Hélène Landemore

然後是真正分歧的問題，也許像賠償之類的。但同時，我們從愛爾蘭最近在墮胎議題上的實驗知道——那是一個非常有爭議的議題，涉及非常根本的價值觀。當人們正面面對——當這件事被框定為一個「問題」，而不是「價值觀分歧」時，即使是反墮胎的人也能趨向共識：好吧，你知道嗎？就法律而言，我們應該將墮胎除罪化。我仍然反對，我仍然覺得這是悲劇、是罪行等等，但女性的性命攸關，你知道，我必須暫時擱置一些個人信念，轉而接受這樣的法律安排。

如果我們在這樣根本性的價值觀上都能做到，或者說在這一類問題上都能做到，為什麼在其他幾乎任何議題上做不到？我只是覺得，我們還沒有觸及可能性的邊界。對。

對。關於這一點，我想說的是，這聽起來像是一個二選一的問題。要嘛你快，要嘛你有包容性，但很多時候，那些聲稱很快、然後以速度之名犧牲包容性的流程，最後交出來的結果很糟。所以也許慢一點、確保所有人都參與進來，最終反而省了時間——因為你最後會得到更好的結果。

我從——嗯，我是說，臺灣和法國在疫情管理上的對比就很說明問題。噢，我們有一個非常，你知道的，積極的、由上而下的政府，接管一切，然後，你知道的，沒有經過任何審議，就實施了一次又一次的封城、宵禁、這個那個，有一段時間非常順從地聽從專家意見。結果呢，我們的表現比臺灣差太多了。那麼，你知道的，速度在我們的案例中到底有多好？

而且也許我們還有其他問題。我確定這個比較並不是在其他條件都相同的情況下進行的；但儘管如此，我懷疑，如果我們一開始就進行更多審議，步調慢一些，也更具包容性，如果我們先了解各個社群的狀況、傾聽不同群體的聲音，而不是只聽專家的，我覺得我們現在的狀況會好得多。疫苗接種也是一樣。

好。所以我自己的看法是，我認為到了某個時候，使用某種自動引導工具將無可避免，因為如果我們要進行全球規模的對話，逐一引導每一組六、七個人，成本——成本實在太高了。而且這項工作看起來不會太難自動化，我是這樣想的。我想，這會是未來強化民主的一種好方法。

然後我看待 AI 的方式——我的意思是，再說一次，我完全不是專家——但對我來說它就像一面改良過的鏡子，我們可以用來更好地探索我們的集體心靈，還有，你知道的，我們的歷史、我們文化與其他文化之間的事實。我認為它是一個了不起的工具，只要我們把它當作工具——就像專家應該是輔助而不是凌駕一切，我認為 AI 應該隨時可用，但不應該控制我們的生活。所以我認為這是每項科技的宿命。從這個角度來看，我不覺得它有什麼不同。

謝謝你，Allan。謝謝。

生生不息，繁榮昌盛，是的。

Can you hear me? Yes? Okay. So first, I’d like to thank my former Yale colleague, Allan Dafoe, for including me in this conversation and for the kind introduction. It’s a great honor to — and an immense pleasure — to be able to meet Audrey Tang, whose work I’ve been admiring for a long time from a distance, for its incredibly technologically visionary aspects, certainly, but even more so for the philosophical principles of openness, transparency, fairness, and fun behind it, as she presented them, which very much resonates with the philosophy of politics I have tried to develop in my own work, including in my recent book, Open Democracy .

So I first heard of Audrey when I was visiting Stanford in 2012, 13. I think the Silicon Valley was perhaps more, you know, aware of what was going on in Taiwan at the time, though I can’t exactly remember who first pointed out her work to me. Around that time, I was discovering and exploring myself the merits of practices like crowdsourcing. I was involved in the design and running an analysis of an experiment in Finland on a crowdsourced law reform of off-road traffic regulation involving the regulation of snowmobile in northern Finland.

And that really taught me something important about how radical principles like openness and transparency can generate a free flow of ideas, collaboration, and genuine creativity, in a way that much more traditional, top-down, rational, planned, centralized methods actually can’t. So from my perspective, there’s a magic to these kinds of principle that it’s actually hard to explain and understand unless you’ve seen them deployed and at work. I think only then you can fully appreciate that it’s something worth pursuing.

And I noticed how hard it is, at least in my home — in my own country of France, and I think in the US as well, how hard it is to sell these principles and the technologies that try to implement them to government representatives who only believe in familiar, rational, top-down, opaque structures where most of the control is in their hands. Most politicians, as it turns out, are deathly afraid of losing control. Audrey Tang, from that point of view, is a rare politician or poetician, as she probably is better labeled, whose ambition seems to want to give control away and empower other people. Again, I think you can only fully understand the radicalness and potential of that approach when you see it working firsthand in some of those experiments.

So for those of you who may not know fully who Audrey is, and despite the brilliant presentation she just gave, I thought I would summarize briefly what I take her work to be about, because it’s so wide-ranging that maybe her presentation didn’t quite cover it all. So I knew some of it myself, but I discovered a lot more while preparing for this encounter.

So the first aspect is definitely the technological aspect. As digital minister of Taiwan, Audrey is famous for using very innovative technological methods to bring legislative agenda questions from the periphery of power to the center and helping solve identified issues in consultative, efficient, fair and fun and legitimate ways. Many of you have — had probably already heard of vTaiwan. So vTaiwan is this process that combines a website where anyone can register and put forward an issue and try to gather votes to push it on the deliberation plate of Audrey and her team. But it’s also a series of meetings and hackathons for problem-solving. And it’s also, vTaiwan, the use of an AI-powered system called Pol.is, as she presented it just before, that allows them to identify the preferences, judgments, and in fact, underlying evolving consensus among large groups of people.

And the goal is both to allow more access for ordinary citizens, especially the youth, into the process of shaping the legislative agenda, and to tap the collective intelligence and creativity of the group to offer better solutions to collective problems. She mentioned the pandemic obviously, but one of their big successes prior to that was the management of Uber and Airbnb’s arrival in Taiwan’s economic ecology. And I love the tagline of that — of the vTaiwan website. It says, “Where do we go as a society? Let’s go and think together.” And to me, that’s really the spirit of my own vision for open democracy, which is, well, as a first take on the problems that the world throw at us, we have to ask ourselves collectively, where do we go as a society, and involve absolutely everyone in the conversation.

So beyond this technological aspect, she’s indeed pushing for new design principles which double as governance principles, and that’s where I think the radical potential is the most visible. So among those are radical transparency, what you could call anarchy or at least flat hierarchy, pluralism, decentralization, and fun. So in terms of transparency, as a minister, unlike most of her peers in advanced democracies, she really makes a habit of posting all her talks online, so there’s no secret conversations that only some people are privy to.

Anarchy or flat hierarchy, as far as you could tell from the pictures, her whole ministry is not really headed by her as much as inspired by her. This reminds me a little bit of the philosophy of the former leader of the Pirate Party in Iceland, Birgitta Jónsdóttir, who herself called herself a poetician, meaning a crossbreed of a poet and a politician, and refused to be seen as the head or representative of the party. So it’s really in these figures, Audrey or Birgitta, we see a different kind of leadership. It’s leadership by example, leadership by kindness.

As to pluralism, it’s this idea that there should not be a monopoly of authority in any area of life, that polyphony is essential in any decision process, which doesn’t mean conflict. And even as we are sort of aiming for a consensus or a consensual solution, we don’t want to silence any voice. So it’s important for her to always include an alternative to the mainstream narrative or the main algorithm or the main source of authority. And all of that seems to me, it seems to translate in a further principle, that of decentralization. There’s not one path to the center of power. In fact, there should be many paths and many centers that sort of self-manage and develop their own logics, which feeds in turn into the pluralism.

And finally, fun is a huge part of her design, both for instrumental and intrinsic reasons. She showed you cute dog pictures. She uses cute cat GIFs, because they make people happy, but also because they can help make viral a government’s tweet about, say, the need to wash your hands frequently. Or because fun is the best way to incentivize or rather spur creativity as well as fulfillment and a sense of community.

There’s a third prong to Audrey’s work, which I think is also essential, which is education. In a world where we know that online experiences are going to be more and more threatened by the presence of deepfakes, technologies that create addictions and manipulate people’s attention, et cetera, et cetera, she aims to inculcate and cultivate digital competence, as she calls it, not literacy, among children, from kindergarten basically, to equip them with the right cognitive capacities for the world they are going to live in. So it’s very much a revolutionary form of education that they’ve piloted in Taiwan, and they’ve tried different kinds of approaches. Many is basically promoting autonomy and self-reliance and the pursuit of one’s personal projects from a very early age on.

So she talks also in other presentations of a pluralism of AI mentors, if we’re going to now live our lives with the help of personalized AI mentors, in the sense that we need to have a way to forge and refine the values that we want, including by playing one set of advice against another, a little bit like we can play one parent against another.

So I now turn to some questions I have for Audrey. There are some questions around design that, you know, uses dichotomies of deliberation versus aggregation, random selection versus self-selection, anonymity versus publicity. I understand that vTaiwan and the other processes she uses blend those things and have different components at different stages. But I have to say still for me as a deliberative democrat, so someone who’s committed to the view that the quality of deliberation crucially matters to the legitimacy of laws and policies, I have tended to prefer in my own conceptualization the central structure of a large randomly selected assembly of citizens, such as, for example, the one that just concluded in France.

A body is called a Citizens’ Convention for Climate, which — and it was a body of 150 randomly selected citizens who spent around 12 months at this point carefully debating ways to curb French green gas emissions in a socially fair way. It was a long, costly process. Can’t say it was a sort of fast iteration, 48 hours hackathon that Audrey is familiar with. Much of it was face to face until basically the pandemic forced the group to move on Zoom. And overall, it was quite low tech, apart from indeed the use of Zoom and a platform where they could exchange some ideas. And it also only involved the larger population in an indirect way.

In the vTaiwan process, by contrast, Audrey relies on the participation of a much larger but essentially self-selected group whose demographic representativeness is questionable. Additionally, whatever deliberation happens is of a more decentralized, anarchical, and distributed nature, taking place over several months, between often anonymous participants who come and go on the platform. So there’s no guarantee that the viewpoints represented are those of the larger population. I think she showed us a slide where I think the number of participants was something like 2,000 people, so a far cry from the 27 million people in Taiwan.

And yet, I have to say, this seems to work beautifully to generate consensus, to work basically. The same way that the process that I observed in Finland worked, even though it was also based on self-selection and sort of little bits of deliberation here and there, but nothing like a centralized moment of actual exchange of arguments. So I have a question for Audrey, which is how do you account for that magic? And what would you say are the benefits of your model compared to a more centralized citizens’ assembly model? If you had used a citizens’ assembly model to deal with the Uber case, for example, or the pandemic, how do you think things would have turned out differently, if at all?

And finally, what’s the value of anonymity of participants versus transparency about their identity in online platforms like vTaiwan or Pol.is, especially regarding the resulting quality of their exchanges? Because in my experience in the Finnish experiment, we didn’t have too many issues with lack of civility because there was facilitation, but we know that anonymity can trigger behavior that are largely reprehensible and detrimental to the quality of other people’s engagement. So how do you decide yourself whether to require anonymity or publicity on any given platform?

I have now another set of questions about accountability. A common objection to citizens’ assemblies use is their assumed lack of accountability because members that are randomly selected are not under the threats or the sanction of elections. So I suppose you must hear the same type of objections to vTaiwan. Who are those people on the internet? What gives them the right to influence policy-making when we hardly know who they are? They don’t have a mandate. They can’t be sanctioned and made accountable by the threat of elections. Who says — who knows if they’re not being captured by large corporations, et cetera? What do you say to people who have these doubts about the process?

On a related note, one hour in a talk with historian Yuval Harari, which I just recently watched, you mentioned the idea of a distributed accountability, whereby every one of us is responsible for keeping track of mask stocks, for example. And all together, our collective efforts end up tracking reality quite efficiently. And I would like you to elaborate on this idea because it’s very hard to sell to, for example, French officials who only trust in electoral accountability and very long chains of commands with a final arbiter at the top, et cetera, et cetera.

So how do we guarantee that individual citizens will do their job instead of free riding on the work or that of others? Sorry — instead of free riding on the work of others. Is there something perhaps peculiar to Taiwanese culture that makes it so that people are responsible and invested? And how do you jumpstart such a culture where it’s lacking or assumed to be lacking, as in France, for example?

I have a question about plurality as well, which is related to this question of accountability as a design principle. In various talks, you mentioned the value of plurality as a design value to encode in institutions and algorithms. And I would like just to hear more about concrete example of how that works in practice, or even better, about how it sometimes doesn’t work or when it creates new kinds of problems. Because I have a hard time imagining it doesn’t sometimes.

The way I hear you describe it, I think of plurality as something somewhat similar to the idea of separation of powers or the checks and balances in the American Constitution. The idea is that if we create multiple poles with equal power and we allow them to check each other, we save ourselves from totalitarian risks or capture by one algorithm or viewpoint. But at the same time, in the American context, this pluralism or this divide to rule kind of approach has also been a recipe for institutional paralysis and status quo bias. So how does your design pluralism avoid such outcomes?

Now I turn to a question about the relationship of mini-publics to maxi-publics. So you say in one of your talks, and in the one you just gave, that if there’s a sufficient amount of cute cats and memes, the ideas worth spreading will spread. And I suppose that’s been true in your experience, and that’s absolutely wonderful, but again, what accounts for this experience on a causal theoretical level? What’s the theory that explain that ideas worth spreading will actually spread when we know that bad ideas and wrong ideas and falsehoods and fake news also spread?

So there’s an optimism running through your work, which is very reminiscent of John Stuart Mill’s faith that the truth will eventually emerge from the free exchange of views in a free market of ideas, or Habermas’ ideal of deliberation in the wild, being magically able to set the agenda for the deliberation of officials in the central decision track of the public sphere. Or I should say my own optimism in Habermas’ idea that the forceless force of the better argument will triumph. So obviously, I buy this vision myself, but I’d like to hear your explanation of that magic and how we can convince people that this can be expected. So in other words, in your experience, what else explains the failure of good ideas to spread besides the absence of cute cats? What’s the key ingredient for this spreading of good ideas to work?

I have another couple of questions, and then I promise I’m done. A question on the transition from here to there. So how do we get from where we, most of us are, not Taiwan obviously, meaning stuck in dated representative governments that are quite closed in their operational principles, to the there of open systems like Taiwan’s or open democracies, as I like to call them? You once said that instead of fighting an old system, you have to make a new system that will make the old system feel obsolete. You don’t have to convince everybody that the old system is obsolete. It’s enough that you convince just a few people.

I like that answer, but at the same time, if we look at the US, these few people have so much power that in such a massive conflict of interest in promoting a reform of the status quo that this is not going to happen. We are not gonna convince Zuckerberg or the GOP or economic elites to relinquish power or at least share it, because they just don’t have an incentive to. So how did you do it in your own country? How did you get past those hurdles and get to involve people that were not included in originally?

Similarly, there’s another more psychological obstacle, I guess, among people, sometimes of older generations, who fear technology or among groups that fear technology sometimes for very good reason, because technology can be racist, can be exclusionary. And so there’s a whole category of people who, for very different reasons, some of them very reasonable, refuse to engage in this participatory modes. So what do we say to them?

I have a question about the translation to developing societies. Okay. Excusez-moi. I apologize. My kids are back from school, and I told them not to rush in, but they did anyway. So a question about translation to developing societies. Many of your recommendations seem to apply to advanced societies with a tech-savvy population. So how do your ideals translate to developing countries with fewer resources and greater digital divides and illiteracies?

Finally, and this is the last question, I promise, I was thinking of the kind of advice you could perhaps give to a government like the French government or perhaps the organizers of this upcoming global citizens’ convention for climate, which will gather 1,000 people online in parallel to the COP26 in Glasgow, I believe, this year. How would you design a citizens’ assembly in a way that connects it to the larger public, the French public in one case, the entire, you know, humanity in the other, in a way that builds on what you’re familiar with basically, all these AI-assisted technologies? Can we have at this point AI facilitators for the deliberations among smaller groups on Zoom or in breakout rooms?

And more importantly, what should a governance platform for a self-ruling body that can’t rely on the classic structures and hierarchies of parties look like? ’Cause a big puzzle for deliberative democrats is how do we get this citizens’ assembly to self-rule as opposed to be managed from the outside by either professional companies that organize these things or a body of outsiders appointed by government in the case, for example, of the French convention for climate. So if you have any idea about how to facilitate the autonomy and independence and basically sovereignty of these randomly selected assemblies, I’d be really, really curious. So I’ll stop here. Thank you so much.

Thank you.

Can you hear me? Yeah. Okay, great. So I just wanted to consider the objections to participatory democracy, you know, on the ground that people are irrational, ignorant. I mean, it’s a very common objection, but to me it’s really premised on a vision of participation that takes place within an existing system which is designed almost to produce ignorance and apathy and lack of participation. So I don’t think you can reason — you can infer the competence or promise of participatory democracy based on the existing system where there’s very little participation.

And so when people think of including people, they think of including voters that are tribal, partisan, ignorant, misinformed, manipulated, et cetera, et cetera. Well, that’s not what we’re talking about. We’re talking about a different ecology, a different set of institutions where people are incentivized to act in a very different way than they are incentivized to act under, you know, the current electoral system, broadly speaking. So I’m less worried, and I think in fact it’s always this very theoretical objection that sort of dissolve when confronted with the empirical reality of success, you know, as demonstrated by Audrey’s work.

That said, the limits to participatory democracy, I’m willing to believe there are some. It’s just that we have so much more to discover before we have to ask this question. I feel that maybe it’s a little stifling to start with this question, I would say. That said, again, one limit could be the time that’s required of participants. It is time-consuming to be invested in these activities, and we don’t have an economy or an ecological sort of environment that necessarily gives people the time they need to do that.

I mean, they still do. You know, Wikipedia is basically born out of free labor. You know, and all these people who participated in the Finnish experiment or on the Uber deliberations, I mean, they did it. I don’t know what, you know, how, between the cracks, between preparing meals for the kids, between going to work, while waiting for the doctor. You know, who knows? But they do it. So we don’t wanna tax that too much, however. So we have to rethink perhaps compensation schemes. Maybe we need a universal basic income for this to be truly feasible and truly inclusive. So there are limits, but I don’t know if they’re inherent to participation so much as to the environment in which we place this participatory ideal. So yeah, I’ll stop here about the — one more thing maybe.

I’m not totally familiar with the book by — 10% Less Democracy , but I think the worry is about referenda and this kind of like purely aggregative judgment elicitation on very specific issues where the citizens do not control the agenda. But I think there are, you know, these methods of generating an agenda on the basis of a sort of a consensus that you sense and build, you know, through technologies, it’s very different. I mean, there’s still the possibility of capture. I’m not saying it’s not there, but it’s less the case than in the Californian system, where indeed the agenda is pretty much in the hands of whoever can buy the most signatures. So I think it’s a little different.

That said, I would be curious to hear Audrey talk about the danger of capture. For example, in the vTaiwan process around Uber, how did you make sure that, you know, Uber didn’t come in the way they kind of did in California, in the referendum on the gig economy, to push for their own interests against the interests of the other stakeholders? So how does your design prevent that kind of capture?

Yeah. If I may add, it’s exactly what we did as well in the Finnish experiment. We had a way to cluster, you know, groups, and to identify where they overlapped in suspicious ways. And so it was about the diversity rather than the num — the headcounts, indeed. Because especially in our issue, it was regulating snowmobiles, and so 80% of people on the platform were men, very libertarian, very anti-regulations. And so we had a — you know, if we had gone by headcounts, the views were really one-sided. But we had also some women, mothers of young teenagers, who are really worried about the speed limits and age restrictions. And so we were able to visualize the different groups, and so we were much more focused on diversity of arguments and perspective than headcount. Absolutely.

Yeah, and I think it’s also worth emphasizing, I think, because we get that objection a lot about the self-selection and the bias and all that. But they’re not deciding. So it’s really like, what’s really neat about what you’re doing, and what can be done in that space, is that you can really segment the process so that there are different steps. There’s ideation, there’s problem-solving, and then there’s decision. And when it comes to the decision phase, that’s probably where we still rely, as you said, on the traditional accountability mechanism of elected officials, who are going to have to endorse this input, which is so much richer than what they get through elections.

So but at the end of the day, there’s still this moment where they have to decide. My addendum would be to say, well, okay, but we don’t necessarily need to have them — I mean, it’s indeed debatable — but even that final step need not be in the hands of elected officials, if we can figuring out a way for randomly selected assemblies, for example, which to my mind are an authentic form of democratic representation, can also be made accountable in some non-electoral ways. So I think, you know, if you combine your ideas and (laughter) visions and some of like the theoretical constructs I propose, I think we have a completely different democracy. We’re outside the representative government model, which is — has shown its limits.