Automatic transcription, there could be errors.
Alessandro Oppo (00:00)
Welcome to another episode of the Democracy Innovators Podcast and our guest of today is Paolo Spada. And thank you Paolo for your time and as first question I would like to ask you if you can share something about yourself and your research path, also like what you worked on.
Paolo Spada (00:24)
Thank you for having me Alessandro. So my research is quite unusual in the field because I started as an economist studying monetary crisis. I finished my undergrad in 2000. My dissertation was with Professor Giavazzi that is quite well known in Italy. But then I read a really famous article from Le Monde
There was talking about participatory budgeting and the social forum in Porto Allegre and I got really interested into it and then I was talking with some friend and he came out that a family friend Giovanni Allegretti was the first Italian to actually have studied participatory budgeting and one of the biggest experts of participatory budgeting in Europe.
random reason because Giovanni was vacationing there when he was 16 because his family was going there because they had connection.
to civil society organizations and religious groups. And so I brought Giovanni and Giovanni told me that he was organizing the first training for civil servants in Trento in Europe on participatory budgeting. That was June, no, maybe July 2000. I went there and I kind of got the bug of democratic innovation.
But at the time I was still studying economics. I began a PhD in economics in Bologna and then I got in a Northwestern. I went in the US to do my PhD in Northwestern. I really didn't like it much.
I don't know if you've read the capital of Piketty, but there is an introduction that really describes also my experience about this mathematical model that reached a level of elegance. I love math, very mathematical elegance, but completely disconnected from reality. And so...
There I was thinking about my life. was coming back in December of my first year on the plane and close to me sits Michael Wallerstein, one of the few analytical Marxists of the US. We discussed for six hours. He was doing model of redistribution and we discussed. He was also a Northwesterner and I started an independent study with him when I came back after vacation. And after a year of independent study, told me, look, if you want to study, what do want to study?
you need to move to political science. Big crisis because I was coming from economics and economists that look down to political scientists. As everybody knows, there is this funny ranking where at the top the physicists that look down to everybody else and it goes down like in the food chain of academia. think political science is probably one of the lowest. I don't know where we are. And big crisis and
But then I decided against my parents. My parents were like, no, what are you doing? No, I decided to switch. So I applied again to a PhD and I moved to Yale with Michael.
to start a new PhD at 27 years old in economics. named politics, sorry. That was a big shock obviously because I didn't know anything about politics, right? I was coming from a complete econ background. And there...
I met people that were studying Latin America. Finally, my little project that I had on participatory budgeting at Northwestern, I was going around professor by professor with this little project asking, are you interested in this? And they're what? No, that's not economics. Finally, I found both economists and political scientists at Yale that didn't know anything about PB, but they were interested. And so I managed to go study participatory budgeting in Brazil. I lived one year in Brazil.
And I contributed to the creation of this data set that is the largest data set in the world on a democratic innovation is called the Participatory Budgeting Census. It's kind of a public book. I really insisted to be a public book. Good. I used the Creative Commons license when nobody was using it. They were talking about eons ago. People use it without citing me all sorts of things, but I didn't care. I kept insisting and I kept redoing
It's a survey we do every four years. We basically call all the 500 cities in Brazil and they ask if they do participatory budgeting or not. And it was used in the sort of first large scale quantitative analysis of diffusion of participatory budgeting and impact.
there is still nothing similar to that data set, we just did the update now. And then from there at Yale they were doing deliberative polls because Jim Fishkin was collaborating with people at Yale. So I started as a research assistant giving sandwich to the participants and then I slowly went up. So I started from the sandwich, then the data entry and then the last one I was actually
designing an experiment and I also got into America Speak. That was one of the early experiments for multi-site deliberation. They were using basically Skype and clickers. We are talking about 2006-2008. To connect multiple cities they were doing deliberative events together. So I organized one together with a Japanese researcher in
at Yale. then Elaine Landemore arrived. Her first job as a professor out of Harvard was at Yale. That was my last year. I just came back from the field in Brazil and I took her class on deliberative democracy and we had this very interesting, lovely conflictual relationship because she's more deliberative and more participatory.
she really opened up a lot of my interest. That was also when I met Hugo Mercier, it's one of the psychologists that works on collective intelligence that does amazing work. That then I collaborate with some grant, that was one of the collaborators with the land. The next step, I was really lucky and I went to Harvard in Arkham Fung and then all sort of complicated life reason for which I couldn't stay in the US, I had to come
back
to.
to Europe and because my family is in Milan now with my ex-wife. And so I worked, but I did a step in Canada with Mark Warren at UBC where I helped organize some citizen assembly. That was my first experience with Canadian style citizen assembly that I really like. I was one of the facilitator of one of the citizen assembly on Grandview Woodland.
It's a district in Vancouver. And then I came to the UK to do what the Fair Citizen Assembly in the UK, they were done in 2015. Then I moved to collaborate again with Giovanni Legretti, that I mentioned before, in an Horizon grant in Portugal in which we developed a digital platform for participation. In the meantime, through Elan, I'd met Mark Lane that had developed the Deliberaturo.
that it's an argument mapping platform. If we're talking about scaling up, that's what I'm gonna talk about because the early experiment before AI and before the current wave on how to scale up large scale discussion online where asynchronous and text-based and argument mapping was really promising and it's still promising as a technology. And then ultimately I got a job at South Africa
I did one year in Italy, L'Aquila, at GSSI, the research center on...
on regional economics where they organized the Festival of Participation for a few years. Now, I think he moved to Bologna and I really like that. But then I got a job at Southampton and I really like Southampton because it was really valuing the Democratic Innovation Scholarship. We now have one of the largest faculty, not just in politics, but in all
Alessandro Oppo (09:29)
I feel it, don't know if it's blocked or not
Paolo Spada (09:50)
department, if you look at the Center of Democratic Futures, we probably have 80 to 100 scholars that work on different kinds of democratic innovation from computer science, politics, economics, all sorts. It's really, really interesting in Southampton right now for this stuff.
Alessandro Oppo (10:10)
I was curious, what is different in relation to citizen assembly, you mentioned the Canadian way, and also another question, during these experiments related to participation, what worked and what didn't work as you expected?
Paolo Spada (10:32)
Okay.
So as the terminology of citizen assembly is traced back to the British Columbia Citizen Assembly that was done in 2004, it's kind of the grandmother of the Citizen Assembly. And the way I define it is a systemic innovation, right?
because it engages multiple audiences at the same time, significantly more systematically than most of the other models. So what's the big difference? Apart that it's longer and bigger. So instead of like having two, three weekends in last one year and they had, I think, 10 weekends, I don't remember. But if you look in participator, there is a nice case study by Mark Warren described. But
On top of that, they had 56 open to all meetings in which the participants of the assembly, the one that were randomly selected, went back to their location because the region was divided in 56 locations and two people were randomly selected from each to come. And they basically act as a facilitator of an open to all assembly together with professional facilitator to bring
the discussion to the maxi public. The key problem, a legitimacy problem of the current generation of mini public is that they're too closed, right? They don't connect sufficiently maxi public. It's really hard to do that connection. With technology, people are trying now, maybe they ask people for comments, but then it's unclear what happened to the comments. And so the Canadian model, when there is enough money and when they're doing a CT
this is an assembly because sometimes they're doing other things as really these sort of multi-channel engagement system in which it has the randomly selected assembly plus open to all events. And I think that works way better. That's my personal opinion than the standard model. A deliberative poll, it's teeny compared, it's completely different system. And the question of what works, what did not work? Are you referring it to participatory batch?
or to the entire galaxy? Because if it's the entire galaxy, we're going to stay here three months.
Alessandro Oppo (13:02)
Ugh.
So
maybe not the entire galaxy but like yeah in your experience something that maybe you were expecting certain
Paolo Spada (13:15)
All right.
If you think about the continuous scale from the most evangelist of democratic innovation and the more sceptic that thinks that nothing works, I am a sort of sceptical friend. I think there is too much hype in the field, in part because the field was minoritarian, particularly in the US, public policy was...
was
completely against deliberative process. They thought they were just mumbo jumbo and an interest group will control everything. If you still find or present democratic innovation stuff to a traditional sort of polyarchy train, US scholar in public scholarship, and it happened to me a couple of times, you feel this perception that what are you talking about? Right? This thing, it's can be, politics is power, right?
stuff make no sense. It's obvious that it doesn't work and this is just... you're deluding yourself to think it works. so I am an empirical scholar.
And so I found myself many times in the position of testing empirically some of the claims that philosophers had put out about the potential of deliberation and not finding a feedback in the empirics. And I was a pioneer of the empirical studies, right? I started in 2004 together with many others.
But so I come from that side, but the field has evolved a lot. Initially, there was a lot of allergic reaction if you were criticizing exactly because it was a minoritarian field, right? So the title of my dissertation was something I forgot what it was, but it was the original version was something about the problem of participatory budgeting or the limit of participatory budgeting. And my external
advisor, Brian Wampler, the most famous researcher on participatory budgeting in the US, told me to change the title. said, you still have to sell participatory budgeting. Nobody believes that does anything. Right. So first we need to establish that it works. And I was, yes, but like I found issues. But I think it's a problem with all the sort of niche growing field.
that are normatively driven because we believe that democratic innovation is good. So there is an action research element, a political element to it. And we talk about it a lot in a paper that we thought...
nobody would have read. We've wrote it like in a month, me and Matt Ryan, after a conference called the failure to examine failure in democratic innovation. But it's actually the most one of my most cited paper, right. And he talks about this our weird field, right. It's a bit strange. There are all sorts of little conflict of interest, because we are doing research on something that it's also a political process we're really passionate about. It's very different from what I
know a person that studies economic crisis as I used to or political parties right and so navigating this conflict of interest is fascinating but also challenging. I'm often asked to sign NDAs.
right, by some of the big players in the field when I'm evaluating for them. And I am an evaluator. And so I, but now I know, like the first time I was puzzled, I had like a big crisis when I was asked by participatory budgeting project in the US. After I helped to evaluate New York City, there was a bigger process US wide. And they said, okay, Paolo, yes, you can evaluate, can help for free.
free,
obviously, but you're going to sign an NDA. And I was, what? An NDA? Where transparency is going, accountability, right? But then three minutes of reflection and you realize that...
there is an important political project that it's minoritarian and it's attacked left and right by very conservative forces. And what do you think it's a minor critique that you are providing to improve the project, right? Something didn't work here, can be instrumentalized by the media, by the opposition to kill the entire process.
And so again, that was my reaction as a young postdoc. Now I actually, when I have a client, I immediately say, look, there are two ways to evaluate here. Do you want an independent evaluation?
Or do you want me in research and development mode? At that point, if I'm R &D, I don't want my name on the evaluation first, because at that point, it's not my academic work, it's another job. I'm here to help you because I believe in your project. And we don't need to publish anything, right? We can decide what to publish or not in your name, right? Because at that point, the publication are way more sketchy. But that's my way of navigating it. But it's complex.
It's not easy. And what works and doesn't work, now there is a vast literature on the impact, but it's still, if you look at it, it's still very limited, right? We still don't have a lot of meta studies, very few replications, and there are trends that are emerging.
but it's a very complicated literature to say exactly what works. So I'm gonna give you my take, but it's an experiential take that it's based on my intuition and I would not be able to prove it academically, right? And so in my opinion, participatory budgeting is the best technology to improve trust at the local level because it is designed structurally to close the feedback loop.
when it works, we're talking about when it works. Because what you're doing is you're letting citizens propose ideas, rank ideas, and then the city is implementing those ideas in a short amount of time.
So there is a direct feedback loop, like in a video game, you push a button and something happens. And in PB is designed for that. And we have some data here and there, but it doesn't always work. But the result is not really that data. The result is the fact that PB, to me the intuition is PB spread like wildfire.
And it's used a lot by charismatic new mayors that come in for the first time and they need to show that they're different from the past. And that's when it really shining works.
Interestingly enough, I have a very different position from a lot of the literature on PB that says the more years you do PB, the better it becomes. I'm actually kind of against institutionalization because apart from a few cases, my research shows that PB has a life cycle and it tends to die after eight years. Like it's really hard for a PB to survive really well at the peak.
after eight years. If you look all the best practice after eight years, they are lower in quality. Like usually the first year are not the best because they need to learn how to do it. Then there is a peak. It's like a life cycle. And the peak is the moment in which the PB achieves his best outcomes. This closes this feedback loop.
Often there are political positive feedback in which the mayor gets re-elected or the group supporting it gets a lot of visibility. But then there is almost like a glass ceiling of what it can do, right? All the easy wins have been achieved in the first few years. And then there are a lot of pesky problems that the citizens would like to solve via PB. But they're really hard to solve due to the complex ecosystem of a
it here, right? There are...
there are problems that are politically extremely complicated, but that appear easy for the citizens and they get channeled through PB and that's when short circuits start happening. then there is all sort of life cycle issues like politicians do not get reelected because it requires a lot of work, it's politically costly. So PB2B is the best at generating trust and it's also the best university for civil servants.
I remember the civil servants were all annoyed at the beginning, but then after a while they were saying, now I know so much more about the city. People recognize me in the city. When I go around, people stop me. So there are very interesting learning for the city machine that are very rarely studied. I keep asking in my grants on PB to study the civil servant learning, but it's not a very...
sexy topic that gets you money. All the topics are about, we're going to make the citizens better, we're going to defend democracy. And instead, think one of the best outcomes is really the PB is the university for the civil servants. Really to open them up to dialogue with the city and to the difficulty of the city. Because if you think about it, it's a sensor. It's a decentralized, low-cost sensor.
in which the civil society organization, the common citizen bring problems and a lot of them cannot be solved, but they're brought in a more propositive way, in a less conflictual way. And also it bypass a bit, not always, but it bypass a existing barrier of communicating this problem.
students has just done a fascinating study on a different topic on a project done by COVID by Boris Johnson. Obviously they didn't know anything and they thought there was no other voluntary sector in the UK. And so from the top they created an app and a system to allocate random volunteers to bring food to health
people they were stuck in the house because for their weak condition they couldn't go buy grocery or they didn't have the money to use the liver. What is
Obviously, there was a massive reaction because the UK is one of the most vibrant, vibrant, voluntary sector. It's just that Boris Johnson had no clue about it. And so obviously, all these voluntarities, existing voluntarities sector at the local level were really annoyed and said, why you didn't talk to us? Right. And also, they were scared that the system will crowd out.
their work, right? Because if you're volunteering for that system, you're not volunteering for me. What was interesting that in the unique situation of COVID, a lot of people didn't have a job anymore, they're working from home. A lot of people volunteer for both first, but
The key, and this is the connection with participatory budgeting, is that for the first time, a lot of minorities volunteered. If we look at the shape of volunteers in the UK, it's most elderly white people that have a lot of time, they're educated, they've been... People from minority do not participate, right? And I think that's the same very nice entry point of participatory budgeting.
Sometimes, not always, it allows for minorities to participate because it's easy to propose ideas. It doesn't require a lot of time. And then it's difficult to win, right? Obviously to win in participatory budgeting, you need the machinery, but to propose ideas, it's easy. Okay, that's it PB. When we are looking at mini publics, I think...
the current generation, right? It's very important to distinguish between the first best we have not achieved yet, and it's still a bit far away, and the current sort of pilots that we have seen. So in the current pilots, to me, the biggest effect around the participants,
and they are an increase in efficacy. The idea that you can contribute effectively to politics, an increase in learning. And if the recruitment has managed to achieve participants that are not already highly interested in politics, not already participating a lot, there is also potentially an increase in what they want to
do after, right? But the political impact of citizen assembly is problematic for structural reasons. It's consultative, right? PB is created to be safe because the city sets aside a little bit of money that is nothing for the city and it lets people play with it. So it's almost like a safe democratic game. It still costs
Mostly
it can still generate problem, but it's safer. With citizen assembly, sometimes they go into very complicated and difficult problems. And so they stay consultative. And then the problem is cherry picking, right? That it's being discussed a million times in which, yes, the citizen assembly does an amazing amount of creation of ideas.
but then there is a second step about implementation and what's gonna happen there, it's unclear. And so that to me, but again, I couldn't prove it, right? This is experiential having organized, I don't know.
200 democratic innovation and seeing more participatory budgeting. think Citizen Assembly, I lost count, but I'm not one of the big implementers of Citizen Assembly, but I've done 30, something like that. I call Citizen Assembly, but in reality it's mostly like a variety of mini publics.
Alessandro Oppo (29:04)
I was thinking also commitment can be an issue in relation to participatory budgeting because as you said most of the time maybe it's just consultative so they can decide if to follow the output of this.
Paolo Spada (29:19)
BB no,
we participatory budgeting, if it's the definition of symptom air, it's not consultative.
So there are consultative PB, particularly in Germany, but they all died. That's kind of established, right? If you do a consultative PB, it dies off. There is a very nice paper again from Giovanni Legretti on Portugal that shows that the first generation of PB in Portugal implemented in 2004 were all consultative, apart a few. When the crisis of 2008 arrives, all the consultative die, the non-consultative stay.
And then Portugal in that moment really exploded, became the country with the most PB in the world apart the countries in which PB is mandated by the government like Peru. But PB is the power in a sense, not by the fact that it's consultative, but by the fact that the budget you give to PB is a joke compared to the budget of the city. I have this story I always tell my students after I make them enamored, I like to confuse my students.
So I usually make them fall in love with something and then I tell them it doesn't work. It's kind of my way. And so after teaching PB, I recount the story of when I helped Josh Lerner, there was the PBP CEO come to Vancouver because I was a postdoc there. I was working for PBP Pro Bono and I invited them to present participatory budgeting because Vancouver was already doing a million participatory planning things, citizen asset.
I was curious, PB was being relaunched by the PBP project. And so Josh comes, he does his usual spill, and the room is full of the Department of Communication, Engagement, Youth Engagement. We are an hour in, and three people in a suit come in, and they sit, they listen for one minute, and there is the end.
and they introduce themselves and they say, I'm the chief of planning, like the people that count, not the communication and engagement. And they ask, how much money do you need? And Josh said, we recommend one million Canadian dollars for a district because it's a nice symbolic amount of money as a minimum.
And the guy of planning for, yeah, you can have it. But do you understand that it's a lipstick on a peg? My budget is two billion dollars. And he leaves. Right. So so PD is sort of is doing this feedback loop.
Alessandro Oppo (32:01)
So.
Paolo Spada (32:06)
but within a safe environment, right? You cannot do much. And if you think about the way also the money is divided, it's one million, but you don't ask one million.
The objective of PB is really designed as a tournament to maximize the winners. It's really well designed from a game design perspective, not game theory, actual game design, right? Because a lot of people win and in a game, if you want to maximize the feeling of sort of satisfaction, you want to have a lot of winner. It's like the first level of a video game. It's very easy. You win easy, then it becomes difficult, right? And so the million dollar,
at maximum you can propose a project of 100, 000. So you have 10 winners, right? And so all those little rules are really, that have evolved over time, are really designed for engagement. But it comes from the origin. When I interviewed in Porto Allegre, Ubiratande and Susa Santos, that it's really the mind behind PB together with Arsene General and a few others.
he told me straight PB is a hook to engage people that are disengaged so that they come to the workers party and we select the leaders that we need to do our process it was very clear and that's how it was started right it was a
from their standpoints that's how the PT used. Then it was modified, it was transformed in a neutral approach that is governance driven. But the genetics is about really making sort of engaging people, identifying leaders, re-engineerize social groups at the local level because you can establish rules that give more advantages
groups, right? That was typical in the first generation of PB. You got more votes if you got more people in your group proposing the idea. And so that pushed civil society growth. It was dragged, but it still pushed it, right? And so it's an incentive system. But
it remained safe in the modern model. In the early Brazilian system, the money was not set aside. So it was more vague, more nebulous, right? And in Porto Alegre, at a certain moment, it was 20 % of the investment of the city. So massive, right? But if you look at...
modern PB and in fact there is the famous paper of Baiocchi and Ganuzza that call it a sort of decathenated PB, right? They have really removed the radical approach. So there is a really nice evolution. PB starts very radical and it becomes less radical over time when it explodes in the world. But it's still very interesting, right? I like to think
of it as a democratic game, again a democratic school. People don't like it, I never published it because they told me, no, you are belittling the democratic innovation, right? If you call it a game, again, this problem of we are in a movement, right? If you're criticizing the movement, it's complicated. So we have one paper that is called Democracy is a Serious Game that was just presented and never published.
because we got such a negative reaction to the idea of calling participatory budget in a democratic game.
Alessandro Oppo (36:09)
And what can be improved? if we think about a possible next generation of... And also this can be maybe related to the concept of scaling, also how to scale this kind of part.
Paolo Spada (36:24)
Yeah, let's go into scaling.
So.
It's really funny because I was one of the very few people within the deliberative democracy that was really doing scaling around 2012-2013. I did this experiment in Italy with a group from Bologna of the Italian Democratic Party in which we used argument mapping online to discuss the electoral law.
that was done in 2013 and...
the process was called Insieme per il PB, it was with the group of Gotzi that now is in France. It was a weird group of Catholic and leftist together, but very, very nice, a very nice process. I was coming from face to face from Brazil, completely skeptical about the tech. That was the first time in which I saw that online deliberation could be better than face to face deliberation when well organized.
And I started saying those things in a field that didn't like those things. And so it was very funny. Like people were no online interaction are terrible. They don't work. You lose so much information. I yes on Facebook.
If you have a community of interested people that want to use the technology and really want to, and you have a good platform, the outcome can be better, right? And I had the data to show it.
And I had seen it, right? The quality of argumentation. Obviously it was a lead process. That was not massive, but it was a lead within a political party. And there was the start, like we had an online deliberation among 600 people. For a month, they built an argument map that it's not easy.
And we had a lot of people participation, like 60 out of 600 contributing to the map. Like almost everybody participated in something. So fascinating, fascinating example. Then the big switch that was fascinating to observe was COVID. And it was driven by the implementers that had been paid to do a face-to-face process and couldn't do it anymore. So they all moved to Zoom with all the
of transparency and using Zoom. And there was almost a flip and now online became ubiquitous, right? And now a lot of citizen assembly have an online moment among the participants, but the citizen assembly are still relatively small, up to 100 to 100 people. There are some experiments that are doing it bigger.
and there are these attempts to involve the maxi public in the city assembly sometimes are done sometimes are not done the one I've seen rarely are done well
You just open up a website, you ask for comments, you generate, if you're lucky, a gazillion comments, and then you don't know what to do with it because how do you give the assembly a gazillion comments, right? So if you have the energy, you summarize them. Now with AI, you can do all sorts of summaries, but the transmission problem is a crucial problem.
And the other issue, if we are really serious about citizen assembly from an epistemic standpoint, and now Elen Landemore is talking about this, I think we cannot have one assembly, right? Because there is group bias and there is a lot of studies about group bias. We wouldn't have to have a galaxy of citizen assembly that work at the same time on the same topic, and then a smart transmission system that combines
it. A little bit like what America Speaks was doing in 2004-2008. That was the early, early time, right? But that to me is the interesting future. And that and so if I had to organize like my gold standard citizen assembly, I would probably do something in 10 cities at the same time on the same topic and with a transmission system plus a mechanism to involve seriously
the maxi public like open to all events and and and the last element is that there is a lot of hype about the engagement system right so these i've wrote a paper that annoyed a lot of people again i like to do that there was on
the over claiming of representation. If you look at the acceptance rate of the vast majority of citizen assembly, by acceptance rate, I mean how many people when invited say, yes, I would like to participate is below 5%. So yes, you send 10, 000 invitation.
two, three hundred people say yes. Then what does the citizen assembly does? They do the randomization even with the new fancy algorithms, right? The one that was published in Nature from my friend. Those are amazing important algorithms that rebalance various things and they make it more fair, but they make it more fair in those among those 200 people, not with respect the overall 10, 000 you have sent.
And so what's the issue? Because you don't have data on those 10, 000. You know very little. There is no way to rebalance things. At most, you know their address. And if you have data on the address, sometimes we have information about the address, right? So you might know the gender. And so there is an issue.
Alessandro Oppo (42:35)
you
Okay.
Paolo Spada (42:48)
on representativity that needs to be solved by spending more money on engagement, by using multiple channels of engagement, by knocking on doors, by using social media, instead of being very cheap on this type of engagement. And one of the most interesting debates is about how much money we should give to participants. Because one of the claims of many practitioners is that if we
Alessandro Oppo (43:08)
Yeah.
Paolo Spada (43:18)
pay them more, we will see the acceptance rate go up. So the problem is purely resource. But there was a very recent case done in Austria by and it's published on the Journal of Sortition in which they paid every participant 1, 500 euros. Right. That was the invitation for weekend. So you got 3, 000 if you came two weekends. Right. That's a stipend. It's the highest pay.
ever done in history for a citizen assembly. The acceptance rate went up to 10%. Still, 90 % of the people did not come. So the reason of not coming are not about the money.
there are other reasons, whatever they are. And there is very little studies about the reason like we have done one, there is another famous qualitative one, but that's a crucial direction. If we want to make this thing legitimate and sell them on the marketing that they are representative, because we could sell them in another way. And that's my point. My point is at this stage, these things are not representative. We cannot say that they're representative, right? We should sell
them on a Republican argument. The randomization allows you to prevent interest group to capture them. That I stand behind. It's really harder to capture a citizen assembly because you cannot flood it with your confederates, right? You have to do very legal things, right? You need to pay the participants, you need to pay the moderators, you need to pay the expert. But those are things that are easier to track than flooding with your confederates.
So I would justify that way and not on representation until we figure out how to push up the acceptance rate.
Alessandro Oppo (45:18)
Yeah, I was not expecting this, to be honest. I also thought that giving more money would raise the participation.
Paolo Spada (45:25)
It's a fascinating show that paper
in the general petition. I said, OK, because I had a discussion with Kyle Redman in a a blog and he said, no, but the problem of a sentence rate is solvable. It's more money and institutionalization make it mandatory. But I already knew that make it mandatory doesn't solve it because we have pulled the work on the juries in the US and the juries in the US are not representative of the population.
because are highly skewed, because there are ways to get away from mandatory stuff, And so I, and the money was still open. Again, it's one data point. So it might be a very unusual case. I talked to the implementer and asked them why, because I don't know, 3000 euros, because the argument, oh no, I want to stay with my kids. Yes, but you know, I pay baby
and then we go doing an amazing vacation with 3000 euros. So I was expecting that at least 50%. By the way, we did 50 % in a citizen assembly acceptance rate on what? On Brexit, right?
So when the topic is in the mind of everybody, you get to 50%. The other person that consistently does 70 % plus acceptance rate is Jim Fishkin, but he spends 200, 000 euros plus on engagement and he uses the same teams that use marketing research that call people a million times and they really call and call and call and call.
And instead, the standard approach that Citi is, but also Certisium Foundation, that is the kind of the biggest provider, is to send one letter.
Alessandro Oppo (47:23)
And I was thinking in relation to political parties because we said that online citizen assembly could work and sometimes also better than maybe offline citizen assembly and so associated to the concept of scaling like because you know also in Italy there was the five-star movement and
and they tried with the Russo platform back then maybe the technology was not very mature like AI didn't exist in the way that we know it and so I'm thinking like how could be like a possible way of scaling in relation to political parties because as just
Paolo Spada (48:11)
So the political
parties.
Alessandro Oppo (48:15)
No, just another thing, because at the moment, citizen assembly participatory processes are usually held in a specific moment. maybe, what about something more continuous, where people can participate always?
Paolo Spada (48:32)
Yeah.
There are now experiences of continuous citizen assembly.
in Italy in theory there is the Milan one, I don't know anything about it, but it's so invisible that I'm doubtful that it's doing much, but I don't know, maybe it's going. The classical experience that everybody cites is the Belgian one, right? And so there are these institutionalized citizen assembly that occur every year. But within political party there is still very little. There is one fascinating thing is that
The researcher I was working in 2012, the one that brought the opportunity to do this experiment with the PD, is the first researcher to talk about doing citizen assembly in political party. He had this terrible name about it. He called D'Oparie. This is Raffaele Calabretta. He has a book about it. And he had a very interesting and sophisticated idea about it because he has been working in political parties for a long time.
time as an activist, his standpoint is that the citizen assembly within a political party was thinking more about deliberative polls and citizen assembly, but the idea is the same, where should be used only once a year and on a topic that really divides the party. Because if the party is not divided on that topic due to the fact that the
parties competing outside with other parties, it is an inefficient system. So it's only when you are truly in doubt and in the left, we are often divided and in doubt among many things, right? So for example, you can think about the, when the, the, the, the, Olivo that it was this sort of less constellation of parties that combined the Catholic and the traditional
former communist was coming together and they had to decide religious issues or moral issues. And for Raffaele, that was a key moment to have a citizen assembly because the leadership was not able to come to a decision. Like his key motivation is when the leadership is in conflict of interest. And that comes actually from the origin of citizen assembly. The way they implemented the citizen assembly in British
on the electoral law was the recognition that the political parties are in conflict of interest when they're deciding the electoral law. so I think that these reflections have disappeared a little bit. Now we tend to do citizen assembly on absolutely anything. And in my opinion, that weakens the instrument. It's only when there is a clear failure
of the politician and the smart politician recognized that themselves, then at that point, the Citizen Assembly can have big impact. Right. And you can think about the Irish case also that way, because in Ireland, the Citizen Assembly that really worked were the ones on same sex marriage and interruption of pregnancy. And those were issues that were stuck.
and that the political system could not solve. And those are the ones that were unstuck, but they discussed a bunch of other things and those did not do anything. They had a lot of impact among the participants and a lot of other impact, but they didn't have change in laws and like the classic public policy stuff.
So for applying internal democracy, it's really important to understand, and it's probably similar for firm, I don't study the firm, but that the agent you're thinking about, the institution you're thinking about, has this survival fundamental objective function through election. And so your citizen assembly needs not to conflict with that survival, need to help.
need to be there. But it's really an underexplorer space, in part because party members don't want to do them. We just did...
a fascinating conjoined experiment in the UK before the election. And funny enough, Labour put in their election manifesto that they wanted to do a citizen assembly. But then when we did a conjoined on party members and party higher echelon and we asked them what is the configuration of internal democracy, they don't like citizen assembly. They don't want they want only to increase the power of
members.
So increased direct democracy, increased oversight. And so that's fascinating, also intuitive ex post. But ex ante, we were expecting a completely different result because Labour was campaigning a bit on more citizen assembly, but for the public.
not for the party. And so it's still a conception of democracy as mobilization, very similar, if you want to the traditional leftist conception of participation. Participation is when you mobilize for me and you come and...
and you protest for me or you do the things that I want. If you're mobilizing in general, I'm a bit skeptical. But yeah, did I only talk about parties? Yeah, I didn't talk about scaling up. You were interested in scaling up.
Alessandro Oppo (54:42)
Yeah,
I was wondering, yeah, in relation to political parties, mean, scaling can be also outside of...
Paolo Spada (54:51)
Yes, there, like
if you look at the discussion in political parties about membership,
and the literature of membership. in Italy, we have the experience of the primaries of the PD, the Democratic Party that I liked it really easily. There is a tension about involving people in the Maxi public that are not members of the party and people within the party. And some people are really in favor to open up primaries to everybody, but some people are
scared
not because the opposing party will come in and hijack, but because it's corruption. You can buy and there were lots of scandals about it. Like you could buy a seat in a primary at the local level with 50, 000 euros. Like somebody have done a study with the prices, right? Because you literally buy people to go for vote for you in the primaries. And there are not enough checks because the level of checks are inferior. But also when
You remember when Renzi won, there was a lot of discussion because a lot of people that vote for Renzi were not the members of the party. Similar when Elly Schlein won. Elly won with the people outside of the party vote. And so there is that tension when you're thinking about political parties of who is the person that should have the decision on the leader.
And when we did the conjoin, that was one of the questions. And we were doing the survey experiment only with members of the party. And they were completely against open primaries. This is the UK. It doesn't really have a tradition of primaries anyway. But it would be, think, within the Democratic Party in Italy, the debates on primaries keep recurring.
Alessandro Oppo (56:44)
you
Paolo Spada (56:57)
divided, right? And often there was this joke that the PD is the only party that consistently loses primaries, right? Because it's always the candidate from the outside that's from the green or from Possibile or from something else that wins the primary. And so you can see why the party members might be annoyed, right? We have been volunteering for the parties for 20 years and there is this new outsider that comes in.
and wins the primaries with the pros and cons that that involves.
Alessandro Oppo (57:33)
Yeah,
of course. so, I mean, there is this tension between, let's say, the leader of the party that, I mean, the party itself, that maybe would like to theoretically engage with citizens, but then maybe they don't like what citizens propose. So...
I'm wondering also how to solve this issue because I see it's very related to scaling.
Paolo Spada (58:08)
And I
think so the participatory budgeting has a lot of sort of funny techniques that have been experienced because it's one of the oldest, right? There are now what 30 years? 1989 were the first 2026. Yeah. And so, for example, one of the thing that is often proposed when you are introducing PB in a highly conflictual environment is the fact
that people can criticize but they always have to propose a solution to their own critique.
And so you can have literally procedural tricks and the literature on conflict resolution is probably the oldest mediation literature. the facilitation, I'm interested in facilitation too. The facilitation literature in deliberative democracy is standing on the shoulder on the conflict resolution and mediation literature in psychology and in law.
It comes, it's older that stuff.
and has been done more for a variety of reasons. so conflict resolution has a lot of procedural approach to reduce. But the issue is to invest in them, right? And also to involve people in a credible dialogue, because people get annoyed when they are involved in a non-credible one. And when you ask them to participate a lot and then nothing happens, then they get for you.
And so, for example, when I consult for a citizen assembly and I know that it's a consultative process, my first conversation is about this and is to say, look, this is a consultative process. I need guaranteed of impact. Otherwise, we should stop here. Right. If you are discussing at the end of the year the transportation program of the city and you're willing
willing to change it because that's the key, not that you're discussing it. If you are willing to change it then and there are not massive political costs in some changes because that's the issue, right? If I've and then it makes sense to open up the discussion about transportation because it would have a little bit of feedback loop, not at the point of participatory budgeting in which you ask for the dog park and two years after
you're taking the dog in your dog park and there is a plaque that says this was for the project that Alessandro proposed, right? That's where the feedback loop shines. But to the point that I participate, I study, I come two, three weekends online, I organize, I write because now they want the participant to write a lot. I do a lot of work and at least
there are some changes that the politicians are willing to think about and at least discuss in a positive way. That to me is the sort of entry point. If they tell me no, I stop and I just leave. But the thing is now there are some topics in which is easier than others.
In a sense, the explosion of certain topics in citizen assembly is due to the fact that this
type of closure of the feedback loop, it's easy on those topics. Like climate change assemblies, now they're changing again, but at a certain moment they had so much explosion, in part because people were righteously pissed about the situation, but in part because the city was already doing a lot of the things that they knew the citizens were going to propose.
Right. And so if you look at what the citizens propose, there is very little innovation. Because in the end, if you think about the big, the really big radical changes are not in the end of the cities.
They are in the end of the production of energy. The transportation decisions are often higher. Healthcare decisions are higher. And so what the city can change is like very little things. And the city is often very open, particularly because I don't know, maybe there are some, but in my experience in the UK, the climate assemblies were all proposed.
by labor government or leave them. They were already very pro-pedestrianization, very pro-providing bicycle, integrated transportation, all the things that within that frame the citizens would support. So it was a little bit, it was an important process of education visibility. It had huge, I think, sort of expansion and that's why the NOCA network does the entire work on this, right? And it's amazing work and we learn a lot.
but it was also a part of less resistance. Nobody has done a study of it again because it's not a sexy topic and it goes against the push of the movement, right? And so criticizing it would be a bit problematic.
But everybody knows it's these funny things, right? Everybody kind of knows these things and a few others. So it's a bit of a weird action research situation.
Alessandro Oppo (1:04:12)
And what could be done to improve scaling? And also because you mentioned some conflict of interest and also in the field people are not always happy when there are crises.
Paolo Spada (1:04:29)
So
I think paradoxically scaling it's easier than the conflict of interest stuff. The difficult stuff is always the politics. There are only political solution to political problem. There is not a technology solution. For scaling, I think...
we are at the point in which if we had enough money, we could scale it up. And Elend talks about it, but others have talked about it. were experiments in the seventies, right? It's just...
Even in the 70s, we could have done a thousand micro-citizen assemblies in a thousand localities across Italy and figure out with a lot of work, manual work, how to integrate them. And if you think about some of the process around the Concilio Vaticano II, the Second Vatican Council, in which all the church across Italy had deliberations about, they were not high quality deliberation,
but
we had capital, capital deliberation of what the former communist party and even the democracy Christiana used to do. They used to have deliberation everywhere, right? And so we could think about doing that in person or online and having like a thousand small cities in assembly that are low cost, that are not expensive. They didn't ask a lot of the participants.
and way to integrate them. And AI, I will be really careful, not because I'm skeptical of AI. I think AI is reaching very good level. The problem is autonomy. If we don't a national AI model that is explainable and that the organizer of the Citizen Assembly understand, we are outsourcing epistemic framing
to big tech in the US, right? And that's extremely problematic. And that is obvious if you use DeepSec. If you use DeepSec and you write something about the Tiananmen protest, it disappears. And obviously, that's obvious, but the biases are there. And there are like biases about the type of source that are recommended.
the type of language that is used, all sorts. And again, we need full control. I would never suggest to use a non-locally installed AI to do this work for a serious political process. Imagine if the Citizen Assembly is used to discuss security issues of Italy or France, right? And the data is in the US.
And before we could delude ourselves that, okay, the US was a good benevolent policeman of the world, but now it's obvious to everybody that we're just deluding ourselves on what they're operating, right? And so that's the key issue around AI. The model will improve. It's also a model that will always contain randomness, right? And so...
it will always need cross checks because by the way it is created, it has randomness on purpose because otherwise it would be a deterministic model, right? So some people describe it as a loaded dice. The loading become better and better and better and better and predicting the next word, predicting the context, but it is a dice, right? So even when you're summarizing arguments of citizens,
you're summarizing them with a random error. And that at that point enters into discussion on the amount of risk that we are willing to have on that discussion. And so if it is a local discussion about do we have the bike lane here on there and we can involve all the minorities that usually do not participate by using
an AI tool, I would probably in the cost benefit analysis say, yes, we had an error, but we are involving this group like the Italian minority, they never participate. And so the cost benefit analysis is a yes. But if we are discussing, I don't know, a topic about a really sensitive topic, like a new law on
sexual violence or whatever in which the voice of victims is fundamental and it needs to be reported perfectly and it cannot be modified in which biases are already present in almost everybody, right? At that point, I would probably say no AI, right? Or
AI with the level of heat very, very low so that it becomes a document retrieval system. Because a lot of people think about AI only as LLM, but there are a million others of machine learning approaches. And there are some that purely doing document retrieval and they have less randomness. And then there is a human that does the summary, does the cross-check. So the technology is there. The problem is ownership.
It's transparency and those right now, if you look at the implementation of most AI, they're using commercial products, right? And that's highly problematic. As when they were using Zoom during COVID to do high level citizen assembly, I was the only one saying, look, Zoom is not GDPR compliant, right? Like we're talking about accountability transparency.
And so, but there is, I don't know, there is still a lot, we're still in the infancy, if you think about it, like the first one was done in 2004. And so it will take time, but the direction is very promising, right? And,
Alessandro Oppo (1:10:52)
Yeah.
Paolo Spada (1:11:05)
And even an important thing that I keep repeating to everybody is that we need to forgive ourselves, right? Because the alternative is really shit. And so even if this stuff has a lot of problems and a lot of mistake, it's better than the alternative. And so, yes, the job of academics like me is to go nitty gritty and super critical on every aspect.
But then the job of the practitioner and implementer is to tell me, go away and do my research and do a cost benefit analysis and say, no, in Bologna now we need a citizen assembly because it's going to break an impasse like in Ireland. We need to try to institutionalize. Even if we know the challenge of institutionalization, we need to try to do it. Right. And that's the action imperative. And that I absolutely recognize and defend.
But I have an academic imperative, right? And so my role is to still figure out how to do it better. I'm very happy now that I'm older to do embargo strategies, to do research and development strategies. When I was a young postdoc, I was way more Taliban in a sense on this. I wanted radical transparency everywhere. And now...
Like the political contest is tough. Democracy is retrenching, right? And so we really need to be extremely careful, extremely careful with critique because they're not gonna care. the people, if the readers of my papers are those that want to destroy Citizen Assembly, I'm not gonna write those papers because I want to improve Citizen Assembly, right? Those paper will remain.
but will be given to those that want to improve city assembly and I'm not gonna circulate them in the public. I was joking, I wasn't in the department at Yale in which a lot of my friends were specializing in violence against civilians, civil war, dictatorship. And we were joking that the only one reading their papers were actually the dictators and those that wanted to repress.
civil society organization because by describing in detail the way civil society could do things, you're also unfortunately providing ideas on how to break down those tactics. Right, so when I'm describing ways to improve citizen assembly, I'm at the same time describing ways to capture it and destroy it. One of the fun exercises
practitioners do in conferences is to ask the audience, how would you capture a mini public, right? What would be the best approach? That's a, if you have never done the exercise, I'm talking to the audience directly right now, think about it. I do it with my students all the time. At Yale, I did an exercise in which I asked how to overthrow the US government. Then I got an email from the head of the department.
That's too much, Paolo. Stick to citizen assemblies. Nobody cares about them. If you're teaching how to do a coup in the US, I get an email from angry administrators. But yet sometimes, and this is, I think, important for action researchers like me, we often think that nobody reads our research, but it depends what you're doing.
And sometimes you are thinking about an audience, but there is also the opposite audience. And it's very important when you do action research to think about the opposite audience, right? It's really, really important. If you are in action research and you want these democratic innovations to become stronger and better.
Alessandro Oppo (1:15:19)
My last question was about if you have a message for practitioners and people in the field but in some way you already gave some suggestion but if you have some others
Paolo Spada (1:15:32)
I think that the...
The message that the message and I keep repeating it is that we are not a minority group anymore. And so I still see a lot of allergic reaction to critique, a lot of evangelism. And to me, that was a position that made a lot of sense when nobody was paying attention to us. Right. But now there is a shift and it's a complex shift. Right. Because we are actually
might need now serious autonomous impact evaluation when it's the European Commission that is doing citizen assembly. And it's not a problem to say that some of these processes are actually highly problematic. And I don't want to criticize the commission. It could be a city or somebody else, right? And so in the moment, there is a moment in which you have grown enough in which there should probably be a reset on how much we are open.
to critique. And so that would be something that probably would be my message. I think we have reached that moment. And so it's important to start thinking larger if you want. We are not a minoritarian movement anymore. Democratic innovation are here to stay. There are a million that are occurring everywhere. And it's a moment also to reflect on
on the problematic ones and when we should use some and when we should use somebody else. It's not always helpful to do it, right? But that reflection is really hard to make because in part some people livelihood depend on selling democratic innovation, all the industrial facilitation of participatory service providers, right? But we have very few reflection on the indices, a couple of books.
but it's an important reflection. But also the counter, right? I was discussing commodification and the problem that this industry is generating. There should be voice from the industry defending actually commodification because commodification that we see as negative most of the time has an element of standardization that allows
cities that don't have enough money to buy a custom process that is way more expensive, to buy an off-the-shelf participatory process that is very cheap, that is outsourced, and so they start their path into learning more about democratic innovation. And so I think these types of really grown-up discussions around commodification are very rarely done, like you have sort of the Marxist tradition that's completely against commodification.
it's terrible, it ruins everything. And on the other side, people don't even engage, right? And we don't care, you're crazy communist. And it's very important just to do as many as possible, right? The objective function of the industry is to do as many as possible. But that's an important discussion to have, but it's not easy. And again, it's the role of academics, right? I think academics should implement less actually. The implementers are better than us.
and implementing. It's good to help and observe and evaluate, but I think it's a good time to have like really redraw some of the boundaries, right? And it's important to co-design and discuss together, but also there are things that we are better at and there are things that they're better at. And it's important to recognize those.
Otherwise there is a naive view of co-design or sometimes it's called democratization of democratic innovation and he has his own risk. I think it's very important to do co-design and meta-deliberation. I'm one of the pioneer of that stuff, but it rarely works. We also have to recognize that it's really hard. We just came out of an horizon that tried to do 11 meta-deliberation, not a single one.
and we had a lot of money to throw at it. It didn't work for a million political reasons, really does. Meta-deliberation is super hard.
So yes, that's my much, Alessandro. And if you want to another one in Italian, if that is helpful as a spin-off, why not? I'm happy to have it. Or a more targeted one, because this was a little bit of flux of wanted to interrupt me.
Alessandro Oppo (1:20:12)
Thank you a lot, Paolo.
Yeah... but-
Member discussion: