Keywords

Introduction

This volume investigates the effect of different city and system characteristics on the implementation of digital citizen participation tools in cities. The way in which characteristics such as pre-existing modes of citizen participation, bureaucratic structures and gentrification processes shape the enactment of these technologies is therefore examined in this book, and how social media or online participation platforms effect the democratic dimension of city governance is discussed. We, in this concluding chapter, elaborate on this discussion using a democratic innovation framework developed in the literature, and ask the question ‘How do different approaches to digitalization of citizen participation influence how well cities perform on the dimensions of inclusiveness, deliberation and popular control in their urban governance?’.

There are good reasons for addressing this question. There is plenty of research on how the digital revolution affects democracy in general, but there is little knowledge on whether and how different uses of digital technologies effect democratic urban governance quality (Medaglia & Zheng, 2017; Smith, 2019, p. 578). This chapter shows that such differences can be quite significant. Furthermore, early technology-optimists had great hopes for digital technologies and the contribution they could make to democracy, particularly in cities, the digital technology vision being that it would allow cities to connect with their citizens in more intimate ways, involve them in problem-solving and in the co-creation of services, and even bring them into their decision-making processes (Effing et al., 2011; Shirky, 2008; Townsend, 2013). Investigating to what extent these expectations have come true is a worthwhile endeavour.

This chapter only deals with a specific aspect of the digitalization of urban governance, namely citizen opportunities to participate in urban policymaking between elections. This aspect has, furthermore, been more specifically narrowed in this chapter onto the channels that cities establish to engage with government, often called ‘invited spaces’ in the literature on citizen participation (Cornwall, 2004). ‘Invited spaces’ should be differentiated from ‘invented spaces’, used for the participatory institutions created outside the state (Miraftab & Wills, 2005). Another term used for digitalized invited spaces is ‘e-participation’ (Macintosh, 2004), which we use throughout the chapter. It can be argued that this term does not encompass the full breadth of the impact of the digital revolution. Citizen participation has, however, become a ubiquitous feature of nearly all forms of urban governance across the globe (Baiocchi & Ganuza, 2017). Between-election participation also lies at the core of nearly all contemporary theories of democracy. These therefore underline the importance of this study.

To be clear, our normative stance is that we do not believe democratic urban governance hinges solely on direct citizen participation. We believe that it requires multiple forms of representation, and that citizens play a greater role than just electing representatives. They should actively participate in collective decision-making. The underlying institutional logic or dynamic of representative institutions is, furthermore, to remove power from ordinary citizens and concentrate it at the apex of government. Representative institutions not controlled and mandated by participatory institutions (in which citizens can directly discuss and decide policies) therefore inevitably lead to the democratic aspects of urban governance being weakened or undermined over time.

The chapter is structured as follows. First, we present our approach to democracy and elaborate on how digitalization can affect the three dimensions of inclusiveness, deliberation and popular control. Then we present the three cases in our study, Oslo, Melbourne and Madrid, and explain how they have implemented digital technologies differently. Finally, we compare these three cases and discuss how these different models of e-participation affect democratic governance, and the implications this has for ongoing discussions about the impact of digitalization on democracy.

E-Participation and Democracy

Assessments of ‘invited spaces’ for citizen participation frequently centre on the three dimensions of inclusiveness, deliberation and popular control—other words sometimes being used (Fung, 2006, 2015; Newig et al., 2018; Smith, 2009; Warren, 2017). The first dimension, inclusiveness, concerns who is given a voice in politics through invited spaces, whether all citizens are eligible to participate, and whether such spaces can mobilize and accommodate the preferences and opinions of previously disenfranchised groups of citizens. The second dimension, deliberation, concerns how participants discuss and decide among themselves in these spaces, whether they are able to form their own will or judgement, or whether they merely are invited to express their preferences on issues predefined by government authorities. The third dimension, popular control, concerns the extent to which participants in these spaces are allowed to influence decisions taken by government, and the importance of these decisions to citizens’ lives.

The digitalization of invited spaces can affect the dimension of inclusiveness in a number of ways. E-participation technologies can mobilize citizens who previously were disengaged between elections, for example youth or parents with small children (Tai et al., 2020), and the use of these technologies by governments to reach large numbers of citizens is, furthermore, less expensive and requires less time than conventional channels such as mass media or public meetings. E-participation is also less ‘expensive’ for citizens in the sense that it requires less time than attending physical town hall meetings or workshops (Gilman & Peixoto, 2019). It, furthermore and perhaps more importantly, requires less network resources, competence, and self-confidence. There are few material barriers to digital participation, at least for citizens in developed countries where mobile phones and internet connection are widely distributed (Jho & Song, 2015).

Most research, however, shows that digital participation processes are populated by their own ‘usual suspects’, mostly white, middle-class men, but often a lot younger in digital than in analogue processes (Touchton et al., 2019). This, in developed countries, is not so much a reflection of unequal access to computers or the internet, but the unequal distribution of the digital skills that make people interested or comfortable with engaging in online processes (Ebbers et al., 2016). A paradoxical effect of e-participation technologies is that they can mobilize more citizens, but at the same time can reinforce existing political inequalities (Tai et al., 2020). A possible solution to this problem could be combining digital and analogue participation channels, which sometimes illustrates the benefit of mobilizing both disadvantaged and disengaged groups in participatory processes (Legard & Goldfrank, 2021). This approach may, however, lead to the risk of the creation of layers of super-participants, a set of participants with the knowledge and resources needed to juggle both sets of channels (Spada & Allegretti, 2020). This is addressed in Chapter 2 of this book.

Urban governments conduct innovative experiments with mini-publics or citizen juries, to promote deliberation among citizens (Beauvais & Warren, 2019), town hall meetings and participatory workshops also often being adopted to promote this (Agger, 2021). Only a fraction of a city’s population is, however, normally involved in these initiatives. The belief that this is inevitable is widespread in democratic theory, the assumption being that it is impossible to involve a very large number of people in in-depth deliberations (Cohen & Fung, 2004). The opposite view is that digital technology has brought the previous time and space constraints on communication to an end (Barber, 1984, p. 246) and that the internet provides, for the first time, a virtual space enabling mass deliberation (Dahlberg, 2001). High-quality deliberation through digital technologies has, however, so far been difficult to achieve (Gilman & Peixoto, 2019, p. 111; Landemore, 2020, p. 65). This is despite much work being invested in developing digital deliberation tools as an alternative to social media (Bravo et al., 2019; Shin & Rask, 2021). E-participation tools can, despite these limitations, contribute to collective will formation in other ways than moderating deliberations. Landemore (2020, p. 182) for example argues that digital technologies are particularly valuable in the crowdsourcing phase of deliberations. It allows the input from thousands or even millions of individuals to be ordered and analysed using digital technologies, and in ways that are comprehensible for those involved in public debate. Discussions on online platforms can play a role in connecting deliberations among citizens to political institutions, despite not satisfying all the criteria set by deliberative democracy theory, so increasing the number of voices and perspectives being heard in the political process (Gastil, 2021).

This brings us to our third topic of whether digital ‘invited spaces’ are equipped to ensure the dimension of popular control in the decision-making processes. Two aspects of e-participation technologies can make these more impactful than analogue methods. The first is that they can bypass traditional gatekeepers in politics (van Dijk & Hacker, 2018), for example the civil servants who normally compile, filter and then present the results of participation processes to policymakers. The other is that digital tools can mobilize masses of residents. This makes it more difficult for policymakers to disregard the outcomes of online participatory processes (particularly when compared with the outcomes of smaller public meetings or face-to-face workshops). Citizen participation is often limited to providing input to urban governments on citizen preferences—the opportunity to influence urban development being restricted by the triviality of what Fung (2015, p. 521) calls ‘the park bench problem’. Governments, furthermore, tend to ‘cherry-pick’ citizen proposals, implementing those that are cheap or do not challenge existing policy (Font et al., 2018).

Data and Methods

We, in the next section, identify three distinct models of e-participation in urban governance. This is based on the depth and breadth of Oslo’s, Melbourne’s and Madrid’s use of digital technologies, the meaning they attribute to digital engagement, and how these digital tools are connected to the policy process. Our method is based on what Skocpol and Somers (1980) call the contrast of contexts approach, which focuses on showing how putatively general propositions are invalidated or affected by particular features of different cases. In our study, this means that the impact of e-participation on urban democratic governance depends on how these technologies are enacted.

Our research questions, however, touch upon so many case aspects, that answering would be impossible without multiple data sources. We therefore apply a multi-method research approach to allow the cases to be analysed holistically, and combine both qualitative and quantitative data (Hunter & Brewer, 2015). Our main data sources are field work in the cities, interviews with key stakeholders connected to the digital spaces, and a survey of local activists in the cities. The field work is most comprehensive and the number of interviews is highest in Oslo, where both authors of this chapter live. We interviewed 48 politicians, bureaucrats, activists and other stakeholders in Oslo, attended a number of meetings, used the digital tools ourselves and followed public discussions on this topic. In Melbourne we interviewed 11 stakeholders and conducted a field visit in April 2019. In Madrid we interviewed 18 and carried out a field visit in May 2019. The survey of local activists in all three cities is described in detail in Chapter 2.Footnote 1

Answering our research question would, however, have been impossible without a heavy reliance on secondary sources. We have most notably used the ‘Survey of living conditions and satisfaction with municipal services’, a biannual survey of a representative sample of Madrid’s population.Footnote 2 We also use sources such as external and internal evaluations of the e-participation processes, official documents and legal regulations in our analysis in all three cases.

Three Models of Urban E-Participation

Oslo: E-Bricolage

The first model we identify is the bricolage model, and is represented in our study by Oslo. This is probably the most common urban e-participation model in developed countries. The model is, as the term ‘bricolage’ suggests, first and foremost characterized by a cautious and eclectic use of e-participation tools, and a focus on the digitalization of services rather than the creation of e-participation channels (Muñoz & Bolívar, 2019; United Nations, 2020). Oslo has a very ambitious digitalization policy which states that ‘all citizen services that can be digitalized, should be digitalized’ (Byrådet i Oslo, 2015, p. 14). But digitalization only includes citizen participation to a limited degree, the implementation of the digitization of citizen participation therefore being incoherent, piecemeal and often ad hoc. The lack of an overarching strategy results in the use of an eclectic variety of tools and technologies which contain elements of social monitoring of municipal service delivery, dialogue between authorities and the public on social media, crowdsourcing technologies such as participatory mapping, typical e-democracy initiatives such as e-petitions, and even more novel practices such as online participatory budgeting. Some of these are institutionalized, such as the right to petition the city council or to be consulted on urban planning issues. Others have become permanent features of the city’s digital infrastructure such as Bymelding, which is equivalent to platforms such as Fix my street in other countries. The remainder are applied when the unit in question thinks that it befits its purposes. The model is influenced by agendas as varied as ‘Open Government’, ‘Smart City’ and ‘Place-making’. They all, however, are administrative agendas that are largely disconnected from politics. Nearly all digital tools applied in Oslo are therefore primarily adopted to allow citizen involvement in different stages of the preparation or implementation of policies, but not in the deciding of political questions.

Melbourne: Digital Crowdsourcing

The crowdsourcing model is more ambitious in the digitization of citizen participation, the goal being that all participatory processes have a digital dimension. This is to ensure that citizens who are not able or willing to attend physically, can still contribute to the process. This approach is represented in this study by Melbourne. All the 31 local councils that make up the city, and the Victorian state government, provide options for online community engagement through their websites or separate participation platforms. The main justification for applying these methods is to ‘enlist the eyes and ears of citizens’ to spot public problems (Fung et al., 2013, p. 42). The governments write that they use online engagement to obtain deeper insights into how they can improve services (Victoria State Government, 2016), and find better solutions to local and city-wide challenges (Melbourne City Council, 2017). This is often referred to as crowdsourcing. The perceived benefits of this are tied to better and more sustainable problem-solving, and more resilient and healthier local communities (Maribyrnong City Council, 2017). The tools the governments use have strikingly similar functions, including map-based feedback, collecting ideas and experiences, user voting, participatory budgeting, discussion forums, polls and surveys. The participation processes combine methods of physical and online participation, as this is believed to enable a more robust and inclusive participatory process. The crowdsourcing model only has weak ties to the political level of government. It is primarily applied by administrative units to the preparation or implementation of policies. The administration plays a strong role in running the municipalities of Melbourne, citizen input from digital platforms potentially giving a greater citizen say in Melbourne than in Oslo. The prime influence behind the model is ‘New Public Governance’ (Baiocchi & Ganuza, 2017), which sees participation as beneficial to the human qualities of urban development and sees citizens as co-creators of their city environment.

Madrid: Online Direct Democracy

Madrid’s model is closest to the online direct democracy envisioned by many techno-optimists in the early stages of the internet (Chadwick, 2006). It is, however, perhaps the rarest in the world of digital participation in urban governance (Steinbach et al., 2019, p. 61). The goal of the direct democracy model is, unlike the bricolage and crowdsourcing models, ‘digital first’ in public engagement. The city government of Madrid established the Decide Madrid platform in 2015. The purpose of this platform was to encompass all major participatory processes at the city-wide level, including citizen proposals, participatory budgeting, voting on policies proposed by the government, citizen-initiated political discussions, and more open consultations on plans and proposed legislations. The platform was to create a democratic city in which ‘all citizens can intervene in the definition, administration and development of fundamental policies’, beyond just voting in elections (Ahora Madrid, 2015, p. 7). The relationship between citizens and politicians should, however, be turned upside down. Elected officials ‘should serve the citizens’ and bureaucrats should learn to ‘work together with the people’ instead of within their offices (ibid.). The government therefore saw digital technologies as being indispensable in including the population. Participating citizens are directly connected, in this model, to the political process and can both propose and decide policies. Influences in Madrid are from the social movements that first developed many of these tools, primarily the Spanish Indignados movement that arose in 2011 in reaction to the austerity measures following the global financial crisis. The Indignados were recognized as innovators of ‘civic tech’, activists in these movements becoming central actors in the government, in platform development.Footnote 3

Comparing the Models

Inclusiveness

What does a comparison of inclusiveness in these models show? We unfortunately do not have reliable data from Oslo or Melbourne on the number or type of participants. We have indicators, however, from Madrid that allow us to give good answers to this question. The ‘digital first’ of Madrid’s direct democracy model seems, in terms of absolute mobilization numbers, to have paid off. The platform in early 2019 had more than 450,000 users, and was visited over 11 million times (ParticipaLab, 2019, p. 23). This is reaffirmed by the data that as many as 20 per cent of the Madrileños in 2019 had recently participated in a consultation held by the municipality. This figure rose from 6.4 per cent in 2012, three years before the introduction of the direct democracy model, which indicates a quite stunning level of mobilization. We show this to be a strength of this model in Table 8.1, on the inclusiveness dimension. The table summarizes the strengths and weaknesses of all three models relative to the democratic dimension of citizen participation.Footnote 4 The age, gender and place of residence of participants in 2019 were representative of the general population. Participants, however, had higher incomes and education levels than the average citizen. Citizens with immigrant backgrounds were underrepresented. Does this mean, as the most pessimistic theories predict, that the online direct democracy model amplifies inequalities in political participation? No. The model seems to just copy the inequalities that existed prior to its implementation. The profile of the participants remained the same, despite Decide Madrid surely animating more citizens and bringing new groups of citizens to participate between elections. The model is therefore vulnerable to pre-existing inequalities in political participation.

Table 8.1 Strengths and weaknesses of e-participation models relative to democratic dimensions of citizen participation

Do the bricolage and the crowdsourcing models fare better in this respect?Footnote 5 One of the main advantages of these models is that they are not married to the idea of digital participation being the main channel for reaching the general population. Planners in Oslo often use digital tools to target what they call ‘weak voices’ such as children, youth, the elderly or people with migrant backgrounds. The ad hoc nature of Oslo’s model furthermore gives the flexibility to target, which is its main strength on this dimension. A particularly interesting example is a district that used paid ‘ambassadors’ to encourage hard-to-reach citizens with migrant backgrounds to participate on its online platform. This mobilized twice as many participants as analogue engagement methods (Melbøe, 2021). Our survey, however, indicates that there still is a tendency in Oslo for those that engage in digital channels to be from the same privileged layers as those who mainly engage in analogue channels. The weakness of the ‘weak voices’ targeting strategy is that it does not seem to work for the city as a whole, but only in isolated cases.

The crowdsourcing model in Melbourne is based on a similar belief that engaging different types of citizens requires different methods of participation. This means that it shares similar strengths and weaknesses. Officials at both the state and local level are aware that ‘digital will only ever get to a certain group’Footnote 6 and that a comprehensive community engagement strategy that combines the digital with face-to-face processes is required to ‘cover the variety of cultures and cohorts’.Footnote 7 Melbourne’s digital and analogue participants have, however, higher levels of education and a more non-migrant background than the average population, as in the other two study cities. This is despite extensive efforts to use a combination of digital and analogue methods of engagement to reach broad segments of the population. The model is therefore vulnerable, as pointed out in Chapter 3 to privileging online users and to reproducing existing inequalities.

Deliberation

How do the different models perform on the deliberative dimension of citizen participation? Oslo’s bricolage and Melbourne’s crowdsourcing models both avoid using digital channels for deliberative tasks. They instead assign these tasks to face-to-face forums that involve small groups of citizens. A lack of faith in the use of e-participation tools for deliberation stems, in Oslo, from negative experiences of dialog on social media and in large public meetings. Both tend to be dominated by white middle-aged and middle-class males, and to be highly biased and polarized discussions. The similar domination of e-participation processes is therefore a fear. The city’s planners therefore use small face-to-face workshops or more creative methods to facilitate deliberation among ‘weak voices’, such as in the co-designing and co-construction of public spaces. The strength of this approach is that it avoids some of the problems of online deliberations. The weakness is that the spaces in which deliberation occurs are fragmented and inaccessible to the general public, but accessible and coherent to a few selected participants.

Public administrators in Melbourne are highly influenced by Australia’s strong standing in the field of deliberative democracy. They are also firm believers in using mini-publics as opposed to online deliberation. This is emphasized in the crowdsourcing aspect of their model, crowds normally being seen as a source of experiences and opinions, but not as the deliberating entities (Howe, 2006). Digital and analogue crowdsourcing is therefore often used to provide input to citizen deliberations that make up the mini-public. Digital tools were, for example, used in the City of Melbourne to receive proposals for a citizens assembly called ‘Future Melbourne’, and which were deliberated on when formulating a 10-year local government plan (Katsonis, 2019). Similar assemblies of randomly selected citizens have also been introduced by other councils. The strength of this approach is that any citizen has the chance of being picked for the citizen assembly. This, when combined with open crowdsourcing processes, furthermore allows all citizens to provide input to these deliberations. The weakness is that such citizens assemblies are used so rarely that very few inhabitants have a chance of being involved in them. Different forms of crowdsourcing are therefore still the dominant method of engagement. These, however, are mostly in the form of a predefined survey, the government therefore sets the agenda and citizen input. The room for citizen collective will formation in these spaces is therefore severely restricted.

Madrid’s direct democracy model took the opposite route and promoted online deliberations on its platform as an alternative to discussions in social or traditional media. This is reflected in the first two features on the platform being debates and proposals through which citizens could discuss among themselves and launch, defend and gather support for their own policy initiatives. The level of activity was impressive. Users had, by early 2019, left more than 25,000 proposals that received more than three million votes of support, and 5630 debate threads that generated around 193,000 comments (ParticipaLab, 2019, p. 23). The platform did not, however, deliver as well as the government hoped. In a critical self-evaluation, Medialab Prado concluded that most platform users were only involved in ‘thin participation’, simple interactions such as voting and clicking and not extensive deliberations. Most users came to the platform only to read, vote or support a proposal. Those who posted content usually did so only once. The majority of debates and proposals were created by the thousands of superusers, who were the most frequent visitors to the platform and generated most of the content (ibid., p. 40). Another challenge was information overload. This resulted in likeminded citizens not finding out about each other’s proposals, so disabling them from gathering enough support for proposals that were very similar. The government therefore decided to establish a randomly selected citizen panel, the Observatorio de la Ciudad, to deliberate over the most popular online initiatives and to see whether they could be passed on to the city council. The reliance on insufficient technical solutions to achieve online deliberations highlights the weakness of this model. The model’s strength is, however, that it establishes a transparent space in which citizens can access relevant discussions and bottom-up initiatives not predefined and controlled by the government.

Popular Control

There are notable differences between the opportunities for citizen participation in the urban decision-making process provided by the different models. Oslo’s and Melbourne’s models primarily use digital channels to inform and consult citizens. The direct democracy model, however, centres on involving citizens directly in decision-making. In Oslo, input from both digital and analogue consultations may or may not be taken into consideration by the elected politicians, who have the final say in deciding the city’s policies. The description provided on the city’s main participation portal for planning issues, Si din mening (‘Give your opinion’), is telling. It emphasizes that opinions voiced through the portal are sent to those who propose plans (private entrepreneurs or public entities) or the planning authorities, and that it is their privilege to ‘decide whether they will take the opinions into account or not’ (Plan- og bygningsetaten, 2020).

The crowdsourcing model, of which the City of Melbourne is a good example, is similar. The rhetoric on the city’s engagement platform, written by the municipality, states that citizens can ‘join the conversation to influence the plans’ and that resident opinions and ideas ‘help shape Council’s decisions’. However, it furthermore explains that comments, ideas and suggestions are collected and used to ‘inform Council decision making processes’ (City of Melbourne, 2020, our emphasizes), officials said that it is up to the administrators to decide whether and in what way the input from online engagement is used. It is therefore ultimately the city council that makes the decisions on most projects put up for consultation. This was put well by a former chief community engagement manager, who said ‘if you are looking at a [citizen participation] spectrum it sorts of [fits] very well into the consult area’.Footnote 8 This is also true for e-participation at the state level and the other councils we studied.

The weakness of both the Oslo and Melbourne models is therefore that developers, bureaucrats and politicians can discard the input received at will. A strength they have in common, on the other hand, is that citizens are often consulted on quite significant developments and issues (of much greater significance than the ‘park benches’ they are allowed to decide directly over). For example, citizens in Melbourne could, using state-level e-participation tools, vote and decide directly on minor issues such as the content of a gift package given to new-born babies or relatively small community funds. They are, however, also consulted on much bigger issues such as redevelopments of social housing estates, hospital plans or large infrastructure projects. Citizens in Oslo have also been invited to take part in participatory budgeting to decide very small community funds. The municipality is, however, obliged to consult citizens on all plans proposed by both government and private developers, which explains why our survey still shows a positive relation in both Oslo and Melbourne between the use of digital platforms and the influence community activists perceive they have. One notable difference between Oslo and Melbourne is, however, that the open feedback solution chosen by Oslo (of using the website Si din mening instead of sending an e-mail) requires citizen expertise in understanding the cases they are to provide feedback on, and skills in formulating meaningful feedback to the planning authorities. Melbourne uses much more accessible tools such as polls and surveys, which makes it potentially possible for ordinary citizens without such resources to exert a degree of influence. We again, however, emphasize that this influence is within the predefined parameters set by the government.

The direct democracy model represented by Madrid marks a deviation from the broader pattern of e-participation being limited to information and consultation. The strength of the model was that participating citizens in Madrid were directly connected to the political process. For example, between 2015 and 2019 more than 346 million euros of investment was reserved by the city council for allocation by citizens through the digital platform. Two citizen initiatives gathered enough support, through the platform, to initiate a binding referendum, and the city government initiated binding votes on other issues, including refurbishment of parks and squares and traffic ordnances. The connection to the political process was upheld by the government, which did not have the legal authority to hold binding votes or referendums, but voluntarily committed to implementing the results of the processes. This e-decision-making seems to be endorsed by public opinion. In 2012, before the introduction of the model, 26 per cent of the general population and 47 per cent of those that had been consulted believed that the municipality facilitated citizen participation in its decision-making processes. These proportions had, however, increased in 2019 to 57 and 81 per cent, respectively. A weakness of the model is, despite this, the government only allows citizens to decide on minor projects, compared to the size of the budget and the major developments taking place in the city. The investment projects were small, and the decisions made by platform users related to renovation of parks and squares, and the public transport ticket system. This is admittedly more due to the city’s limited authority over urban development processes and public services. It does, however, illustrate that it is perhaps easier to give citizens decision-making authority in minor projects than for large budgets or strategically important urban planning.

Concluding Discussion

A number of conclusions can be drawn from this analysis. The first is that it is not just democracy at large, but also specific democratic practices at specific places that is digitalized. A second conclusion is that the impact of digitalization depends on the way in which e-participation technologies are enacted. Each of these enactments has their strengths and weaknesses, as the differences between Oslo’s, Melbourne’s and Madrid’s models show. The most notable differences in this study are between the direct democracy model and the other two. Madrid’s e-participation approach was, for example, able to mobilize a much larger number of citizens into the political process than the other two cities. This was not just due to the technology used, but factors such as the general level of civil society mobilization and the way in which citizens were allowed to affect policy outcomes. It is, however, unlikely that this level of participation could have been achieved without a digital platform.

The sad overall fact is, however, that despite this exception none of the models have been able to counteract the reality that citizens have different levels of ability and willingness to participate through ‘invited spaces’, whether digital or not (Gaxie, 2014, p. 23). Madrid’s high participation numbers seem, in fact, to verify the paradox that digital participation (even in the best case scenario) can mobilize more citizens, but simultaneously reproduces existing political inequalities. The importance of digital channels in reaching new groups should not, however, be underestimated, even if the most frequent users of these digital channels are the ‘usual suspects’. E-participation tools can also give vulnerable groups a channel for voicing their concerns, but only where their voices are not drowned out by others, and where the city government listens to every voice rather than those who shout loudest. A combination of the broad participation found in Madrid, with the specially designed tools and processes found in Oslo or Melbourne, is probably the best solution for promoting inclusiveness.

Madrid’s attempt to create a digital deliberative space on its platform also deserves attention. This space opened up the process to relatively autonomous deliberations on topics that potentially could affect policymaking directly. The failure of this initiative unfortunately the claim that not even the most promising digital platforms have so far been able to design solutions that allow for mass online deliberation. E-participation can, even so, be valuable in collective will formation through crowdsourcing. This can allow more experiences and views to be considered in public deliberations than otherwise would be the case. Melbourne’s and Madrid’s attempts to design processes in which crowdsourcing takes place first online, followed by a phase of deliberation by randomly selected citizens, represent innovative and promising solutions. This has also been recently attempted in the constitutional process of Iceland and in the citizen panel on climate change in France (Landemore, 2020). Realizing technical, political and social solutions that can allow greater numbers of citizens to take active part in this will formation process, without it being dominated by the ‘usual suspects’ still, however, seems to be a challenge. Finally, Madrid contributed more to the popular control dimension than the other two models. This was more due to the way in which it was connected to the political institutions, than to its digital dimension. The scope of many of the issues that users of Decide Madrid were allowed to decide was limited. This, however, shows how e-participation tools can be used to involve citizens in collective decision-making, even in a modern metropolis.

It should be noted that participation through digital technologies had a limited impact on power relations in all three cities. We believe the explanation of this is twofold. The first is that e-participation takes place within political, administrative and economic structures that are not easily changed by the introduction of digital technologies, the specialized and fragmented bureaucratic structure of modern city government being acknowledged by civil servants (in all three cases) to be a strong barrier to implementing the outcomes of the participatory processes (see Chapter 4 of this book). The multilevel structure also sets limits on what issues the cities invite citizens to influence. The responsibilities of local councils in Melbourne, and the municipality in Madrid, are very limited, all major urban development issues for example being decided by state or regional levels of government. This is one reason why citizens, in at least these cities, are predominantly invited to give their opinions or vote on minor or trivial issues such as parks and park benches. Another reason is that urban democratic governance takes place in settings dominated by private capital and developers. All three digital models exist within largely neoliberal regimes of urban governance, governments in this being expected to act as entrepreneurs, make the city business-friendly and provide the backdrop for large-scale investments. Planning has therefore largely been transferred to outside market-actors, the civil servants of the three cities believing that developers or business organizations are the most powerful stakeholders (Chapter 4).

The findings presented in this chapter further our knowledge on whether and in what way digital technologies affect the quality of democratic governance. Digital technologies can enable cities to reach out to more people and strengthen the citizens’ role in politics. Our study, however, uncovered limitations of digital participation, confirming previous findings that these technologies often reinforce existing inequalities, and that high-quality deliberation is difficult to achieve in digital spaces. The impact of digital participatory processes is, furthermore, ultimately dependent on the willingness of politicians and civil servants to share power, and on the scope of authority they can share. Finally, we wish to emphasize that these models were not arbitrarily chosen by city bureaucrats or politicians. They are, instead, the outcome of path-dependent processes, or processes contingent upon specific events. For example, the economic crisis in Spain created a rebellious and internet-savvy movement that later took office in Madrid. We, nevertheless, believe that our assessment of the strengths and weaknesses of these models can be useful to practitioners in other cities, as they set out to design systems of citizen participation and consider which e-participation tools and practices to implement.