In Conversation with (Redux): Paolo Spada

Over several years, beginning in 2015, PlaceSpeak presented a series of Q&A with experts in public engagement and civic technology each month.  In 2023, we circled back to ask those experts for an update on their original contributions.

Next up is Paolo Spada, currently a Lecturer at University of Southampton focused on Collective Intelligence, Participatory & Deliberative Processes, and Methods, from his original Conversation in March 2017.

Paolo’s work explores how to optimize participatory systems, complex platforms that combine a variety of digital and in-person participatory processes. Paolo is a member of Participedia (CA), of the research board of the Participatory Budgeting Project (USA), a member of the World Bank’s Digital Engagement Evaluation Team, and the Democracy Matters team (UK). His current projects focus on how to promote better and more meaningful consent forms in digital engagement practices, gamification, and how to improve comments in online news by adapting deliberative practices.

Q. We had our last discussion in 2017, what have you been up to in these 6 years?

A. The last time we spoke I was in the midst of the EMPATIA project, I was living in Portugal. Since then, I moved first to Italy for a year at GSSI, and then I found a permanent home at the University of Southampton within the Rebooting Democracy project.

Q. Tell us more about the lessons of the EMPATIA project?

A. EMPATIA was a typical big multi-university project that generated a lot of innovative ideas and numerous prototypes and an interesting platform that had a spike of adoption during the lifetime of the grant and then slowly disappeared.

The main lesson of EMPATIA for me was a design philosophy lesson, and it emerge in comparison with platforms such as CONSUL and DECIDIM that were developing at the same time. EMPATIA was a platform designed to do everything from the start, it was designed for participatory and deliberative systems and maximum flexibility, but it was ultimately a top-down project led by a bunch of crazy academics and an equally crazy developer that fall in love with the project. Some of the ideas of EMPATIA were simply too advanced, and are still too advanced for most real-life political applications.

CONSUL, and later its fork DECIDIM, instead started from specific community processes, mostly the Spanish model of PB, they were initially extremely simple, and frankly not very good given that the Spanish PB model is not particularly strong (see Ganuza, E. and Baiocchi, G., 2012) and they could only do a few processes with very limited flexibility. EMPATIA was born in part as a critique of these first generation PB platforms that were adding technology constraints on top of the already existing political constraints to existing democratic innovations. They were basically telling participant that PB was their very restricted version of PB, when instead as we know democratic innovations get constantly redesigned and reinvented to better adapt to local conditions. However, behind the latter two platforms there was a political project and a proper community of developers that slowly expanded the platform improving it and making it more flexible and secure. Both platforms were adapted to many use case scenario and became flexible overtime.

I think these are two different strategies with different pros and cons. The top-down strategy might generate more innovation and can leapfrog current cutting-edge technology, but it can also become disconnected from reality and lack a market for the final product. EMPATIA was like a Lego system for participation, but in most applications, it was used with the exact same combination of limited pieces because political constraints prevented the creation of complex deliberative and participatory systems that currently only exist in academic dreams.

The second strategy is more realistic and generates a product that meets a clear need. It starts from a limited and constrained platform, but as we have seen in Spain, if the community learns and develops, it can generate more innovative and flexible products. Decidim is now a very strong platform, particularly for large-scale processes.

The diffusion and success of platforms designed with the first approach depends on the academics’ ability to obtain grants and encourage adoption. EMPATIA was adopted by around 50 cities during its first life, then shrunk down to a few places in Portugal. However, it is being reactivated because we won another grant last year and now, we will promote its adoption again in the PHOENIX project.

In the second case, the diffusion of the platform depends on the political support of cities that fund the development communities. CONSUL effectively disappeared after a change in the coalition in Madrid, but hopefully, the DECIDIM foundation will figure out ways to support its community when the current coalition loses the elections. DECIDIM currently has one of the largest numbers of installations in the world due to the aggressive marketing of the city of Barcelona, the quality of the product, which is now one of the best, and the creation of an ecosystem of firms that support the installation of the platform around the world. We hope that this combination of ingredients will support its resilience.

Q. Any other lesson from the EMPATIA project?

A. Another interesting lesson from the EMPATIA project was the difficulty of effectively interfacing social scientists who are experts in participation and technology developers. This is a common mistake that occurs in interdisciplinary teams without significant experience in technology development. My lesson is simple: translating social science requirements into tech requirements is the most challenging part of the project, and therefore, it should receive the most time and resources. Since 2017, I have been involved in several projects where not enough time and resources were given to this translation because both parties underestimated what was easy and what was difficult. More and more, my role in projects is to try to translate these languages and slow things down at the beginning and require more testing. Everything else is easy, the combination of languages is extremely hard.

Q. What trends do you see in the field of online public participation?

A. I think there are 5 main trends some started before covid and developed during covid, some other instead are almost a direct consequence of covid:

1) The stigma against online deliberation has disappeared

Prior to COVID, online deliberation had significant stigma surrounding it. I remember being one of the few researchers publicly stating that I had seen online deliberations, based on well-moderated forums or argument maps, work better than face-to-face ones. I also believed that the media was only part of the story and not the most important factor. However, the pandemic forced most countries to move to remote schooling via digital platforms, resulting in a better understanding of the pros and cons of digital deliberation among the general public. This completely altered the culture around online dialogue, which is now significantly more accepted. In the UK, for example, many small dialogues similar to Citizens’ Juries are now conducted online for convenience, and with high-quality moderation, the main drawbacks of online deliberation have been reduced. However, there are interesting new areas of research relating to the impact of commercially available platform designs on dialogue, such as the impact of constantly seeing ourselves while talking.

2) Citizens Assemblies fashion is generating a backlash

Currently, we are witnessing a trend where everyone wants to call their process a “Citizens Assembly” (CA), similar to the Participatory Budgeting (PB) trend of the early 2000s. This is due to the success of the Irish process and the visibility of the French processes. However, as with PB, the original CA model that emerged in Canada with the British Columbia CA has been “decaffeinated”. Processes that are as short as two weekends and have no institutionalisation are being called CA’s, when the original model was a system that combined a year of meetings of a randomly selected assembly, 56 meetings open to all citizens, and a referendum. This phenomenon is almost inevitable for successful innovations, but we know from the experience generated by the PB case that it generates a backlash, and only after the backlash do things adjust to a novel equilibrium. To my knowledge, all PB’s implemented before or in 2000 around the world have been abandoned. I recently discussed the problem of selling CA as leveraging a representative sample of the population in a blog post with Tiago Peixoto which you can find here.

3) We are close to have AI powered argument mining technology that will allow novel e-deliberation platforms

I have been working on argument mapping in collaboration with Mark Klein and others since 2011 (see for example: this experiment we conducted in 2012 with the Italian Democratic Party), and now I am quite excited about the recent developments of AI and how we can apply it to the field. One of the main challenges of online synchronous (video-based) or asynchronous (text-based) deliberation is managing large amounts of information and transmitting it across small groups. For example, if we were planning to organize an ideation process with 1000 people, we could divide them into 100 groups of 10 participants, and each group could generate high-quality deliberation. However, the transmission of information across groups can be problematic. If each group is asked to come up with 5 solutions to a problem and a few arguments in favor and against each solution, the resulting 500 solutions and thousands of arguments would take weeks, if not months, to summarize. Furthermore, groups often provide very similar solutions and arguments with minor variations.

Collaborative argument mapping platforms such as Deliberatorium emerged more than ten years ago as a possible solution (see my paper on the topic). The argument map naturally organizes the discussion, ideas appear only once, and argumentation  s promoted. However, they have a steep learning curve, require moderation, and people make mistakes. ChatGPT 4 is highly effective at extracting arguments from text. Therefore, we are already experimenting with asynchronous text-based discussions that are then transformed into argument maps by artificial intelligence.

What is important to note is that ChatGPT is more creative than what we need, yet it is doing quite a good job. Therefore, when a more refined AI specialized in argumentation emerges, we will have even better performance.

In summary, I think we are very close to seeing a mix of video platforms for deliberation that automatically transcribe the participant discussion and automatically generate an argument map that combines similar ideas. Such technology will be a game changer to scale up deliberation to levels we have not seen before. The platform will still require moderation, but the number of moderators will be significantly reduced, and the time required to harvest the content will also be reduced.

4) It is surprising how little interest privacy concern generate in most countries, are we in equilibrium or something will change?

Only a few countries have managed to educate citizens on the importance of data privacy. In the countries I operate only Germany and Switzerland seem to have generated some cultural movement. In all other countries there is almost no interest from participatory service providers in using digital platforms that have strong privacy protection and most online deliberations are done via zoom or MS-teams, quite a few cases in which the data is managed in a way that is not GDPR compliant making effectively a process introduced by a city or a national government illegal according the laws of such country. I am not sure if we will continue this way forever, or if some sort of big litigation case will emerge that will generate a tectonic shift in the landscape.

5) We are still summing processes together in an attempt to solve the democratic malaise, are we in equilibrium or something will change?

I started exploring the idea of participatory systems in my dissertation between 2009 and 2012. To my eyes Brazilian PB was a complex system, particularly when it was struggling to combine online and offline channels of engagement. In 2012 I published with Allegretti a paper about multichannel democratic innovations that was fleshing out the limitations of the emerging aggregative approach to participatory systems. We were drawing from the experience of Brazilian PB’s and many other examples in which channels of engagement had entered in conflict instead of reinforcing each other. The EMPATIA project we won in 2015 was a direct continuation of our work on multichannel democratic innovations and participatory systems. EMPATIA was originally intended to introduce real time optimization strategies for participatory systems. What is interesting for me is that the mainstream approach is still more or less the same, cities are still multiplying many weak channels of engagement in an attempt to promote inclusion and visibility, instead of trying to create less processes, but more impactful ones. As with the privacy concern described in the previous point I do not know if we are stuck in some sort of stable equilibrium, or if we are getting close to some major shock to the system that will generate an abandonment of the current approach.

Q. What are you working on at the moment? 

A. At the moment I have a few concurrent projects. The aforementioned Rebooting Democracy is the umbrella project at Southampton that investigates how technology and in particular AI can be used to enhance democratic innovation. We just completed an online dialogue on sentencing that included an experiment that was testing the possible usage of argument mapping in video synchronous discussion as a note taking device, we are supporting a series of participatory and deliberative processes in Southampton itself, and in a spinoff called DEMOPLAY we are exploring the role of gamification in democratic innovations. The latter again is a continuation of work initiated during EMPATIA when we developed EMPAVILLE and we started working on Democracy as a serious game. Then I am working in a new project called PHOENIX that aims at adapting democratic innovations to better support green transition discussions and their trade-offs. In Phoenix we will evaluate 11 different pilots across Europe. And lastly, I continue supporting the Participedia project and the school of Collective Intelligence in Morocco.

Know someone who we should feature? Leave us a comment below or tweet us @PlaceSpeak.