Online participation. A three-dimensional approach to study digital political platforms

Introduction

This article departs from an old but still central question in the social science literature: how extensive is participation in internet-based political processes, and how should these tools be categorised in political science? Since the early debates on “electronic democracy”, scholars have proposed typologies that map platforms onto large normative families such as deliberative, direct, liberal, or neo-republican democracy (Dahlberg, 2001; Päivärinta & Sæbø, 2006; Pelloso, 2020; van Dijk, 2000). While this work is indispensable, it tends to stabilise a few big political categories and to overlook the more fine-grained sociotechnical arrangements through which participatory platforms operate. Empirical research, by contrast, reveals a wide spectrum of online participatory embodiments, from purely consultative portals to devices that reconfigure party organisation or urban policy-making, and shows that similar labels may hide very different practices and power relations (Deseriis, 2020; Doustaly, 2019).

Historically, online political participation emerged in the mid-1990s alongside the diffusion of the internet as a mass communication medium (Berg & Hofmann, 2021)¹. Early accounts emphasised its potential to lower participation costs, link citizens and representatives, and expand deliberation (Grossman, 1995; Rodota, 1999; van de Donk et al., 1995). A more sceptical literature soon stressed continuity with existing hierarchies, the resilience of digital divides, and the integration of online tools into traditional repertoires (Hindman, 2008; Margolis & Resnick, 2000; Norris, 2001). Since the 2008 financial crisis, however, the rapid digitalisation of politics and the rise of “digital parties” (Barberà et al., 2021; Gerbaudo, 2019b) have made participatory platforms central to intra-party decision-making, urban governance and social movements, renewing the need for analytical and evaluative tools.

This paper brings together two strands of scholarship that have often evolved separately and approaches them with a third strand. The first develops typologies of digital democracy and e-participation (Dahlberg, 2011; Päivärinta & Sæbø, 2006; Pelloso, 2020; van Dijk, 2000). The second proposes ladders or models to assess degrees of participation and decision-making impact, from classic work on citizen participation (Arnstein, 1969; Fung, 2006) to more recent indicators and meta-syntheses proposing evaluative frameworks of online participatory platforms (Kneuer, 2016; Santini & Carvalho, 2019). In parallel, a growing body of research introduces science and technology studies (STS) into the analysis of participatory platforms, focusing on the interdependence between sociopolitical and technical levels (Badouard, 2014; Deseriis, 2021, 2023; Moats & Tseng, 2023).

Building on these three strands, I propose a new comparative tool, both analytical and normative, to assess participatory platforms along three axes: (1) the relationship between organisers and online participants, (2) software architectures and affordances, and (3) infrastructural underpinnings such as servers, cloud services, and data management. In this article, I explicitly take their degree of participation as a central problem: how open are these platforms? What are the limits of citizen deliberation and decision-making in political, software, and infrastructure matters? The interest of such an approach is not to replace existing large categories, but to situate them at different stages or configurations within a multidimensional space. Epistemologically, the multiplicity of frameworks is not necessarily a problem, as long as it allows researchers and practitioners to see the same object in a prismatic way.

To propose such a three-dimensional framework, I extend the recent dialogue between political sociology, communication and digital sociology, and STS. While Badouard (2014) and Deseriis (2021, 2023) have already theorised and practiced the junction of participation studies and STS by looking at the different sociotechnical configurations of online political participation, the material and institutional underpinnings of these platforms (how they are hosted and governed) remain largely invisible. At the same time, platform and infrastructure studies have shown how actors such as Google or Facebook operate as infrastructural layers in public life, opening up a strand of research on the standards, hardware, and social worlds that turn around digital underpinnings (Plantin et al., 2018; Srnicek, 2017; van Dijck et al., 2018). Applying a similar lens to participatory platforms, this paper contributes to a reorientation that examines the politics of server dependencies in online participation: how infrastructural invisibility is produced, what normative choices are embedded in infrastructural design, and how these choices condition the legitimacy and effectiveness of participation (Li Vigni, In press).

This orientation responds not only to academic debates but also to concerns expressed by practitioners and activist-researchers. As Antonio Calleja-Lopez, political scientist and co-founder of Decidim Barcelona, has argued in one of our exchanges about the Catalan platform, the problem today is less the lack of diagnoses than the absence of robust evaluative devices capable of distinguishing “participation-washing” from high-quality processes, and of equipping institutions and social movements with practical benchmarks². While activists like him may also have a rhetorical (besides an analytical) interest in developing frameworks and indicators, the literature review by Santini and Carvalho (2019) shows that most existing evaluation frameworks for online participation are thin, state-centred, and rarely assess whether platforms actually influence political decision-making, with around 70% of the 179 cases they review lacking any measure of decision impact. They respond by proposing an “Online Political Participation Assessment Model” that distinguishes government- and citizen-led initiatives across six types of participation, and they find that only a small minority of platforms achieve meaningful collaborative or deliberative impact rather than mere “participation-washing”. Facing such concerns, this article offers a new tool for a more nuanced, interdisciplinary assessment tool, which includes the governance of the platform itself.

Empirically, the study is based on forty qualitative interviews of around one hour each, conducted between 2019 and 2025 with actors directly involved in the design, deployment, or promotion of online participatory tools. Interviews were carried out in person (Paris, Barcelona, Berlin, Copenhagen) and by videoconference. The sample is deliberately diverse in terms of geography, institutional affiliation, and professional role. The interviews cover actors located in Barcelona, Boston, Brasília, Brussels, Buenos Aires, Copenhagen, Geneva, Graz, Helsinki, Mexico City, Milan, Monterrey, Paris, Taipei, and Yaoundé. Interviewees work in municipal and regional administrations, public innovation agencies, political parties, NGOs, and civic-tech companies. They include software developers, platform designers, participation consultants, public servants, civic-tech activists, and NGO directors. The interviews were semi-structured, ensuring coverage of core themes while leaving room for open-ended accounts. This material underpins the small typology developed in the article and is used illustratively through quotations that specify the interviewee’s role, city, and platform affiliation.

The remainder of the article proceeds in two steps. The first part sets out the three-dimensional framework and introduces a comparative cube that locates participatory platforms along the axes of political governance, software affordances, and infrastructural dependencies. The second part applies this cube to four paradigmatic cases – Italy’s Rousseau, Barcelona’s Decidim, Argentina’s Consulta Pública, and Taiwan’s vTaiwan – showing how their different sociotechnical configurations help explain the variable effects of online participation.

The comparative cube: conceptualising a framework for the assessment of online political participation

According to Google Ngram, the terms “cyberdemocracy” (Kaczmarczyk, 2010) and “digital democracy” (Hacker & van Dijk, 2000) appeared almost simultaneously at the beginning of the 1990s, whereas “e-democracy” (Coleman & Norris, 2005) emerged after 1995. From 2003 onwards, the latter term largely dominated the others, before being gradually caught up by “digital democracy” after 2020. By contrast, “cyberdemocracy” has remained a marginal term and has continued to decline since the 2000s. Here we choose to avoid these three synonyms in favour of the more generic “online participation” or “e-participation” (Steinbach et al., 2019), which makes it possible to encompass a variety of practices, including those implemented by authoritarian regimes (Li Vigni, 2025). Online participation can be broadly defined as the use of digital tools to allow participatory, deliberative, and decision-making processes in political governance (Congge et al., 2023; Santini & Carvalho, 2019). This definition encompasses a wide range of practices, from online petitions and participatory budgeting to crowdsourced policymaking (Li Vigni & Wojcik, 2025), implemented by tools often called “civic tech” including platforms and applications (Knight Foundation, 2013; Schrock, 2019). These tools are digital infrastructures that network people around common objectives of all kinds (economic, artistic, political, etc.) (Gillespie, 2010; Srnicek, 2017). Even though their promoters claim the will to disintermediate, platforms actually constitute a new kind of intermediary that seeks to constantly innovate in terms of the functionalities offered. In our case, political platforms do not sell products or services: They serve to create consensus and mobilisation, and can be deployed to prefigure and operationalise public action.

In order to answer the initial question about how extensive is participation in internet-based political processes, it is first necessary to identify the fundamental dimensions of the object under study. Participatory platforms are complex sociotechnical arrangements in which political actors, software architectures, and underlying infrastructures are tightly interwoven. The objective of assessing degrees and limits of participation in these arrangements cannot be adequately addressed from within a single disciplinary lens. Political sociology clarifies who organises participation and with what forms of authority; media and digital sociology help grasp how algorithms and affordances orient behaviour; science and technology studies (STS), particularly the strand on infrastructures, illuminate how technical artefacts embed norms and power relations. Only by combining these perspectives can we render visible the different layers through which online participation is configured and constrained.

Over the past decade, some works have already fruitfully articulated online participation studies and STS. Badouard’s “design-oriented” approach and Deseriis’s work on decision-making software both show how code stabilise different models of participation (Badouard, 2014; Deseriis, 2021, 2023). Yet, even in this literature, the material and institutional underpinnings of platforms – which cloud services are used and how they are governed – remain in the background, treated as neutral infrastructure rather than as politically charged choices. In the remainder of this section, I therefore review strands of scholarship that appear most useful for asking precise questions about online participation along three analytical axes that will structure the comparative cube proposed in this article: political governance, software affordances, and infrastructural dependencies.

Political governance: co-decisional, consultative, or informative

The first step is to question the type of governance that parties put in place through participation technologies, a dimension that can be analysed using the conceptual tools of political sociology and its debates on the renewal (or not) of representative institutions. STS has developed a rich tradition of research on technical democracy (Chilvers & Kearnes, 2015), but some authors have noted that, in its Latourian/ANT variant, the emphasis on flat ontologies and coproduction can make it harder to grasp more traditional questions of institutional politics (Marquet, 2017; Pestre, 2012). Hence the need to enrich this literature with political sociology works on democracy. Coupling participation studies and STS allows us to claim that platforms are not coded and used in the same way, depending on whether one subscribes to “libertarian municipalism” (Thompson, 2021) or “liquid democracy” (Bertone et al., 2015), or whether one uses elections or drawing lots (Sintomer, 2023).

A political sociology perspective raises several fundamental questions about platforms and their users. First, do these platforms foster deliberation and co-decision at various levels of political action – within parties, government, legislation, and budgeting – or do they instead promote plebiscitary dynamics (Gerbaudo, 2019a)? Do they allow citizens to shape both the content and the process of decision-making, or are these structures imposed from above (Deseriis, 2020)? Second, how do digitally active citizens interact with these tools, and in what ways do they appropriate and reshape them (Lefebvre, 2020)? Perhaps the most critical question is whether participatory platforms genuinely shift the balance of power between citizens and representative institutions, or whether they ultimately reinforce the existing status quo (Douay, 2016).

Software affordances: deliberative, interactive, or unilateral

As Badouard points out, “Web designers who create participatory sites build […] models of participation. By arranging different applications that encourage action within the same site, they promote certain activities and define a way of carrying them out” (Badouard, 2014, p. 36). In this regard, digital sociology provides the tools to detect how normative conceptions are part of and reinforced by algorithms (Grosman & Reigeluth, 2019; Lee & Björklund Larsen, 2019; Marres, 2017). In the case of online participatory platforms, this involves first listing the available functions (debate, voting, drawing lots, etc.), establishing the code policy (open, proprietary, etc.), and understanding who determines what to discuss and how (Deseriis & Vittori, 2019). The structure of digital tools can be analysed in relation to their effects on online debate, their influence on public policies, and the ways in which citizens reappropriate or contest these platforms.

A digital sociology perspective, according to which we fabricate the algorithms that fabricate us, raises several key questions about the relationship between platforms, citizens, and platforms’ promoters (Cardon, 2015). First, what aspects of political debate do platforms eliminate, modify, or enable (Esau et al., 2017)? To what extent do they allow for systemic critique – are there dedicated channels for citizens to express dissent or suggestions (Aragón et al., 2017)? Second, how do online activities interact with offline engagement, including demonstrations and assemblies (Mabi, 2019)? What kinds of demands and constraints do platforms impose on users, and how do citizens respond to the processing and structuring of their contributions (Bravo et al., 2019)? Finally, do citizens develop counter-uses to bypass platform constraints, and what forms of resistance emerge against the exclusions and biases embedded in algorithmic systems (Marchand et al., 2019)?

Infrastructural dependencies: private, public, or commons

Infrastructures, often relegated to the background until a breakdown exposes them, encompass the material objects, organisational procedures, and symbolic standards frameworks that underpin a vast range of human activities. They have been a central focus of STS, where scholars have sought to repoliticise them by showing both what they enable and constrain (Peters, 2016; Plantin & Punathambekar, 2019; Slota & Bowker, 2017). Far from being neutral artifacts, infrastructures are deeply embedded in, and constitutive of, different social and political orders. In the case of online political participation, the characteristics, management, location, and governance of servers may play a role in shaping the social existence of platforms. For example, bandwidth congestion during single-day voting events – a recurrent issue on the Rousseau platform of the Five Star Movement in Italy – often forced users to retry their votes multiple times before ultimately abandoning the process³. Such technical failures contribute to citizen disengagement and provoke public criticism regarding the platform’s reliability and the competence of its promoters (Biondo & Canestrari, 2018).

Infrastructure studies, particularly the STS strand working on internet governance, raise several critical questions about the relationship between platforms, service providers, and users (Epstein et al., 2016). How do platform operators interact with service providers, and to what extent do Internet users have control over the digital infrastructures they rely on (Musiani et al., 2016)? Do platform promoters opt for private solutions, invest in public infrastructures, or contribute to the development of digital commons (Dulong de Rosnay & Stalder, 2020)? Additionally, what challenges and obstacles do platforms face, including public criticism, legal disputes, and internal conflicts (Mabi, 2019)? How in the end do these factors shape the usability, credibility, and political effects of participatory platforms?

The comparative cube

The three-dimensional approach outlined above is designed first and foremost to address a specific research question: how do different participatory platforms compare in terms of the degree of participation they enable? In line with work on participatory democracy that emphasises both the promise and the ambivalence of the contemporary “participatory imperative” (Blondiaux, 2017), this article treats participation neither as an abstract norm nor as a self-evident good, but as a set of situated arrangements that can be more or less inclusive and influential for citizens. Participation is here operationalised as the bundle of opportunities offered to citizens to be informed, consulted, to deliberate, and in some cases to co-decide on public issues (Steinbach et al., 2019). The question “how open are these platforms?” is therefore made more precise: to what extent do they move beyond mere information provision, invite meaningful input, support horizontal deliberation, and effectively share agenda-setting and decision-making power? The comparative device proposed here has been constructed from this participation-centred question in the form of a cube – a tool already used in several disciplines (Frericks et al., 2018; Fung, 2006; Hitt, 1967; Hung, 2000; Kvist, 2007; Van Der Sluijs et al., 2008) – but it may readily be adapted to other evaluative questions, such as the degree of social inclusiveness or the extent to which contestation is taken into account.

Existing typologies in the civic tech field already rely on two-dimensional spaces, but they tend to foreground other distinctions. The Knight Foundation, for example, distinguishes civic tech projects according to whether they are oriented towards “open government” (transparency, open data, institutional participation) or towards “community action” (peer-to-peer collaboration, civic crowdfunding, neighbourhood problem-solving) (Knight Foundation, 2013). Alternatively, Mabi’s typology maps civic tech initiatives along two axes: the degree of institutionalisation and proximity to public authorities, on the one hand, and the ambition of social transformation (from incremental deepening of representative democracy to more radical reorganisation), on the other. These bidimensional mappings are highly useful for situating actors and strategies, but they do not directly operationalise participation itself, nor do they systematically integrate software design and infrastructural choices into the analysis.

In contrast, the comparative cube proposed here brings together insights from participation studies and STS to articulate three axes that are all directly related to participatory performance: political governance, software affordances, and infrastructural dependencies. On the political axis, I draw on a long tradition that distinguishes information, consultation, and more empowered forms of involvement (Arnstein, 1969; Macintosh et al., 2002). Concretely, this axis ranges from platforms used primarily for one-way information provision, through consultative devices that collect feedback on pre-defined agendas, to more co-decisional arrangements where citizens have recognised influence over outcomes (for instance, binding online primaries or decisive participatory budgeting votes). At the same time, I take seriously the critiques formulated by Chilvers and Kearnes and others against rigid, hierarchical models such as Arnstein’s “ladder of participation”, which can obscure the multiplicity and situatedness of participatory practices. In this article, the information–consultation–co-decision scale is thus used heuristically, as a way to order cases along degrees of citizen power, while remaining attentive to ambivalences and hybrid arrangements.

On the software axis, the focus shifts to the affordances of the platform itself. Here I build on STS-inspired work that shows how designers “build models of participation” into interfaces and functionalities (Badouard, 2014; Deseriis, 2021). In the cube, this axis moves from more unilateral configurations – where the main function is to transmit comments or votes upwards and horizontal exchange is minimal – to interactive platforms that allow users to react to one another but do not particularly encourage sustained debate, and finally to more deliberative environments where discussion tools (threaded debates, argumentation features, collaborative editing, etc.) are central and explicitly oriented towards disagreement, reason-giving, and collective problem-solving – what specialists of deliberation call “reciprocity” (Gutmann & Thompson, 2004).

The infrastructural axis addresses a layer that has been comparatively neglected in online participation research. In this cube, the infrastructural axis does not evaluate technical performance (such as bandwidth), but rather degrees of dependency and control: from private solutions where platforms rely on commercial, proprietary infrastructures; through public infrastructures operated by state or municipal bodies; to arrangements closer to digital commons, where hosting and development would be organised as shared, non-state public goods. While the literature on digital commons rightly emphasises broader aspects – such as community governance and non-commodified knowledge (Fuchs, 2021) – the present model isolates one specific infrastructural facet: ownership and business model of the underlying technical infrastructure. It should therefore be read as a narrow operationalisation of commons, not as an exhaustive treatment of the concept.

These three axes can be visualised as a cube (Figure 1), within which each participatory platform occupies a position defined by its political mode of governance (informative, consultative, co-decisional), its software affordances (unilateral, interactive, deliberative) and its infrastructural regime (private, public, commons-oriented). The aim is not to produce definitive labels but to construct a comparative space in which technopolitical arrangements can be classified and contrasted in terms of their participatory performance. In the next section, I use this cube to situate four empirical cases – Rousseau, Decidim, Consulta Pública and vTaiwan – and to discuss how different combinations of governance, code, and infrastructure shape the actual openness of online participation.

Operationalising the cube on four emblematic cases

I deliberately limited the analysis to four platforms in order to maximise variation in sociotechnical configurations (party vs municipal, Europe vs Latin America vs East Asia, different software and hosting choices) while keeping the comparative exercise intelligible and manageable in length. Including many more cases would have increased redundancy and made the visualisation of positions in the cube harder to read, without necessarily adding new combinations beyond those already illustrated by these four paradigmatic examples. The first one, no longer active under the form studied, comes from Italy: the platform Rousseau analysed here has been replaced by a new one called Movimento5stelle.eu in 2021, but remains historically interesting for its influence on governmental, legislative, and intraparty decisions during 2018-2020. The second one, platform Decidim, is one of the most spread open-source tools of this kind around the world: still active in Barcelona, it implements participatory budgeting among other processes. The third case study, Consulta Pública is a tool for law crowdsourcing and is provided to the Argentine parliament by a civic tech organisation called Democracia en Red using its digital toolbox DemocraciaOS. The fourth case, vTaiwan, is the name of a participatory process which leans on different digital tools in order to bring consensual proposals from the active Taiwanese citizens to the national government.

Rousseau ⁴

Founded by comedian Beppe Grillo and IT entrepreneur Gianroberto Casaleggio in 2009, the Five Star Movement (FSM) was in power in Italy in a national unity government supporting Prime Minister Mario Draghi until July 2022; it previously governed with the League (2018-2019) and the Democratic Party (2019-2021). Since its founding, the FSM has aimed to disintermediate politics through platforms such as LiquidFeedback, Meetup and, since 2013, its Operational System, which became Rousseau in 2016. Following deep disagreements between the Movement’s elected officials and the Rousseau Foundation, which managed the platform, in April 2021 the party led by lawyer Giuseppe Conte severed ties with Rousseau’s providers and migrated user data and features to a new site (https://www.movimento5stelle.eu/), linking to different companies providing different services (e-voting, data storage, etc.). I focus here on Rousseau, as this is the platform that was in place when the FSM was in power.

Political axis. The FSM is an unprecedented political force that defines itself as post-ideological, embracing at once ecological, Keynesian, anti-establishment, anti-immigration (later abandoned after the 2019 party’s official rally to the centre-left), Eurosceptic, and direct democracy claims (Bordignon & Ceccarini, 2015; Franzosi et al., 2015; Mosca & Tronconi, 2019). However, several scholars have highlighted its centralising tendencies (De Rosa, 2013; Sæbø et al., 2015) and have characterised its conception of participation as “direct parliamentarism” (Deseriis, 2017). According to Deseriis, who introduced this oxymoron, this mode of organisation “allows party members to entertain an ostensibly direct relationship with the party in public office, at the expense, however, of deliberative processes that may allow them to influence the party agenda” (Deseriis, 2017, p. 47). While it is true that decisions taken by an absolute majority of online users on the Rousseau platform were supposed to be binding for the party’s elected officials – but have not always been respected (Gerbaudo, 2019b) – votes on programmatic points and other aspects of the FSM’s internal life were implemented in a top-down manner, based on “questions that are predefined by the platform’s organizers” (Mosca, 2015, p. 42).

Software axis. In 2019, Rousseau underwent a revamp, expanding its features, which included: a) the ability to propose a bill at the local, national, or European level; b) the possibility to discuss legislative proposals submitted by elected representatives or other users; c) tools for coordinating actions on the ground; and d) the ability to vote on various topics. One key feature, “Lex Eletti”, allowed citizens to discuss legislative proposals in order to integrate ideas, modifications, or corrections. Here, the “direct parliamentarism” described by Deseriis manifested in the commitment of elected representatives in the Assembly to vote in accordance with the will of their “base”. However, users were unable to communicate with each other, nor could they interact with the elected officials or determine the priority of issues to be debated. More critically, votes on intra-party and other political decisions often took on a plebiscitary nature, heavily influenced by the charismatic co-founder of the FSM, comedian Beppe Grillo, who failed to uphold electoral neutrality by expressing himself through his blog or YouTube videos (Gerbaudo, 2019a). Before being replaced by the new platform, Rousseau had more than 180,000 registered users, seven legislative proposals under discussion at the national level, and over 400 closed ones. Co-decision is thus constrained by these limitations, plus the lack of real-world deliberative meetings and the relatively low level of participation. Indeed, while high-profile issues – such as joining a government coalition – could attract tens of thousands of votes, contributions to the co-creation of legislation rarely exceeded a few dozen comments.

Infrastructural axis. The code of Rousseau was never open, unlike that of its “app” version for smartphones. The software and hardware infrastructures used by this platform included a mix of open-source programs and languages (owned by private entities)⁵, such as Keycloak⁶and GraphQL⁷, as well as paid services from companies and foundations like Amazon Web Services⁸or the Cloud Native Computing Foundation, which provides Kubernetes⁹. Before the platform’s promoters switched to AWS for server management and scalability, users frequently experienced system crashes on voting days due to excessive traffic. This not only hindered the quantity of participation but also undermined the credibility of the voting results¹⁰. Yet, credibility was jeopardised also because the Casaleggio Associati, the private firm providing the platform, maintained exclusive control over the database of registered voters and voting records. Without an auditable system in place, Rousseau provided no assurances regarding vote integrity or protection against potential cyberattacks – vulnerabilities that were ultimately exploited by both white and black hats (Loucaides, 2020).

Decidim ¹¹

Decidim (which means “we decide” in Catalan) is an online platform implemented by Ada Colau’s administration in Barcelona in 2015, supported by a coalition of five left-wing parties and movements. The platform emerged as the technical convergence of at least two major social movements: the 15M (Indignados), which advocated for participatory democracy and produced civic hack tools, and the PAH (Platform for People Affected by Mortgages), from which Ada Colau rose to prominence, both shaping the platform’s emphasis on citizen engagement and collective decision-making. The community using the platform is constantly growing: more than 450 instances in 30 countries, 60% of which are governmental and 40% grass root¹². Their number continues to rise, given the software’s open-source nature, its flexible features, its technical stability, and its prominence in the international activist and media spheres.

Political axis. The coalition supporting Ada Colau – called Barcelona en Comú – draws inspiration, much like the researchers and activists behind the Decidim platform, from ecofeminism and Murray Bookchin’s libertarian municipalism, among other sources. These actors are also strongly committed to social justice, direct and participatory democracy, and welcoming migrants (Bookchin et al., 2019; Hamou, 2026). The political theory of municipalism fosters self-government within local communities. Here, participation is thus understood in a stronger sense, as ongoing deliberation leading to co-decision-making with citizens. Decidim’s participatory processes typically involve a few dozen participants, both online and in physical assemblies. Thus, overall participation remains limited, as it primarily attracts the most politically engaged citizens in a city with a strong tradition of political activism. The key difference is generational: online participants tend to be younger, while offline participation is more common among retirees involved in neighbourhood associations and activists from social movements such as LGBTQ+, migrant, and environmental. Despite these limitations, Decidim Barcelona stands out globally as one of the cases where participation reaches comparatively the highest levels. Active citizens can influence urban planning, municipal regulations, and the introduction of new practices within public services.

Software axis. From a design perspective, this platform allows citizens to conduct surveys, contact elected officials, organise offline meetings, participate in the city’s participatory budgeting, visualise data, track municipal projects, and propose new “participatory processes”. Decidim gives access to technical documents related to ongoing processes and allows modifications to projects, debate these proposals in an interactive way with other users and the administration, and categorise arguments as favourable, unfavourable, or neutral. As some of the cofounders have shown in an influential paper, explicit possibility to express disagreement increases reciprocity among internet users and thus the quality of deliberation (Aragón et al., 2017). Unlike the FSM, the Barcelona administration systematically provides a reasoned explanation for integrating, evaluating, or rejecting citizen proposals (Li Vigni, 2022). According to an interviewee, ensuring the traceability of citizen proposals in Decidim was a way for the administration to compensate for the fact that Spanish law prohibits making participatory mechanisms legally binding. Once proposals collected through the platform are deemed compatible with the legal framework and the administration’s political roadmap, they are evaluated by municipal councillors. If accepted, the city justifies their alignment with the broader public policy agenda and specifies where they are incorporated (norm, article, paragraph), or it indicates the law or plan where they were already included. At the time of writing, Decidim Barcelona has over 140,000 registered users, with 15 active processes and 109 past ones. The administration proudly announces that more than 70% of citizen-proposed projects and initiatives have been approved by the city council, although this figure also includes proposals that were already part of Ada Colau’s political programme.

Infrastructural axis. Even though the Barcelona city administration has a technical team to manage its platform, a number of public institutions and developer cooperatives collaborate on GitHub¹³to contribute to the platform’s code, aiming to establish it as a digital common. These teams coordinate through the MetaDecidim instance, around which an association has been created. However, while the software layer constitutes a digital common, the hardware layer relies on a private solution. Decidim’s infrastructure is built on a combination of hosting providers and services, reflecting both practical constraints and strategic choices. The platform’s main instance, MetaDecidim, is hosted on Heroku¹⁴, as determined by the public tender process managed by the Barcelona City Council, which oversees its maintenance. Meanwhile, the association’s other services to spread Decidim internationally run on Hetzner¹⁵, while additional services rely on other providers, often taking advantage of free-tier offers available to non-profits and open-source projects. For instance, Google services (via TechSoup)¹⁶are used for email, while Netlify¹⁷hosts the Decidim website and documentation, and GitHub manages the code repository. This set of solutions is largely dictated by resource limitations: the association currently has only one system administrator, making the adoption of fully self-managed servers impractical; according to an interviewee, the aim of Decidim’s team “is not to be perfect, but to provide participation”.

vTaiwan ¹⁸

The wave of social movements under the Ma Ying-jeou administration (2008–2016), culminating in the 2014 Sunflower Movement, revealed a pressing demand for greater citizen participation in Taiwan’s policymaking. This momentum led to the development of new participatory mechanisms, including vTaiwan. Launched in 2014 as a collaboration between Taiwan’s civic hacktivist community “g0v” (gov zero) and administrative institutions, vTaiwan – where “v” stands for “vision”, “voice”, “vote”, and “virtual” – was conceived as a deliberative process to engage citizens and stakeholders in digital policymaking. Due to its use of sorting and consensus-building algorithms, vTaiwan has inspired normative debates on the potential role of “artificial intelligence” in scaling deliberation (Landemore, 2024). However, its influence on public policy has diminished over the years due to the absence of a formal governmental coordinator. The administration’s attention has shifted instead to Join, an e-petitioning platform with strong institutional backing but limited deliberative depth (Tseng, 2022).

Political axis. Its promoters describe vTaiwan as an open deliberative process aimed at strengthening digital participatory democracy in Taiwan. Initially backed by Jacqueline Tsai, a Minister Without Portfolio, the platform later gained the support of Audrey Tang, who coordinated it from 2016 until 2022, before taking charge of the Ministry of Digital Affairs (MoDA) (2022–2024). However, following this, no successor was appointed to oversee vTaiwan’s governmental integration, leading to a decline in its institutional ties. Today, the process operates independently, run entirely by volunteers, primarily from the g0v civic tech community. In response to its weakened governmental connection, the platform has shifted its focus towards collaboration with NGOs and civic organisations to maintain its influence on policymaking. Ideologically, Audrey Tang embodies cyberlibertarian ideals, advocating for decentralised technological solutions to minimise state bureaucracy (Frenkiel, 2017). Their progressive and civic-tech-oriented political stance aligns with the Democratic Progressive Party (DPP), which held both the presidency and government during Tang’s tenure. The DPP is generally regarded as centre-left, pro-democracy, pro-Taiwan sovereignty, and socially liberal. Within this political landscape, some scholars characterise vTaiwan’s positioning as straddling the line between technocracy and democracy (Nabben, 2020).

Software axis. vTaiwan is a methodology combining multiple digital tools for collaborative decision-making, involving: a) an online phase through a discussion tool that helps map diverse opinions and identify consensus points; b) a hybrid deliberative phase, where stakeholders and citizens engage in discussions to refine proposals; c) a reporting phase, where synthesised conclusions are published and shared with decision-makers (Hsiao et al., 2018). Open source Pol.is platform has long been the core tool of the entire process, for it gathers user opinions through short comments expressing only one idea in response to a prompt set by organisers (Small et al., 2021). Moderators can seed discussions with sample comments and may edit or remove those that are unclear, repetitive, or express multiple ideas. Importantly, users are asked not to respond to other users in order to help the algorithm be more efficacious. Users then vote on a semi-random selection of past comments with agree, disagree, or pass options. This collaborative model mixes up top-down and bottom-up processes, while seeking to identify opposing groups. Some scholars have criticised Pol.is for its “algorithmically-forced separation of participants into in-or-out groups, which may pre-empt more heterogeneous groupings” (Moats & Tseng, 2023). However, vTaiwan organises face-to-face meetings in addition to its digital deliberations approximately every two to three months, depending on volunteer capacity. In terms of political effects, although the platform was initially envisioned as applicable to all areas of law, its implementation has remained confined to digital policy discussions. While not legally binding, the platform has facilitated discussions on 28 cases, with official figures stating that 80% resulted in some form of decisive government action¹⁹.

Infrastructural axis. vTaiwan operates on a hybrid public digital infrastructure maintained by both government and civic tech communities. The vtaiwan.tw server and domain are managed by the g0v community, while some consultations using Pol.is are hosted on Taiwan’s Public Digital Innovation Space (PDIS) and Computational Democracy’s Pol.is servers. Unlike Rousseau, which suffered from system crashes, vTaiwan has not encountered major scalability issues, as participatory processes unfold gradually rather than through sudden peaks in participation. However, there exist other challenges like data accessibility. While Pol.is recently introduced raw CSV download capabilities, enhancing platform independence and open data practices, access to datasets for in-depth analysis can still be constrained when government infrastructure is involved. Indeed, volunteers sometimes struggle to retrieve datasets for deeper analysis of the online contributions from the government staff. Although vTaiwan is now marginalised, its methodologies have influenced institutional efforts like join.gov.tw. The half dozen vTaiwan’s activists continue to meet weekly to discuss ways to improve the overall process. Recent developments include experiments with AI-assisted deliberation, in which ChatGPT is used to analyse large datasets from Pol.is discussions.

Consulta Pública ²⁰

Consulta Pública is an implementation of one of the tools provided by the civic tech organisation Democracia en Red. Originally developed in Argentina in 2012 by the DemocracyOS foundation based in Palo Alto (California), the homonymous platform DemocracyOS was designed as an open-source deliberation and voting tool to be used by the Partido de la Red (Net Party). Similarly to the FSM, this digital party aimed to have elected representatives vote according to the decisions made by citizens online (Mancini, 2015). While the party dissolved within a few years, the development of the DemocracyOS platform has been continued under the name of DemocraciaOS by a civic tech organisation called Democracia en Red founded in 2014. This shifted its focus from a single tool to a modular set of participatory democracy solutions for political parties, governments, and civil society organisations. Democracia en Red now operates exclusively in Latin America, after having abandoned a global outreach strategy for resource constraints and the need to adapt to contextual specificities. The organisation funds its operations through consulting contracts, offering customised implementations and process facilitation services to its contractors.

Political axis. Originally, DemocracyOS was deeply rooted in the ideals of direct democracy, with its founders envisioning a platform that would enable real-time citizen decision-making in legislative processes. Drawing inspiration from models like the Pirate Party’s Liquid Democracy system (Bertone et al., 2015), the Argentine platform was designed to allow citizens to propose, debate, and vote on legislation, positioning elected representatives as mere conduits for these collective decisions. This approach is closely aligned with FSM’s “direct parliamentarism”. However, over time, Democracia en Red moved away from this model, acknowledging the limitations of techno-solutionism. Instead, the organisation redefined DemocraciaOS as a participatory democracy toolkit with four tools – participatory budgeting, goal tracking, collaborative lawmaking, and digital consultations. A prime example of the last category is Consulta Pública, which presents itself as a “channel for dialogue and debate that enables interaction between the [Argentinian] government and the community”²¹. However, a closer examination suggests that government moderators selectively engage with feedback. In controversial projects, such as Total’s proposed offshore oil platform in Argentina, the platform’s moderators welcome supportive comments while merely acknowledging opposition without meaningful engagement²². Deputies provide information to those asking neutral or technical questions but systematically deflect arguments calling for the project’s rejection, reiterating that the consultation process is legally required for all environmentally impactful projects and is not the forum for debating the project’s approval or cancellation.

Software axis. DemocraciaOS digital consultations are online forums where citizens can discuss and express preferences on policy proposals using various voting methods (approval, ranking, survey formats, etc.). Structurally, Consulta Pública’s design permits exchanges between users and the administration through a comment system. However, responses remain collapsed unless actively expanded by clicking on a small arrow. This design choice appears to limit reciprocity between citizens, as discussions tend to occur primarily between individuals and the administration rather than fostering peer-to-peer deliberation. Furthermore, discussion topics are exclusively determined by the government, without any bottom-up agenda-setting. As a result, deliberation is relatively constrained, positioning Consulta Pública between “informative” and “consultive”, that is far from “citizen control” or “co-decision”. Initiated in 2016, this platform engaged just 500 participants, generating 1,000 comments by 2021; as of today, it has 2,604 participants and 5,557 comments.²³Given Argentina’s 47 million inhabitants, these numbers remain extremely modest. It has to be noted that this limited reach depends on the institutional actors deploying the tool more than on Democracia en Red itself. Without publicity, only the most politically active citizens engage.

Infrastructural axis. Unlike Decidim, DemocraciaOS lacks an international developer community dedicated to maintaining its codebase. Initially hosted on Microsoft Azure²⁴through an NGO grant, the platform later migrated to DigitalOcean²⁵to reduce costs. Democracia en Red encourages governments to self-host the tools they use, fostering institutional commitment and ensuring long-term sustainability – Consulta Pública is a prime example of this approach. However, for NGOs and smaller organisations with limited technical capacity, Democracia en Red provides both hosting and ongoing technical support. Scalability has been a recurring challenge across the organisation’s various projects, particularly during high-traffic events such as participatory budgeting processes, where servers occasionally struggled to handle peak demand. Although DemocraciaOS does not incorporate dedicated scalability solutions, the team has mitigated resource constraints by deactivating staging instances to optimise performance. That said, Consulta Pública never posed any such problem, operating on a modest infrastructure, as participation levels remain easily manageable.

Discussion: mapping participatory performance across three axes

The comparative cube makes it possible to bring together in a single analytical space the heterogeneous traits of the four platforms and to see more clearly what kind of participation they actually sustain (Fig. 2). Plotted along the three axes of political governance, software affordances, and infrastructural dependencies, Rousseau, Decidim, vTaiwan and Consulta Pública cluster in a way that reveals a broader pattern where only the Barcelonian platform tends to overcome mere consultation and to engage citizens on multiple levels.

On the political governance axis, the cube helps distinguish between platforms that merely reinforce consultative logics and those that create recognisable spaces of co-decision. Rousseau and vTaiwan both present themselves as devices of direct or radically participatory democracy, and Consulta Pública advertises itself as a channel for dialogue between citizens and institutions. Yet, once positioned in the cube, their participatory promises appear more modest: agenda-setting is tightly controlled by political elites, legal constraints, or administrative staff; binding effects are limited, episodic, or easily circumvented; and citizen influence remains largely advisory. Only Decidim systematically channels online participation into concrete decisions on urban planning, participatory budgets, or municipal regulations, while documenting how proposals are integrated, rejected, or already in progress. By situating these platforms along the information–consultation–co-decision scale rather than relying on their public narratives, the cube shows that, despite their rhetoric of participation, three of the four cases function primarily as sophisticated consultation tools, whereas only one approximates a more consequential co-decisional regime.

On the software axis, the cube makes visible how different “models of participation” are embedded in the platforms’ affordances (Badouard, 2014). Rousseau, placed close to the unilateral pole, allows users to comment and vote, but offers no horizontal interaction between participants and concentrates attention on plebiscitary votes framed by a central leadership. Consulta Pública sits nearer the interactive centre: its interface allows citizens to comment and react, but default collapsing of threads and the government’s control over topics limit peer-to-peer deliberation. vTaiwan lies somewhere between interactive and unilateral: the combination of Pol.is with face-to-face meetings enables a sophisticated process of opinion clustering and consensus-building, yet online interaction is structured as individual reactions to statements rather than as dialogic exchanges. Decidim, by contrast, is placed towards the deliberative pole: it affords threaded debates, explicit marking of supporting and opposing arguments, integration of online and offline assemblies, and systematic feedback from the administration. Bringing these software properties onto a single axis underscores that differences in “degree of participation” are not only institutional but also encoded in functional architecture.

The infrastructural axis, finally, highlights a dimension that is usually kept in the background in studies of online participation, but proves crucial in terms of platforms efficiency and credibility. Rousseau relies on a strongly privatised model: proprietary code for the main platform, opaque database control by a private firm, and cloud hosting on large commercial providers. Decidim combines an open source, community-driven software layer with a hardware layer that still depends on private cloud services contracted via public tenders, reflecting resource constraints and the absence of robust public or common infrastructures. Consulta Pública and vTaiwan are closer to the public pole: they are hosted on government-owned or publicly managed servers, but with limited capacity, uneven maintenance, and persistent dependencies on external services. All in all, choices about cloud solutions and maintenance models embody trade-offs between scalability, cost, sovereignty, and democratic accountability that condition what participation can be in practice (Li Vigni, In press).

Seen through the cube, the limits of online political participation appear more clearly. The three platforms that remain closer to consultation (Rousseau, vTaiwan, Consulta Pública) are those that combine intermediate political commitments (between informative and co-decisional), weak or constrained deliberative affordances (unilateral or only mildly interactive), and strong infrastructural dependencies (on private or minimally resourced public servers). Decidim’s relative exceptionality stems from its distinct position on all three axes: a stronger co-decisional ambition, richer deliberative tools, and at least a partial commitment to commons-oriented infrastructure. The cube thus helps at the same time to visualise differences and to explain why platforms that speak the language of co-decision so often end up operating as enhanced consultation devices, and where precisely their participatory performance is curtailed.

Conclusion

This article has pursued a programmatic objective: to propose a three-dimensional framework for studying political participatory platforms that integrates governance, software, and infrastructures into a single comparative device. Instead of assigning platforms to a few large normative families, the cube invites a more fine-grained analysis of how participation is organised, encoded, and materially supported. Applied to four emblematic cases, it shows that co-decision remains rare, that deliberation is fragile and often marginal, and that commons-based infrastructural choices are yet absent from the online participation landscape.

Several limitations, however, need to be underlined. First, the selection strategy aimed at maximising variation rather than representativeness; as a result, the typology remains exploratory and does not claim statistical generalisation across the numerous online participation initiatives that exist worldwide. Second, the cube is built from ideal-typical categories such as “co-decisional” or “commons-oriented” that simplify complex configurations and may obscure internal heterogeneity or temporal change within each platform. Third, the analysis relies primarily on the perspectives of designers, public officials, and civic-tech professionals, with less systematic inclusion of citizen experiences and no quantitative measurement of demographic biases or long-term policy impacts. Finally, infrastructural dependencies are reconstructed through interviews and documentary traces rather than through direct technical audits of server configurations.

Precisely because of these limitations, the framework should be read as a heuristic and an invitation to further work rather than as a definitive evaluative grid. Its main contribution is to suggest that any assessment of how participatory a platform is must combine at least three lenses: who governs participation and with what degree of commitment to sharing power; how interfaces and algorithms structure interaction, deliberation, and feedback; and how infrastructures distribute control behind the scenes. The cube makes it possible to re-use these lenses for other analytical questions. Beyond degrees of participation, researchers could, for instance, adapt the axes to study how platforms accommodate contestation or how they include or exclude marginalised groups. A next step would be to translate the cube into a more elaborate evaluative framework, including numerical indicators. Combining benchmarks within the three axes would allow scholars and practitioners to move from a qualitative positioning of platforms in the cube to semi-quantitative profiles, or even to composite indices tailored to specific scientific or policy questions. The cube is thus intended as an open, adaptable device that can be enriched by other researchers, public administrations, and activist communities working on online participation. In that sense, it is less a finished model than a proposal for a shared language to describe and compare platforms across different sociopolitical contexts.

Acknowledgements

The author thanks Céline Vaslin, research engineer at the Centre Internet et Société, for her work on the ideal-typical cube, the reviewers and editors of the Internet Policy Review for their valuable suggestions, as well as the many colleagues and friends whose critiques and insights have helped refine the interdisciplinary theoretical approach presented here over the last years.