think of sense-making tools as not just making this process zero friction, but also making writing back not just to those five lucky ones, but rather to literally everyone.

Has Carl talked to you about Waves?

Not yet, no.

So this is the project we’re launching that I mentioned, in May. It has an AI application which is about processing the evidence and the sense-making piece.

Very good.

And then it has a kind of scaled, with a video element, deliberative aspect to it. And the “waves” are the waves of deliberation.

As in video conference online?

Yeah.

Oh, very good. Okay. Yeah. I’ve been working with Lawrence Lessig of Harvard on Frankly.org, which used to be called Kazm. It is now an entirely open-source, out-in-the-open clone-ish version of the Stanford deliberative platform that we used in Taiwan. And so, yeah, because now any open-source project benefits from what’s called “vibe coding.” I added Spanish and Mandarin translation just by talking to my computer and saying, “Oh, let’s add translation.” And then after 10 minutes, it adds the translation. So I think it’s very good now.

Could you have imagined that 10 years ago?

No, not at all. Not at all.

This is a really extraordinary time. How many years ago would you have not been able to imagine that?

I think five years ago, around 2019-20, was the time that we really figured out scaling laws really work. So I visited OpenAI before the Anthropic fork. Jack Clark and others were writing very publicly about that, I think around 2019. So that’s when I realized that, oh, we’re eventually getting to this point. But 2015? No. I don’t think anyone expected that to come within a decade.

It’s so funny. I talk to a lot of MPs who are using AI to draft responses to their constituents, but they’re not thinking about how to use AI to get across the incoming.

Exactly. It’s all about broadcasting.

Broadcast rather than listening.

Yeah. That’s interesting. I think broad listening, of all those memes, has the most political potential to me, maybe not just in Japan and California, because that speaks to a real pain point of anyone working in politics today. Because the petitions, the people’s ideas, are now all AI-assisted or even AI-generated. If you don’t use AI, there’s simply no way to consume that fire hose.

Yeah. How are we doing for time? I’m fine. Anything else you’d say, just going back to that first question around the enabling conditions? It’s a very clear answer: the urgency, the air cover. I kind of like, just coming back to that idea about scale of ambition…

It’s like the Tokyo Olympics 2020. They changed the Olympic slogan, which used to be “Faster, Higher, Stronger,” to “Faster, Higher, Stronger – Together.”

Yeah, exactly. And what I find inspiring from everything I’ve heard about what you achieved together is that it told a different story about the kind of democracy Taiwan was. What do you think gets you to that tipping point of getting political buy-in and the urgency to make it more than the sum of its parts—part of the story about the country you are, as well as the stuff that the government does?

Yeah, I think one big part of it is the intergenerational pact. I was introduced as a young reverse mentor to the cabinet when I was 33. At the time, the central government institutionalized this idea of people under 35 mentoring cabinet ministers about where young people want to go. This was because, again, they were at a point where intergenerational polarization threatened to tear the fabric apart. At the time, I think among young people, the KMT president had a very low approval rate. Nationally, it was 9%, as we’ve discussed.

The point here is that instead of saying, “The young people have demands and the government should please them,” it flipped it on its head and said, “The young people have directions and the senior people should follow them and provide them with resources.” That turned out to be the Pygmalion effect; it was the only way to restore trust because the young people really did have better ideas of where to go. The career public service then started to see those young people who joined as reverse mentors as comrades who were in it for the long haul. Instead of just demanding, we moved to the supply side. Instead of just opposing, we moved to proposing. One of the most lasting changes, which we worked on together before the Uber case, was to design the curriculum with the children, not for them. A generation of children in Taiwan now grow up knowing, before they turn 18, that they can participate fully in national agenda-setting, and maybe even become cabinet-level advisors before 18.

So in 2022, according to the International Civic and Citizenship Education Study (ICCS), Taiwanese 15-year-olds were top of the world when it comes to the feeling of agency, belonging, and care they feel around planetary and people issues. This horizontal trust, I think, was the most enduring legacy. The other parts kind of followed, like we are now among the world’s lowest in terms of social, ethnic and religious polarization. So no matter what political party polarization plays out, the underlying fabric of trust has been rebuilt and healed.

What’s the best data on trust that speaks to that? What’s the source of the data?

Yeah, so the Plurality book cites rigorous data on all these, especially affective polarization.

Which chapter?

Chapter 2-2. Economically, we’ve also done very well, and the Youth Development Administration within the Ministry of Education tackles issues like social housing, mental health, and many other things that young people feel are worthy of national agenda-setting. The young people themselves are the facilitators who conduct the broad listening countrywide. Every year they set an agenda, so there’s plenty of data and statistics, I think especially around the social housing one. But I assume loneliness and mental health data would also be useful here as well.

Very useful, yeah. Brilliant. That’s so helpful. I kind of feel like I want to go and reread the book to draw out the evidence. I don’t remember the individual pieces of evidence now. What’s the most compelling evidence, from Taiwan or elsewhere, for this approach for the government?

Well, I think the World Happiness Report. We’ve been the happiest in East Asia for two years now. That was also why when I was in Finland, Sitra (Finnish Innovation Fund) people really wanted this connection, because Finland is, of course, the happiest overall. They also want their process to expand across the Nordic, so they can shape policies like AI governance using Polis-like technologies, because they also see happiness fundamentally relying on the social fabric.

When they ran their nationwide “What do you think, Finland?” consultation, that enabled them to very easily do a horizontal scan on upcoming polarization points. For example, I remember the most polarized point was older generations feeling that the language they were taught and used throughout their lives was now being abandoned or even frowned upon by the younger generation, and they were tired of getting their words corrected all the time. I mean, obviously, they’re the happiest if that’s their most polarizing debate.

But I think this continuous horizon scanning is another model that we really like because then it doesn’t need to be urgent issue after urgent issue. Rather, exactly as you said, it could be part of the deliberative system that enables, for example, their city councils to just read what the most emerging polarizing issue was in the past month and handle that every month, kind of like a post on their polity.

This is what Google Jigsaw is now doing in Bowling Green, Kentucky, WhatCouldBGBe.com. That used this idea of committed listeners to the people’s wish in Bowling Green without a single triggering condition.

Have you just heard the term epistemic security from Elizabeth?

Yes, in the Minderoo context.

That idea is getting more political traction. I think there’s more comfort in thinking about it in those terms at the moment.

That is really good. And it connects to the idea of epistemic injustice and how to correct that, which is, I think, a very good theoretical idea because it’s like lay governance, right? Just because people lack certain training and certain use of language doesn’t mean that they don’t have the most kind of “rare earth” of lived-in experience to contribute. In a sense, they were suffering from epistemic injustice from the existing democratic institutions. So epistemic security is about enabling them to feel more connected and thus more secure. I think that is a much more positive frame, more like a life, liberty, pursuit of happiness claim, instead of counter-misinformation.

And because counter-misinformation is such a… I think it’s one small part of something, and it starts to… It’s very judgmental of the sources of that information.

It’s very vertical, elitist.

Yeah, yeah. And people… That is losing ground now. The term I used years ago, thinking about this, was “unnewsed”—like “unbanked.”

Yeah, That.

So the term unbanked is for people who don’t have a bank and therefore can’t access the economy. If you’re unnewsed… I was a journalist at the time. Now I think about it more as, you know…

Yeah, epistemic inclusion, like financial inclusion.

And in terms of how democracy is functioning in Taiwan now, has it fulfilled your wildest dreams?

Oh, yeah, definitely.

Would you want it to go further towards direct democracy?

I think we’re at a point where referendum and recall, informed by sensemaking, are already very close to direct democracy. I think we’re at a point where MPs in Taiwan do know that they kind of have to work with these horizontal power centers. We had a strategy of introducing young people to the Fediverse, specifically Threads, but any place on the Fediverse would do. We offered real-time responses on Threads. So on Threads there are now more Taiwanese people than people of any other country. We’re now the largest segment.

I think that also shows it’s not limited to Taiwan. If Taiwan by itself goes even more frontier, and the rest of the world kind of backslides, then we’re in a very dangerous place. Because then democratic innovation and renewal would be seen as something odd, and antisocial extremism would be seen as the norm. Then everybody would just be scrambling to defend against civilizational collapse. Which is partly why, instead of focusing only on Taiwan, I’m now focusing on, as you know, Tokyo, California, Finland, and other places.

So that’s fascinating because, I mean, Trump is on one hand rejecting the whole world and on the other hand, really building a political movement across the world. Do you see it as your mission now to do similar with Taiwan?

I think in a sense, yes, with Taiwan, not necessarily for Taiwan.

Yeah, I think you always do with .

That’s right. And yeah, kind of horizontal leadership, right? Also, I think fundamentally there’s something in the MAGA movement, which says we need to address this “unnewsed” phenomenon, not by moderating speech that offends, but by giving voice back to local communities. Instead of simply saying, “You’re unnewsed, so we’re taking away your speech, we’re moderating you, shadowbanning you.” So I think that the pluralism ideas and the MAGA right kind of have this one value in common. And that is what I’m also talking to the investor of BitChute, hopefully also reaching out to Truth Social (which is also open source), so that they can connect with the Bluesky part of the population.

Because the criticism of Bluesky and everyone moving on, or kind of everyone of a political persuasion moving onto Bluesky, is that they were ghettoizing.

Right, exactly. And by building across the Fediverse this kind of pro-social bridging, then you can have your community, you can have your safe space. But then when the surprising validator comes, you also won’t miss it. So we’re building an alternate relay for Bluesky at the moment to enable this kind of cross-pollination with Truth Social and so on.

When you use the term bridging, do you mean it specifically in the social capital bridging term? Or do you mean it more broadly than that?

So I mean it in the bridging-as-opposed-to-bonding term. And also I mean bridging algorithms like Community Notes that compute the diverse groups with diverse voting patterns, like the left wing and the right wing. Then the bridging system discovers the uncommon ground—the things that are upvoted by both sides. That is the content that has the most bridging capital in this system. So it’s kind of an applied way of using bridging capital.

Because one of the big, like where this kind of segues into social capital theory for us, is that, you know, you see the bonds of trust crumbling all over the place, except in very small groups in this country. So family groups, community groups, those. And that, you know, that for me feels like a source of hope, but also a source… like, to not be kind of starry-eyed about that, because it’s also about atomized communities.

So there is less bridging. And there’s also internal polarization. If a person whose spiritual community, professional community, family community, and so on, only have this person in common, then in this person, there’s also a fragmentation, a balkanization of the spirit, so to speak. That is what we also try to formulate in the new paper, “Prosocial Media,” what we call coherent pluralism. The idea is that just as advertisers would like to pay for brand-building opportunities like Apple during the Super Bowl or something like that, where everybody shares in a common experience (which is increasingly impossible in this fragmented, itemized, individual precision targeting of advertisement and social media that strip-mines the social fabric), now people would also like to kind of sponsor content that heals the divide between their communities. So that if my spiritual community and my professional community see this piece of content that shows to me that both communities are enjoying it, and it’s a piece of content I can then use to bridge my two communities together, the theory is that people are willing to subscribe, to pay for that sort of thing. It provides a different revenue model as opposed to the current strip-mining, individualized advertisement auction model.

Can I ask you, is this kind of going slightly in a different direction? But I want to ask you about the kind of cultural change that’s needed within the government to achieve what you need to do. We published some work a couple of weeks ago about the UK government’s missions. They’ve got five missions, they want to deliver them. On many different levels, these missions are not getting started as a way of working. But I think the fundamental bit that’s broken about it is that as a government, they only know how to do vertical. They don’t know how to provide a platform or build the conditions or kind of enable other people to do it. This work was looking at, from a business, citizen, and civil society point of view, kind of like, what could the government do? And then there’s another layer to it that even my team, when they were saying, “What can a government do?” they reach for policies, rules, or things—

Levers.

Levers. Everybody wants a lever. Whereas actually what’s needed for this to happen is, every time two people sit together, they do something different than what they’re doing now. I feel like the challenge is in the nature of expectations of vertical rather than horizontal relationships and the culture change that’s needed to do that differently.

Yeah, I hear you. And I think part of the reason why Taiwan works this way was that, first of all, we collect, I think, one of the least taxed percentages of GDP in all OECD countries. So we simply don’t have that sort of top-down vertical resource. Everything we do has to rely on collaboration, especially with the private but also the social sectors. We always think in terms of incentive design and things like that. We don’t think, “Let’s build this data center over there.” We always think, “Oh, let’s invest $10 billion that we just earned from TSMC into the next round of domain-specific AI models with capital, private investment, and equities from the private sector.” So that’s kind of our default mode of policy action.

And it’s always been like that?

It’s always been like that. I mean, TSMC wasn’t state-owned, right? But the state created the conditions for its success. And so I think this… So in the UK, like the widely studied GDS (Government Digital Service) movement, I think it was also like that, but that was in a time of austerity.

Yeah.

Like just because, you know, post-financial crisis, nobody had the budget to build their new IT infrastructures. So they trusted GDS to do the design systems and everything. So I think some sort of frugal innovation frame is good as an enabling condition.

We can do that.

And without which, I don’t think it’d be felt as urgent. People would just say, “Oh, let’s put some of those extra resources into some vertical thing.” But if you say, “We simply don’t have the resources to do all that, and now we have to work across sectors,” I think the bureaucrats do understand that there’s simply no way to fulfill all the missions using just top-down methods.

There are definitely people in government that understand that’s the position they’re in, but they don’t know what that looks like in practice or how to do it. Or, you know, the muscle isn’t there. What could we do to support them to grow that kind of enabling muscle or instinct?

So I thought GDS is a good story, but maybe that’s too long ago?

I don’t know, because also GDS is quite kind of centralized.

Yeah, it really is.

And I think thinking about some of the mission delivery, it needs to be really kind of letting go at the centre and almost giving permission. Like, there are councils around this country that are doing the most extraordinary horizontal work with people. I did a lot of work in children’s social care. So when families aren’t working out and the state needs to step in, there are councils around the country doing extraordinary work to invest support for families and to give power to families to find their solutions and kind of not be done to. This is in a system that has traditionally coerced families into behaviour and then, when they failed, removed their children. It has been a brutal and very difficult system. But the councils that have done that differently are few and far between. They’re proven, they’re saving money, they’re saving lives, their families are flourishing in those contexts. But spreading that practice… There are so many blockers in people’s imaginations or feelings of empowerment to make that change. For those councils I’ve spoken to, it has always come from leadership creating the permission for everyone to innovate and take risks. And most of this country is so risk-averse.

Yeah, and I’m surprised that you would use the “risk-taking” frame. I mean, 10 years ago, certainly you were taking on some risks if you tried broad listening. I think in this day and age, certainly the way we sold it in California, it is sold as purely a risk-reducing move. In the sense that all the technologies are mature. Not just mature, they’re open source. You can really audit them. And we enable people who are on the ground—the Eaton and Palisades people, the Department of Angels, the social sector, and so on. So we’re not bypassing the existing self-organized civil society organizations; we’re enabling their voice to be amplified, connecting them with other conversation networks. By, first of all, working with the existing civil society organizations and city landscape, and second, by saying we’re just offering tools to save you time and reduce your risk, I think that is a kind of Pareto improvement ticket. Instead of “you’re taking on some risk to try something that may save you more time downstream,” it is “something that saves you a little bit of time and reduces a little bit of risk, right now.” And I think that is much easier to introduce, especially to city councils.

Who’s inspiring you at the moment? What inspires you? Coming back to the original question around democratic innovation around the world. It sounds like you are just on an amazing global odyssey, all over the place learning about what’s going on. What’s inspiring?

Well, there are many things inspiring me now. I do think that Tokyo and Japan in particular—Japan generally, and Tokyo specifically—is going through a remarkable phase of renewal because they have all the enabling conditions. Their young people are thoroughly non-partisan. Because they have a shrinking population, like a reverse pyramid, the young people also feel there’s no way they can outvote the senior people because there just aren’t many of them around. They increasingly feel that they have all the right directions figured out—they’re very smart—but the existing political apparatus simply does not provide the cover to listen. So I was speaking to the third-party leader in Japan who really got this prominent position now because neither of the two main parties has an overall majority. So they’re the kingmakers, similar to Taiwan in the constituency sense. And he completely endorsed the “Anno stack,” as we call it. So it has the political conditions now nationally, as well as in the city realm.

The other enabling condition is a little bit like the UK. Japan has the most amazing small community-level citizen participation governance, like community practice. About a decade or so ago, the idea of mini-publics and citizen assemblies took root in Japan. So now you can find, not a majority, but a significant amount of neighborhoods that have tried this kind of sortition-based democracy before. People are not afraid when the central government introduces something like this, because some of them have now, either as jury duty or as citizen assembly duty, understood that peak experience.

Do they have a citizen assembly duty then?

Yeah, in some neighborhoods, not all, but certainly more than any other country that I know of. So I’m very excited about Japan right now.

That’s really exciting. We must talk about that tomorrow with the MP. Was it Japan or was it California where they’re using the voice clone?

It was in Tokyo. Yeah.

Tokyo, yeah, with Anno-san. I think digging into that example tomorrow will feel quite tangible to MPs; they will understand the application.

Yes, yes. I think for anyone who wants to run for re-election, this really is a net plus. There is nothing to lose because you have to handle all those citizens’ campaigns and petitions anyway.

Is there any other type of democratic innovation that is beyond digital democracy and deliberative processes that’s inspiring you at the moment?

Well, so I have a recent preprint paper with Deb Roy and Larry Lessig about Conversation Networks. Part of the idea here is not an infrastructure-level play, as we just talked about, but rather just to enable more small group conversations—not necessarily in a binding way, but making it kind of the norm, for example, civic classes in education, for youth networks, and things like that. What we were just talking about are decision-making technologies that translate into collective action that somehow closes the loop. But there’s also, I think, a lot of realization now in the U.S. that simply listening to people on the opposing side and making it a habit, almost like just sharing food together, is what a country desperately needs now. So I think this very, very small-d democracy is not about policy at all—none of its output is policy.

It’s about relationships.

Right, exactly. And the healing of the social fabric. I think that excites me more and more. Especially now, I think many people feel… It’s like jury duty. You enter feeling very difficult, you have to set aside a lot of time, but you sometimes come out of it feeling reinvigorated about society. If we can deliver this small delight, I think that is also a very good part.

Did you read a very, very, very long piece in The Atlantic called “The Antisocial Century”?

I have no idea.

Yeah, it came out a few weeks ago.

Okay.

And I think it speaks to that. It’s describing an emerging generational trend of people feeling like other people take energy away rather than give you energy. It gives examples like the rise of noise-canceling headphones where people feel distractions are hard. But there’s an amazing experiment in this article where they asked a group of citizens—I can’t remember which country—“Would you like to get on the train carriage where everybody is talking to each other or the one where nobody’s talking?” Every single person said, “No way, I want the quiet one.” And then they split them into groups. The ones that went on the chatty one came out feeling completely uplifted and engaged in the world. I think that kind of difference in the idea of feeling that coming into contact with people will somehow cost you something or take things away, rather than add to the sum of your experience… And of course, that’s different for different people at different times. But I think there’s so much value in that.

Yes. Pro-sociality. I think for people who are like me, who are bitroverts, we have a choice, right? We can get energy from focusing; we can get energy from uplifting conversations. How to balance the diet, like nutrition labels for interaction—I think that is an important skill to have. Indeed, maybe in this vibe-coding AI era, that may be the main thing that gives humanity meaning now. Because the previous meaning centers—being able to cognitively perform well, to answer standardized tests well—become like… I don’t know, like the jobs that take input from one machine into another machine. These kinds of jobs, like coding, are going to be completely replaced one way or another. The sooner we can move people’s meaning center from outracing a car to that of building meaning across people, the social fabric, and the sooner we can say AI should be off-limits when it comes to replacing people-to-people ties… We shouldn’t all just talk to our chatbots and send them to deliberation. It’s like sending our robots to the gym to lift the weights: very impressive, but no civic muscle grows this way.

So that’s an AI application you don’t like.

Exactly. To me, assistive technologies should be first and foremost like eyeglasses that improve human-to-human connection. Or they could be augmentative. I got laser-blending vision now, so the eyeglasses are etched into my eyes.

Amazing.

It’s really amazing. I now see very clearly near, middle, and far fields, which is a little bit transhuman because naturally, eyes don’t grow like this. But at the end of the day, this is not replacing this experience with your avatar uploaded to the cloud and projected into my retina with flying ads.

I talk to many people in government, new people who have gone into government, and I feel like they are experiencing that feeling of being so overwhelmed by the responsibility and just by the pressures of the job that you can see them wanting to shut things out rather than open up and listen. And you can see that because she was so overwhelmed. Did you meet any blockers like that along the way, politicians who were just like, “We don’t need to”?

In a sense, right, there’s this expression, “a minister’s schedule.”

A minister’s schedule.

Yeah. It’s always packed back to back. And I think for the career bureaucrats, which I keep going back to, it is also a matter of upward management. Like, how much do you want the minister to process? It’s like you have to really pay attention to the traffic when driving the car because as the driver, you have the full responsibility if there’s a crash. So the dashboard, so to speak, tends to be all over the place because there’s a tendency for career public service to essentially manage the minister’s attention such that anything that could potentially be political fallout is at least kind of in the ambient vision of the minister. Because otherwise, the career public service is to blame, and no one wants that.

My system with the career public service was entirely the opposite. I publicly said in 2016 that I entered as a servant to the public service. At the end of the day, I wanted this to be a self-driving car. I wanted the bureaucracy to be able to very clearly sense-make with the citizens, not just for the citizens. So in a sense, as a minister, I had the easiest job (well, until the pandemic). But anyway, the easiest job because I just held the space. I did not actually need to be personally aware of each individual detail going on, but I could go back to the radically transparent reports whenever I wanted to. So I think this nudge thing is a good first step. Because if we can get the nudge to be not just about a policy measure, like turning off electricity to save energy, but rather something that is like the Ice Bucket Challenge that goes viral on its own, then in a sense, the minister doesn’t have to do that much work. The minister would just hold this conceptual integrity, and then everything else, the implementation part, could be figured out by the career public service in association with civil society.

I’ve just realised we’re up for time. Thank you so much for this.