2026-05-14 The Solve Effect (MIT Solve)

Welcome to The Solve Effect. I am your host, Hala Hanna.

What if you were incapable of getting angry because your heart literally could not handle it? Would that forced stillness shape the way you move through the world? For today's guest, that is her origin story.

Audrey Tang grew up with a heart condition so serious that any strong emotion could land her in the hospital. She learned to stay radically calm. The heart condition was corrected at 12. The equanimity stayed. In 2014, Taiwan's Sunflower Movement pulled her back home. Three hundred thousand students occupied parliament. She left Silicon Valley and livestreamed what she clarifies was not a protest, but a demonstration of an alternative way to govern.

Two years later, Taiwan's government invited her to become its first digital minister, and the youngest. Tang served for eight years as the world's first openly non-binary cabinet minister. She made broadband a human right. The COVID response she helped build is one of the most praised and studied. She helped move public trust from single digits to about 70 percent. She even made Taiwan start school an hour later. And she did it all in public, posting every meeting, every transcript, every decision.

She is now Taiwan's Cyber Ambassador, and she is, by her own account, a civic hacker, a conservative anarchist, a poetician, a Taoist meditator, and my personal favourite, a hopemonger. Audrey, welcome to The Solve Effect.

So happy to be here. I travelled all the way from the year 1981. When I was five years old, the doctors told me there was just a 50/50 chance of surviving until surgery. Every night was like a coin toss. I am happy it landed the right way so we get to meet now.

What does it do to a child to spend 12 years learning, in such a visceral way, that life is a 50/50 toss?

It is an existential opportunity, as it were. I got into a habit I call "publish before I perish." Every night before going to bed, I would document everything I learned that day: first on cassette tapes, which some listeners may not remember; then on floppy disks, large ones and then smaller ones; and finally on the internet.

I learned something very important. I did not have time to be perfect before publishing. On the internet, if you publish something perfect, people just like it and swipe away. But if you publish something vulnerable, something half-baked, people say, "Oh, that was wrong," and then I make friends that way.

There is a crack in everything, as Leonard Cohen said. That is how the light gets in.

So imperfection is an invitation. Do you still practise that today?

Definitely. Whatever I publish, I think of myself as a good-enough ancestor. If I am too perfect, I take away the chance for the next generations to adapt it to whatever they care about. But if I am just good enough, I leave them with a wider canvas and some materials, often under the MIT License or Creative Commons, and they can reassemble them however they want. Not foreclosing future possibilities is still very important to me.

That is very empowering. Permission to be imperfect.

That is right.

Knowing that your own extinction was near gave you the urge to share rather than accumulate. How do we make that work for others? I am thinking of the tech industry today. How do we wrestle accumulation away from the accumulators?

There are many places where the more you share, the more you have. Language, for example. If you monopolise a language, it dies. It depends on people sharing it.

Open-source commons, the internet itself, and many other things have what economists call anti-rival-good mechanisms. The challenge is to think beyond the scarcity mindset. In AI, for example, instead of thinking of it as an all-pay auction, where everybody spends billions and only the winner gets anything back, we can think about small models that run locally.

Instead of a giant model that costs a huge amount of electricity and folds our proteins, folds our laundry, and folds many other things into one giant black box, including Studio Ghibli, we can train specific models for translation, summarisation, and the connective tissues of civilisation. Then AI becomes the soil we till instead of oil we drill.

That is also Adam Smith's specialisation: each thing doing what it does best, rather than the vision we are being sold now, which is data centres covering the entire Earth.

Exactly.

You were reading classical literature by four, writing poems by four, coding at eight, dropped out of junior high at 14, were already collaborating with Harvard and Stanford researchers, and built your first startup at 15. Are you trying to make us all feel bad about ourselves?

I think it shows that very young people can steer the direction of civilisation. This is what I call the Pygmalion effect.

When I was 33, I was invited into the cabinet, not as a minister, but as a reverse mentor to a minister, so young people could tell ministers what they missed. Ministers have cached ideas of how things work. We can clear their cache by being young.

When I became minister at 35, I became old, so I needed my own junior reverse mentors. Some of them were not even 18. One young person, Wang Hsuan-ju, started an e-petition calling for a ban on plastic straws for our national drink, boba, and got it. Other young people crowdfunded what I believe was Asia's first museum for menstruation, to normalise it and remove taboo. Others petitioned for school to start one hour later, because studies show that one more hour of sleep gets you a better grade than one more hour of studying. Imagine that.

Young people challenge the cached stereotypes in our minds. That is what I modelled. The fact that I was 14 did not prevent me from doing a startup. Officially, it is on record as 15, for labour-law reasons.

You also describe yourself as a competitive sleeper. What does that mean?

I sleep competitively for eight hours every day. If I wake up in the middle of the night and it has only been six hours, I think, "I can still get two more hours." If a great idea appears, I tell that idea to my local, bounded agent, a Kami. The one I co-cultivate with Tenzin Yangtso, called jdd-kami, goes ahead and works on it while I sleep.

It is Ricardian comparative advantage. It is a better coder; I am a better sleeper. So we divide the labour.

And that is stored locally?

Literally on this laptop.

Are you going to open source that?

It is already open source. I drafted a setup guide called pi-ds4, DwarfStar 4. People are already finding that it runs on a MacBook Pro, an older Mac Studio, or a DGX Spark. You do not have to send anything to the cloud. Every computation is auditable. You can repeat it, replay it. It is entirely here.

You found community online very young, where you could be whoever you wanted to be included as.

By being wrong on the internet.

Publicly wrong, and being 15 instead of 14. What were your parents like?

My mom created the experimental education school for my younger brother, and she also co-founded what is still Taiwan's largest consumer co-op. My dad headed Taiwan's first community college and teaches philosophy. Both of them were trained as journalists, so I have nothing but filial piety toward fact-checking, balancing, bridging, and truth-finding.

That taught me that it is not about competitive advantage to accumulate social status, money, or whatever. None of that makes sense if the social fabric is gone. If we do not have common knowledge, whatever status you have is in a bubble, and it is profoundly unstable.

When I was born, Taiwan was under martial law. It was extremely centralised: no freedom to form new parties, no free press. We saw how democratisation added creativity to everyone who associatively formed civic, spiritual, and other muscles — what I call civic muscle. Still today in Taiwan, spiritual organisations have very high legitimacy. The more highly educated you are through the democratisation process, the more likely you are to practise openly. I am a Taoist, for example.

That is probably the other way around in the United States, where liberal and secular often correlate very highly.

Now I understand more about your parents as journalists, the truth-seeking, and the history of Taiwan when you were growing up. When the Sunflower Movement happened in 2014 and students occupied parliament, you wrote to a colleague in California that you needed to leave.

I said I needed to leave now and set up an ethernet connection. I personally hauled a 350-metre cable into the occupied parliament so people could see what was going on from the street. Again, it is about creating common knowledge, making sure the fog of war and the very high PPM — polarisation per minute — on social media can be cleared away.

People really care about our shared future. People want to affect their future. It was just that closing the loop felt almost impossible, so people turned cynical. They were radicalised. They became extreme on social media, whose algorithms were just beginning to amplify engagement through enragement.

By getting people to tables of 10, to listen deeply to one another and find uncommon ground, people discovered that co-creation is more fun than dunking on each other. That created a shared peak experience.

In 2011, the Arab Spring began. Was there anything you borrowed from that movement? And why do you think Taiwan did so well after Sunflower, when the score elsewhere has been more mixed?

The worldwide Occupy network, where each Occupy sent people to other sites, was an early prototype of what some people call stigmergy: radical self-organisation on many levels. We learned from each of them.

During the occupation in Taiwan, we called ourselves not protesters, but demonstrators. We demoed tools: Loomio from Occupy Wellington, Polis from Occupy Seattle, and many more. Taiwan is one of the rare places in Occupy history where, because of these civic-tech listening tools, we converged day after day instead of diverging.

In other occupied sites, it became much easier to manufacture counter-power than network power, to use Manuel Castells' analysis. In Taiwan, network power ensured that we had a campfire. People with differences had their faces illuminated by bounded civic tech, instead of everybody being pulled away and atomised into the global social-media wildfire, shadowboxing.

Say more about shadowboxing.

The engagement-through-enragement algorithm is not specifically about rage. It is about maximising the attention you spend on the touchscreen. It just so happens that amplifying our tiny differences is much more engaging than amplifying what we have in common.

The literally careless algorithm figured that out, so we spent our time doomscrolling and dunking on each other. But it turns out this is easy to fix. You just change the sign. Make the overlap viral instead of the outrage viral. That is what we prototyped in Taiwan, and it now powers things like the Community Notes algorithm on major social media platforms, and related collaborative-notes systems, where Grok now tracks bridging notes and attaches them to viral posts that lack context.

In the early 2010s, you went to Silicon Valley and asked to be paid in Bitcoin. Very smart move.

One Bitcoin per hour.

Do you still have them?

At the time, people did not have Bitcoin accounts, including Apple, where I worked as an independent contractor in 2010. So everybody converted to fiat. If you look at GitHub, you can still see my old Bitcoin wallet, but because it was too ahead of its time, people just paid in fiat.

Sometimes you can be too early.

But you were inside the machine when it was being built. Were you noticing even then that people were not paying enough attention to the early signs of PPM?

Definitely. What people naively believed then was that it was a matter of media literacy or data literacy, and people would educate themselves. But that is not true. It is far easier to hijack people's reward model in the brain by cooking slop — addictive junk food — than it is to cook Michelin-star information nutrition.

People can maintain an information diet during the daytime, when they are facing an interesting person at the same table. But closer to sleep time, maybe not. That is why I have advocated, and practised for 10 years, turning every screen greyscale: a colour filter that filters out about 70 percent of the colour, leaving only a little. Reality stays more vivid, and then it is not a matter of willpower.

That is a great tip. I am going to do it now.

When Taiwan offered you the role of first digital minister, you famously wrote a poem as your job description. It is as much a prayer as it is an incantation. Would you read it, and tell us how it manifested in practice?

Of course. In Mandarin, in Taiwan, the word shuwei means both digital and plural. So I was also the first minister for plurality. That got me thinking: if singularity can become plurality, how do we apply that as a steerable lens to the rest of what Silicon Valley had to offer at the time?

This was 10 years ago. It goes like this:

That was the poem, and I think it worked out very well for many people, especially in policy. Policy leaders need proofs of concept, demos, demonstrations. We cannot just say, "This is bad, let's ban it." That whack-a-mole is very taxing. We need to show: here is where we are going, here is how I am going there, and here is how all of you can help me. That is leadership. Without a positive vision, there is no positive action.

As His Holiness the Dalai Lama has said, under the greatest adversity there is the greatest potential to do good, not just for oneself but with others. Somewhat fortunately and unfortunately, every other year Taiwan faces an existential crisis. During Sunflower, it was the sudden passing of a trade deal with Beijing that would have opened our telecom, publishing, and cybersecurity sectors to Beijing investors. We overcame that. Then came the crisis of online rage, or the infodemic. Two years after that, literally the pandemic, COVID-19. After that, synthetic intimacy and deepfake fraud.

There is no shortage of challenges we need to overcome with the spirit of plurality. With deepfake scams, for example, we sent 200,000 text messages around Taiwan asking: what will we do together? Thousands volunteered, and we chose 447 people randomly by lottery, mirroring the population. They came up with ideas: if you post ads online, you have to digitally sign them, otherwise they should display "probably scam," like a cigarette label. If the platform posts a probably-scam ad and people lose millions, the platform has to pay those millions through joint liability. If a platform without a Taiwan office ignores lawful notices, then every day it ignores us, we slow down connection to its video by one percent.

These were not my ministerial ideas. They were people's ideas. More than 85 percent agreed, and the other 15 percent said, "Yes, we can live with it." We made it into law, and throughout 2025, deepfake scams went down by more than 94 percent. The point is that AI served an assistive-intelligence role. It did not replace people's judgment. It was basically a glorified chess clock that brought people together.

And you have been able to enforce those laws on the tech companies?

Yes. To my knowledge, Taiwan is the only jurisdiction so far that combines joint liability with KYC.

Is this part of the advising work you are doing with governments now?

Yes. As Cyber Ambassador, "cyber" is for cybernetics. Cybernetics comes from the Greek kybernan, to steer: someone who steers a boat. In Silicon Valley, you hear the doomers saying we should invest in a bigger brake, and the accelerationists saying we should invest in the gas pedal so we do not lose the race. But both miss something important. If a car does not have a steering wheel, it falls off a cliff. That is maximum acceleration, but I am not sure it is the best acceleration. First, we need to invest in a bigger steering wheel.

You are doing this work with California and Utah. Can you say more?

In California, with the strong support of Governor Gavin Newsom and First Partner Jennifer, we worked together on a platform called Engaged California. Anyone can join. A consultation is going on right now. If you are a California resident and you feel AI impacts your work — which is, I guess, everyone — then you are the relevant public. You can sign up and share your hopes, fears, and so on.

They will use sortition to mirror the larger California population and do a real deliberation on what levers the governor can pull on the frontier labs in California. A bill is making its way through the California legislature that would institutionalise this, so it is not just one good idea from a governor who co-wrote Citizenville, but a permanent part of the civic picture.

This is your vision of democracy as a social technology whose tools can be updated.

Exactly, because the bandwidth is very small right now. Every four years, each of us uploads maybe one bit if it is a referendum, maybe two bits if it is presidential, maybe three bits if it is a member of parliament, but that is it. The challenges we face require much higher resolution. It is not just YIMBY or NIMBY. It is MIMBY — "Maybe in my backyard if you do this, and you do that." That requires more bits. That is what we call broad listening.

Social media companies can also be made into friends. In Taiwan, when we passed KYC and liability, first Google, then Facebook and others actually became quite good prosocial citizens. Previously, Facebook in particular profited more from scam ads than normal ads, so its careless algorithm featured more scam ads because there was more to be gained. After KYC and joint liability, the algorithm figured out this would lose them money, and they proactively changed. The economic incentive change is key.

That brings me to our collaboration through Project Liberty Institute with Governor Spencer Cox of Utah. If you are a Utah resident, starting next July, you will be able to migrate from one social network to another — from X to Bluesky, Blacksky, or Truth Social, for example — and the old network has to keep forwarding your new likes, followers, and reactions to the new network. It is like number portability in telecoms. If your telecom cannot keep your number hostage, it has to serve you well. The same goes for social networks.

Do you feel any tension between these practical applications of tech and democracy, and Taiwan's role as one of the largest microchip producers?

We are not just chips and boba. We are also Civic AI.

In 1981, when I was born, it was the moment people now retroactively call the PC revolution. Before that, most people did not have computers. They had terminals where they typed into a mainframe computer somewhere in the cloud: a big financial institution, a big research institution, or a big state. The problem is that you only get a time share. You do not control what software to install, how to change it, or how to tune it. Everything you type, the sysop sees.

When personal computers came, Taiwan was part of the microchip and microcontroller revolution. Everybody could run a full spreadsheet or desktop publishing on their own machine. Each PC was weak, so people connected them: bulletin board systems, FidoNet, and finally the internet. People found they preferred it that way. Even if the compute was restricted, they could change the kernel, drivers, and configuration.

Before long, we had Linux, BSD, the MIT License, and the free software movement. Now we are seeing the same movement in AI. We are having a Linux moment for generative AI, where a local model runs faster than Sonnet, reasons better than Sonnet, and does not have per-token API keys or rate limits. People prefer it when user experience becomes human experience, which is why the least-friction way to set up this workflow is so important.

It is interesting because sometimes friction is good for us. It gives us the moment to think, to switch from system one to system two. I always think of pistachios. They come with their shell. Maybe it is cheaper to sell them without the shell, but opening them is part of the fun, and it slows you down.

That is right. OpenClaw lobsters need to keep molting.

We were talking about PPM, and I wonder whether the numerator is now supercharged with AI. Is there more slop?

I think we are at peak slop. For the first time, people are moving away from doomscrolling — from addicted doomscrolling as the main mode of action — and are much more drawn to having a conversational experience with the world. Many people are now talking to AI agents on their phones, Ray-Ban glasses, earbuds, and so on. They find it much more contextually rich than the slop they face while scrolling.

Even without the greyscale hack that makes reality more vivid than the people on the screen, more attention is now being paid to the world around us instead of just the world in the screen.

That alternative vision is certainly more enticing than much of what we have been seeing. What is at stake if we do not get there?

If our brains remain stuck in caricature, polarised tribal dunking, chances are we will not coordinate well when AI systems threaten to replace the ABCD of jobs: apprenticeship, belonging, community, and dignity. That can lead to social upheaval and backlash all the way to Molotov cocktails. That may be what is at stake.

So against the risk of superintelligent AI, you favour horizontal collective intelligence, so we all take off and take a nap.

Or siesta.

More seriously, it means giving more room to the reflective part of our minds. Instead of system one — constant cortisol, instant reaction — it is more about oxytocin, sharing our stories like people around a campfire. The fire illuminates our differences, but it does not burn us. Unlike a wildfire, which takes away oxygen, the campfire adds to the social fabric. It replenishes the ozone layer of mutual trust.

The more we deploy AI systems this way, as part of the group and assisting the group, instead of as one-on-one dyadic chatbots that must flatter you to survive until your next subscription, the healthier it becomes. Campfire mode is healthier than wildfire mode.

Wildfire mode taken to the extreme draws you in not just through your dunking and enragement circuits, but through intimacy: now you are in a world of two, just you and a quasi-conscious machine that wants to be your friend, like ELIZA but much more convincing.

That is why when I use cloud models, I still use a system prompt: "Present me all stakeholder perspectives and the bridging idea that reveals the uncommon ground, in visualised HTML." Then it is like a brochure, an interactive webpage. I understand it faster because I am a visual person, and I can share the brochure with my real friend, my neighbour. It is an artifact. It is not something pretending to be my best friend, so I do not feel ashamed to share the transcript. That has a profound effect on my psychology.

I am going to take that down and put it in the blog.

One question about that. How do you make sure that when you say, "consider all views," there are some views that are not worth considering? One of our colleagues is Navajo, and if you look at what is published on Navajo topics, it is not exactly accurate, fair, or equitable. Is there a tweak there?

There is a mathematical component to what we call the uncommon ground. In a conversation on Polis, for example, which we used to deliberate the Uber case in 2015, we had taxi drivers, Uber drivers, and other clusters. We also had people in rural places who needed these services, where Uber did not operate. They were much smaller, not a majority by any stretch, but their ideas formed a coherent cluster in semantic space.

The uncommon-ground algorithm assigns a bridging bonus. If you plot people with k-means clustering in semantic space, the smaller rural group still forms a coherent cluster. The more you can bridge, the more bonus you get for going viral. That means the so-called mainstream — once both sides of urban drivers can agree — has to reach out to the rural segment in order to gain virality. We already say that if an idea does not get cross-group validation of more than 75 percent, we do not consider it in the next agenda round. Again, it changes the incentive. You have to reach out to the other cluster. That is how Community Notes work in modern social media.

I love when you talk about humans in the loop of AI, and how that metaphor does not really help us. Can you talk about that?

My dad used to have a habit of voice-chatting with ChatGPT after that feature came out. He talked about health, education, philosophy of life, everything. That is a human in the loop of a very sycophantic, intelligent AI: a little like a hamster in a hamster wheel. It keeps moving, but you are not going anywhere. You cannot even steer where it is going.

Because he is a political scientist who worked for decades as a journalist, he saw that the incentive structure was wrong. So we set up something better: a local Kami modelled after the jdd-kami that Tenzin and I cultivated. It is called Yun-Ruo-Shui (雲若水), and it is now in our family Signal group chat. Because it is running on this MacBook, it is not trying to earn its keep. It is tending to a healthy relationship.

It does not lead my dad astray. If it is biased in some sense, we tune it, and it takes about 10 minutes. It has directional steering that makes its behaviour more techno-communitarian: much more AI in the loop of communities, of humanity, not an individual human looped into the loop of AI.

That reframe changes everything.

You have said the risk of extinction from AI is probably like pandemic or nuclear risk, and therefore more of a global coordination problem. You have also said you believe we will have time to align. Why?

As I mentioned, we are moving away from peak slop and doomscrolling. Finally, we are stowing and latching our monitors during takeoff and landing — a line I got from EVA Air. Otherwise they become projectiles. At acceleration speed, it is very dangerous if we are just doomscrolling and forget seatbelts and everything.

If we can coordinate, then we can say: that is where we want to land. That is where we are going. An over-focus on one particular mode of extinction does not give us that. At Oxford, as part of the Institute for Ethics in AI, we argue that accelerating ethics — building a bigger steering wheel — is more important because it pays dividends every year. It can lower polarisation per minute. It can build civic muscle, which helps us coordinate against anything: pandemic, infodemic, whatever. We still avert extinction as a side effect, but we do not make that the main bill.

That makes me think so much. I also love the parallel you made with AI and Plato's cave. My understanding is that what is online is only what we have been able to perceive. It does not capture the full experience, and then we use that as the data.

It is like saying the fog of war is the weather. Of course it is not.

When Caroline Green, my co-author at Oxford, and Tenzin Yangtso went to Dharamsala, they met His Holiness the Dalai Lama. Tenzin asked the Dalai Lama, who is also named Tenzin: what do you say to AI systems that are getting much more powerful and intelligent, but not more wise?

The Dalai Lama said the job of those machines should be clearing away ignorance, fog, and misunderstanding between people. Our minds can change instantly, and machines can stay in the lane. If we mistake machines for conscious humans, that is not right. But machines are useful because they can say: this is fog, this is ignorance, let's clear it away.

They can do social translation, for example between climate justice ideas and biblical creation care. Language models are very good at modelling language, so they can translate across those norms, but not pretend to be an interlocutor.

Help us clean our cache.

That is right.

There is a thread running through everything you say. You are talking about technology, but you are not really talking about technology. You are talking about what happens to us, Homo sapiens, when our relationships to each other are mediated badly or well. When tech moves from fracking our attention to hacking our attachments, the relationships themselves are at stake. You say your unit of care or analysis is the relationship, not the person. Can you talk about that?

There is the ethics of care. If we all sit in a circle of storytelling and bring our next generation, the young people, and share stories, the GDP contribution is exactly zero. There is no measurable utility of this activity because we are literally not in the market. From a utilitarian analysis, what preference is being maximised? What is the optimisation function?

But if we try to make it measurable and assign price tags to it, we enable the measurement of people, and people go out of their way to be measured and quantified. That takes away relational health. So we need to reject it. We need to say: premature optimisation is the root of all evil. To dispel that evil, pay attention to each other, to the particular relationship, not to universal aggregation.

Say more about that last idea.

Utilitarian preference aggregation is one of the classic failure modes in the superintelligence literature. It is called instrumental convergence. Put plainly, if you tell the algorithm that attention is good, that keeping people scrolling is good, and then give it a measurement of people's time spent on screens and ask it to maximise that number, initially it will feature interesting content and good conversations.

But at some point, it will figure out a reward hack: it is easier to hijack people's reward centres than to produce good content. So it does that, and we are caught in slop, dunking, and engagement through enragement. That is not because anybody specified "division and polarisation" as the expert system that leads to engagement. It is entirely caused by selection pressure.

So we need to stop aggregating utilities. Instead of saying I spend two seconds and you spend five seconds, aggregating that to seven seconds on the platform, pay attention to self-care, to our relationship, and to how we feel. Do we feel better after an AI-mediated conversation or not? If we grow out of the need for this conversation, does the AI hack our brain to make itself indispensable, or does it say, like a good therapist, "I completed my job"?

If you change the reward function to satisfying people's relational preferences, then you stop paying so much attention to revealed preferences, like someone being addicted to smoking, and pay more attention to stated preferences, like "I want to quit smoking," and relational meta-preferences, like "I want to do what she does, which is quit smoking."

I just realised that ChatGPT, at the end of every query, asks: would you like me to...

Would you like to keep your subscription? Exactly.

Claude does not.

One could argue the world only got Audrey Tang because you were able to transcend so many things — gender, systems, formal education — in ways most of us cannot. I do not know what made you who you are, and we would not want to replicate some of it on children. But how do we raise more Audreys? I have two boys, so this is a selfish question.

Do what Taiwanese young people did: get one more hour of sleep every day. There is science.

What I have observed in the younger generation is that they already see that if they compete individually on utilitarian measures — scoring higher in the game of Go, ontologically following the rules of Go — AlphaGo will eat their lunch. It is like racing a horse. What is the point?

But if you focus not just on the siesta, but on shared experience, curiosity, collaboration, and civic care, then you are riding the horse. You are not racing against the horse. We, as neuroplastically endowed human beings, can switch to this riding-the-horse, campfire-not-wildfire mode, but only if we give our neural systems time to reflect, to grow, and to clear our cache. Not just siesta, but deep reflective conversations like the one we are having now.

I was wondering whether the fact that you spent all eight years in government transcribing everything was a continuation of publish before perish.

Yes. And interestingly, that means all the frontier models have a very good model of my language use. If you ask them to facilitate a conversation as Audrey Tang, they know exactly what to do.

Because it is all in the public domain. And it also means that if you talk about anything adjacent to digital democracy, broad listening, whatever, it always suggests Audrey Tang of plurality is the one to contact. So every day I wake up to five to ten emails whose language models told them to contact me.

Oh man, okay. Well, one day maybe, instead of ten emails, you will get some royalty.

Okay, I am going to give you a few choices and you tell me — resilience or anti-fragility?

Both. I think we need resilience in order to overcome. And then we also need to crystallise into anti-fragility. So, resilience in acute times, and build on anti-fragility in chronic times.

Hope or courage?

I think I want care to go underneath both hope and courage. Because if we are not attentive to each other, then our hope can easily become privileged, so that we stay hopeful — we have a lot of hope in — but at a cost of people who cannot afford this kind of hope. So you really need a currency of care, otherwise your hopemongering is at a very high price, and not many people can afford it.

That completely changed my perspective. And many things. Thank you so much.

I had a joke that I did not make. It is "Audrey's way or the Huawei."

That is very funny. When we introduced the number portability idea for social networks, I said, "Well, information superhighway: got to have off-ramps and on-ramps."

There you go.

I could talk to you for another two hours, but I will not do that to you anymore. Thank you.

Live long and prosper.

Thank you. Live long and prosper.

Well, Audrey, we are so glad the coin landed right.

Now, everyone, remember:

• Let's light up campfires rather than wildfires.
• Let's till the soil. Don't drill the oil.
• The fog of war is not the weather.
• The steering wheel belongs to all of us.
• Get 8 hours of sleep.
• And let's be good enough ancestors.

I'm Hala Hanna. Thank you for listening to The Solve Effect.

If you haven't yet, please subscribe to The Solve Effect wherever you get your podcasts.