1Today constitutional modernisation has taken a deliberative democracy turn; that is constitutional changes through processes of democratic deliberations that include not only constitution-makers but also – parts of – the general public [1]. This turn comes in the wake of a larger deliberative movement [2]. Democratic innovation through deliberation is a topic that has greatly regained attention in the last decades, in a time where current democratic institutions seem to be unable to secure the support from the people and to bring solutions to their problems [3]. Several scholars believe that ordinary citizen deliberation has the potential to generate legitimate political decisions, even when there are strong disagreements on the objectives and aims a policy should promote [4]. Nevertheless, for deliberative democracy to contribute to the legitimacy of modern political systems, these processes of deliberative democracy must reflect principles of legitimacy themselves [5].

2The aim of this paper is to analyse recent attempts of constitutional modernisation through deliberative democracy. Four cases – Iceland, Ireland, Luxembourg and Nepal – provide the empirical ground of our assessment that will be performed through the lens of three different types of legitimacy in the wake of works [6] inspired by Easton: input legitimacy, throughput legitimacy, and output legitimacy [7].

1 – Analytical Framework

3The framework that will be used to analyse Iceland, Ireland, Luxembourg and Nepal in their efforts of Constitutional modernisation and deliberative democracy is a framework that assesses the legitimacy of Constitutional reform in a deliberative environment via three types of legitimacy: input, throughput, and output [8].

4Input legitimacy refers to the nature of representation that deliberative democracy allows for. Essentially, input legitimacy deals with citizens’ opportunities to influence the process and the outcomes of deliberation [9]. The question of « who deliberates » is crucial in this respect. The second dimension that is paramount to input legitimacy is the agenda-setting dimension: which questions will be deliberated? Naturally, the whole dynamic of deliberative democracy is different when participants are faced with a closed agenda (i.e. participants can only debate questions that were predetermined) or open agenda (i.e. participants can determine themselves what topics will be discussed). Finally, the last dimension that input legitimacy is interested in is the question of epistemic completeness, which refers to the level of information that participants possess. It also refers to the tools made available to the participants that enable them to acquire good information on the selected topics [10].

5Throughput legitimacy focuses essentially on the deliberative process itself, the shape and form that deliberation takes: to what extent were participants able to take part? Did every participant have an equal voice and an equal amount of talking time, was every one able to bring out his or her experience and perspective, and what does the group composition look like? Moreover, throughput legitimacy also looks at the « quality » of decision-making, and « how » the deliberative process translates into a decision: the idea is to examine how the participants reach a consensus, and in case they do not, what method is chosen to arrive at a decision (e.g. voting) [11].

6Finally, output legitimacy essentially focuses on three criteria: public endorsement, weight of the results, and responsiveness and accountability. Output legitimacy aims to explain how decisions taken by a few individuals can be generalised and explained to the entirety of the population in general, to the « maxi public ». Indeed, decisions taken by a small group of individuals still have to be justified to the maxi public that did not take part in the discussions and debates. The key dimension examined in terms of output legitimacy is concretely what is done with the outcomes and results of the discussions and debates. For the outputs of deliberative democracy to be legitimate, they have to be linked in some way to formal political decision-making. The corollary of this examination is generally to assess whether there is any accountability on behalf of the political power towards the citizens. This of course depends on what the political power agreed to do based on the results of the deliberation [12].

2 – The Case of Iceland

7One of the most impressive examples of deliberative democracy is the case of citizen panels in Iceland following the aftermath of the 2008 financial crisis. The Icelanders demanded the organisation of extraordinary elections that would result in a Constituent Assembly responsible for the elaboration of a new draft Constitution. Under the pressure of a disgruntled population, the leftist government agreed to a three phased mechanism of constitutional revision. First, a « national forum » of 1,000 randomly selected citizens was created to deliberate for the duration of one day to set the principles and values upon which the new Constitution would be based. At this stage, the idea was not to promote a contradictory debate, but rather to identify the main questions and issues that Iceland would face in its Constitutional revision. Next, a « Constitutional Committee », composed of seven members appointed by the government was created to collect information and analyse the issues and problems highlighted by the « National Forum » with the aim of making suggestions of constitutional revision. This phase resulted in a 700-page report that would serve as a starting point for deliberation. Finally, the Parliament of Iceland called for the creation of a « Constituent Assembly » composed of 25 members, that would be elected through national elections, in which 522 candidates competed. The aim of this Constituent Assembly was to revise the Constitution, or alternatively draft a completely new Constitution using the work of the « National forum » and of the « Constitutional Committee » as a foundation [13].

8However, several events raised questions about the legitimacy of the process. For example, an increase in the amount of polling stations due to the increased complexity and duration of the votes led to the cancellation of the elections for the Constituent Assembly in January 2011, and this for purely technical reasons. Following these cancelled elections, the government decided to appoint individuals that could have been part of the Constituent Assembly to be part of a freshly created body called the « Constitutional Council », which would have more or less the same responsibilities and tasks [14].

9In terms of input legitimacy (i.e. who deliberates, level of information of participants, and how the agenda is set), the results for Iceland’s case are mixed: in this three-phase process (i.e. randomly selected citizens forming a national forum, appointed experts forming a Constitutional Committee, and finally the election of a Constituent Assembly), various elements must be taken into account.

10First of all, the « National Forum » formed in 2009 gave citizens the opportunity to express their opinions and thoughts. In practice, 900 citizens were randomly selected, and 300 additional representatives were selected amongst various interest groups and associations in civil society. These 1200 individuals participating in the « National Forum » were divided into small « round table discussion groups » which were managed and moderated by professionals. At the end of the process, a document summarising the results of the deliberation was published. This document highlighted the values and principles that these 1200 citizens put forward: honesty, equal rights, respect, justice, responsibility, liberty, durability, democracy [15].

11However, in practice, a second « National Forum » was created in 2010. This time, the initiative came from the government instead of civil society. At this stage, 950 citizens were randomly selected to participate, but levels of mobilisation and public outcry were no longer as strong, making this second « National Forum » less mediatised [16].

12Next, concerning the « Constitutional Committee », it was composed of experts chosen by the government that represented various political tendencies in the country. It does not come as a surprise that this body was therefore strongly politicised [17].

13Finally, the « Constituent Assembly » composed of 25 members, was elected by the nationals of Iceland, but few citizens participated in this extraordinary election, weakening the input legitimacy of this third phase of the process [18].

14In conclusion, input legitimacy for the case of Iceland was greatly damaged by the cancellation of the elections for the Constituent Assembly, and its subsequent replacement by a mechanism of appointment by the government. Moreover, the use of government nomination for yet another body, the « Constitutional Committee », instead of a more inclusive method, weakened the input legitimacy. Furthermore, the low turnout in the extraordinary election did not really strengthen input legitimacy in the Icelandic process. However, the method chosen to form the two « National Forums », i.e. random selection, reinforces input legitimacy of the Icelandic case [19].

15Concerning throughput legitimacy, which as a reminder refers to the shape that the deliberative process takes (i.e. how individuals participate, presence or not of discussion moderators, and how the deliberation translates into a decision), the Icelandic case opened the work on constitutional reform to the general public. Indeed, the Constitutional Council made calls for contributions on social networks and media such as Facebook and Twitter. Some started to talk about a « crowd-sourced » Constitution because of the vast possibilities of involvement in the process for the general population. However, this optimism of a « crowd-sourced » Constitution must be called into question. Despite the numerous resources at its disposal, the Constitutional Council was not able to thoroughly examine the entirety of the contributions because the Council only had four months to achieve all this work [20].

16In terms of methodology of the deliberative process in Iceland, an iterative method was adopted. This means that documents were completed gradually, step by step, or round per round. Concretely, three different committees worked each on their own portfolios. These committees would then meet all together in a plenary session to debate and deliberate on the various modifications and amendments that each committee thought of. Through this method, the document was progressively refined in order to finally take the form of a new draft for a Constitution. Subsequent to this method, the decision making process for the Constitutional council was a mix of deliberation and voting [21]. Despite the vigorous debates, the Constitutional Council ended up by unanimously adopting the new draft of a Constitution. This is an important element to outline because it showed a united front against the expected resistance of the political class in Iceland. Experts tend to agree that the throughput legitimacy for the Icelandic case was rather strong because of its innovative and participative method [22].

17Finally, concerning output legitimacy, which refers to the explanation and generalisation of decisions made in a small group to the wider public, results are once again mixed. On the one hand, uncertainty reigned amongst political parties and in the Icelandic Parliament because the Constitutional Council decided to work without cooperating with political parties, and without keeping them in the loop. As a consequence, both the Parliament and the political parties felt alienated from the process, and directly after the publication of the new draft of the Constitution, political quarrels began [23].

18The Icelandic Parliament opted for an advisory referendum with six questions on the ballot. In October 2012, the citizens of Iceland were primarily asked whether this new draft written by the Constitutional Council should serve as the foundation of the new Icelandic Constitution. After that, five sub-questions dealing with debated articles of the new draft were also asked during the same referendum. These five additional questions concerned public ownership of natural resources; increased use of referendums; increased personal voting; equal weights of votes; and an article on the status of the church. Approximately half the electorate showed up to the polling stations, of which around two thirds accepted the draft for a new Constitution as the basis for a new Constitution, which the Parliament would have the task to complete [24].

19The fate of this new Constitution rested in the hands of the Parliament. Running short on time, due to the incoming elections of April 2013, the government agreed to delegate the decision concerning the new Constitution to the next Parliament. However, the issue was that the question that dominated the electoral campaign and the agenda for these elections was not this draft for a new Constitution, but rather the « Icesave » conflict between Iceland and the United Kingdom. Hence, output legitimacy seemed strong at the beginning of the process, but this impression must be put into question due to the disappearance of the draft Constitution from the political agenda during these crucial elections [25].

20Furthermore, output legitimacy in the Icelandic Case is weakened by the fact that the new government that saw the light of the day after the 2013 elections decided to discreetly abandon the constitutional process to instead appoint its own Constitutional Committee, which was politicised and composed essentially of conservative lawyers. In September 2014, the Chairman of this committee resigned and declared that there was neither interest nor reason to change the current Constitution. However, something to note is that his successor declared that the committee would continue its work on the basis of the constitutional drafting process. What is more, some political parties of the opposition stated that they would try to ratify this new Constitution, should they be in government. The Icelandic process is possibly still ongoing, so additional research is necessary to make a full assessment of the Icelandic case [26].

21As a conclusion for the Icelandic case, it would be appropriate to note that despite the results in terms of legitimacy, the process served as a sort of healing exercise for the Icelandic society after the financial crisis. Indeed, the process stimulated wide public discussions and expectations for greater public participation. Moreover, this exercise attracted substantial attention from abroad. Consequently, the spirit of the Icelandic process has widely spread abroad, counting as significant output legitimacy [27].

3 – The Case of Ireland

22In December 2012, the Irish government established the Irish Constitutional Convention and joined the increasingly common trend of citizen involvement in processes of Constitutional reform. Like in the Icelandic case, it is in a context of financial crisis that a process of constitutional reform took place. The Irish Constitutional Convention is the product of both the financial crisis of 2008, of a disgruntled population, but also of a compromise between two political parties; Fine Gael and Labour [28].

23The Irish case is interesting because of the fact that during the elections of February 2011, all the Irish political parties had included in their election manifestos proposals for the establishment of citizen-oriented deliberative forums. This decision was arguably a response to the numerous public debates that questioned whether the Irish political system was fit for purpose [29].

24The design of the Irish Constitutional Convention was largely inspired by the example of citizen assemblies on electoral reform that took place in the Canadian provinces of British Columbia (2004) and Ontario (2007), and by the We the citizens (2011) pilot project which was organised by Irish political scientists with the aim of proving to political authorities that Irish citizens were able to debate over complex questions and issues [30].

25In terms of input legitimacy, the Irish Convention was composed of 100 members: 66 citizens, 33 elected legislators, and one President: Tom Arnold. The citizens were selected randomly in order to secure a reasonable reflection of the Irish population with regard to gender, age, region, level of education, and socioeconomic status. In addition, a series of extra citizens were selected as substitute members. Some disadvantages were that there were no homeless people included in the sample, due to the recruiting technique employed and the small size of the sample, and that women were underrepresented. Now for the 33 elected legislators, political parties determined their composition. Fine Gael left its share of legislators to be appointed by its party whip, whereas Labour decided to proceed via vote to determine its delegates [31].

26Concerning participation in general, numerous efforts were made in order to make the maxi public participate: calls for contributions were relayed in social medias, the press, radio, and television. Contentious questions tended to receive a substantial amount of contributions. In total, the Convention received over 2500 contributions, ranging from well-developed propositions to short messages. Various associations and interest groups played the role of amplifiers to mobilise the wider public, increasing input legitimacy [32].

27In terms of agenda-setting, which is an important element to take into account when assessing input legitimacy, the agenda was predetermined by the Irish houses of Parliament (« Houses of the Oireachtas »). This tends to weaken input legitimacy since the participants are not free to choose the themes to discuss. Moreover, the agenda seemed rather full considering the time that was available to both inform the participants, and then debate every theme, which tends to once again weaken input legitimacy. However, this statement must be put into perspective, because the participants to the Convention were innovative and did not hesitate to stretch the remit of the Convention beyond the confines set by the government. Indeed, the members of the Convention often considered options that often went beyond what either the government or chosen experts recommended. For example, instead of reducing the minimum age to vote to 17 years old, as recommended by experts, the members of the Convention agreed on the age of 16 [33].

28To inform the members of the Convention, four political scientists and one legal scholar worked together with the members of the Convention. The role of this group was to work in tandem with the Convention Secretariat to suggest the intervention of various experts that would display a broad spectrum of opinions and positions on the themes that were to be discussed. Concretely, this group of five scholars had the task of preparing and vulgarising briefing documents to make information easy to understand and accessible for the participants, and to prepare presentations for the members of the Convention on the addressed topics. However, in terms of information sharing, a balance had to be found between, on the one hand, information sessions and, on the other hand, debates in small groups. Hence, it was observed that only little time was allocated to informing participants because ten topics had to be discussed in only eight weekends, which weakens input legitimacy [34].

29When it comes to throughput legitimacy, the Irish process can be summarised as debates where each and every one was invited to participate in a constructive and respectful discussion. The 99 participants were arranged into roundtables of approximately eight individuals per table. Each table was composed of a majority of citizens, some politicians, one note-taker, and one trained facilitator. The facilitator had many tasks: making sure that the debate did not digress too far into other topics, making sure everyone had an equal opportunity to participate, and that the discussion remained respectful of everyone’s views and opinions. Moreover, the Convention tried to make the process gender-inclusive by including enough women at each table. According to Karpowitz, Mendelberg, and Shaker, women participate less when in a minority [35]. However, the issue was that there were fewer female Convention members than male members [36].

30In terms of translation of the deliberation into a decision, the Convention operated through a mixture of deliberation and vote, resulting in a good balance between, on the one hand, inclusiveness and depth of reasoning during debates, and on the other hand, equal voice during votes [37].

31Finally, concerning output legitimacy, the results can be considered quite satisfactory. The Convention sent reports to the Irish houses of Parliament and the government, which had the obligation to respond to these reports within four months of the reception of these reports. The Irish houses of Parliament and the government had the obligation to discuss the contents of the reports submitted by the Convention and had to give feedback. Furthermore, a positive feature of the Irish case resides in the fact that the Convention was partly composed of politicians, which minimised the risk of a « disconnect » between the political class and the Convention, a problem that is often observed in other cases of deliberative democracy. Actually, several Irish politicians made references to this Convention during parliamentary debates, meaning that the Convention indeed did not get disconnected from the political class. What is more, a large proportion of the participants admitted that during the experience, they changed their mind and opinion on several topics, boosting the output legitimacy of the process [38].

32The impact of the Convention should be assessed both on polity and on society. On the one hand, the Convention had an advisory impact on polity, since the Convention was mandated to make recommendations only. It seems that the government took the Convention seriously since they answered reports made by the Convention in a timely fashion. Moreover, two referenda did indeed take place in Spring 2015 on questions raised by the Convention. However, it would be wise to reconsider the Irish case after seeing the evolution of the debates in the Irish houses of Parliament to make a final assessment of the Irish case. On the other hand, the Convention had an impact on society since the Convention’s website has archived papers, submissions, presentations made by experts, which will serve as a substantial educational resource for the wider public during the next referendum campaigns [39].

33To conclude on the Irish case, it should be noted that quite ironically, it is in the areas that were initially critiqued (i.e. its composition and closed agenda) that the Convention turned out to be innovative, and increased all three types of legitimacy. On the one hand, the inclusion of politicians in the Convention was a valuable source of knowledge of technical and institutional aspects, and on the other hand the remit of the Convention was spontaneously widened by the participants in order to discuss the topics in a broader way [40].

4 – The Case of Luxembourg

34Luxembourg’s experience with constitutional reform through deliberative democracy is set in the context of a referendum campaign over the revision of the Luxembourgish Constitution. This campaign can be divided in two phases: first, the citizens of Luxembourg were invited to voice their opinion on 7 June 2015 on four advisory referenda questions that were prepared by the government. Next, at the end of 2016, the citizens of Luxembourg will have to cast their votes, this time with legally binding force, on the whole project of constitutional revision. Between these two referenda, the deputies of the Luxembourgish Parliament will have to decide what gets revised in the Constitution, and what stays the same, by taking into account the results of the advisory referendum held in June 2015. In order to make these decisions, a majority of two-thirds is necessary [41].

35The four questions at the heart of the Luxembourgish process were the following ones: right to vote at age 16, voting rights for foreigners residing in Luxembourg, a 10-year limitation on a ministerial mandate, and finally the funding of ministries of cults [42].

36In terms of input legitimacy, a citizen panel was organised by a group of scholars. 27 citizens were chosen to form a group reflecting the population of Luxembourg. These participants were chosen respecting various criteria such as age, gender, level of education, and nationality (i.e individuals that did not have Luxembourgish nationality but who resided in Luxembourg were included in the process) [43].

37Concretely, the process began with the participants obtaining informative documents and fact sheets regarding the topics that would be subject to a referendum in June 2015. This information-sharing step greatly increases the input legitimacy of the process. In terms of agenda setting, the case of Luxembourg differs slightly from the Icelandic and the Irish cases, because the citizens did not have any say in the choice of topics that would be subject to referenda. The questions that were asked during the referenda were determined by the government, partly in collaboration with the political parties of the opposition. This feature tends to weaken the input legitimacy of the Luxembourgish process [44].

38Concerning throughput legitimacy, the 27 participants were divided into three groups of nine: one group working on the limitation of ministerial mandates, another group examining the question of the extension of voting rights to individuals aged 16 to 18, and finally one group working on the question of funding of cult ministries [45].

39The aim of the deliberative process was to arrive to a summary of the main arguments in favour and against for each topic, and to think about potential alternatives. The discussions were moderated by professional facilitators that were given the task of making sure that the debates went well and that every participant had time to voice their thoughts and opinions. Moreover, three experts were present to answer questions of the participants and to help them formulate their ideas and opinions [46].

40Globally, the aforementioned objective was reached: the debates on every topic were sufficiently inclusive to make a variety of opinions emerge and to enable a confrontation of ideas. This was confirmed through an analysis of the content of the debates as well as by the evaluation of the process by the participants [47].

41Nonetheless, some features weaken the throughput legitimacy of the Luxembourgish case, essentially the lack of time and the linguistic barrier. On the one hand, only one day was allocated for the whole deliberative process, and the majority of participants agreed that the time allocated to the exercise was insufficient. On the other hand, the environment in which the process took place was mostly French speaking, so the participants who wished to express themselves in Luxembourgian felt that taking part was difficult. However, this last element must be put into perspective because the professional facilitators that were present also had a translating task in order to make sure that participants could express themselves in the language of their choice [48].

42Finally, in terms of output legitimacy, the citizens of Luxembourg first took part in several non-binding referenda on 7 June. For each referendum, the result was a landslide victory for the « No » option. [49] The results of these referenda were not legally binding, but the government of the time held itself as politically responsible in regard to the results.

43However, in order to make a conclusive assessment of the Luxembourgish case, it would be wise to wait for the end of the process towards the end of the year 2016, when the binding referenda should in theory take place.

5 – The Case of Nepal

44Nepal’s experience with constitutional reform through deliberative democracy begins in 2008, in the aftermath of a 10-year civil war and a popular disgruntlement against monarchical rule that led to the election of a Constituent Assembly that declared Nepal a secular federal democratic republic, marking a stark contrast to over 200 years of unitary and often authoritarian rule [50]. However, it took many years to promulgate a Constitution that would institutionalise these new ideals. During over seven years, a deliberative and participatory process was more or less followed with the aim of obtaining consensus among the public and political parties about the new shape and structure of the state and its democratic system [51].

45In order to better understand the Nepalese experience of constitutional reform through deliberative processes, a short summary of the country’s characteristics is helpful: Nepal is a culturally and geographically diverse country of over 26 million people and over 100 « identity groups » based on a combination of language, ethnicity, and caste [52]. Nevertheless, Nepal’s political and social structures do not reflect this diversity, since they have been historically dominated by specific identity groups – for instance the high caste Pahadis [53]. Geographically, Nepal is characterised by mountainous terrain and poor infrastructure which has created remote and isolated regions that are economically and socially isolated. In these regions, government has had little to no reach [54].

46Moreover, Nepali political culture has often been characterised as « hierarchical fatalism », following the expression given by Bista [55]. Some groups’ culture is dictated by an absolute belief in fatalism: an individual has no control over his life circumstances, since they are determined through an external agency. This fatalism is reinforced by an extremely hierarchical caste system, ultimately displacing responsibility over one’s own life circumstances.

47Concerning the deliberative democracy experience that took place since 2008, it contrasts heavily with all the previous Constitutions in Nepal, which were all prepared by the dominant elite [56]. This time, to the contrary, the Constituent Assembly was quite inclusive, because it comprised a highly proportional contingent originating from each of the different identity groups, included a larger proportion of women, and numerous programs and forums were established to provide opportunities for information and deliberation. Concretely, the deliberations took place at two different levels: on the one hand on a popular level, where suggestions were collected following numerous discussions, and on the other hand at the elite level through the Constituent Assembly and the structures linked to it [57].

48Breen has collected data that can be used to assess the legitimacy of the deliberative process [58]. In terms of input legitimacy (i.e. who deliberates, level of information of participants, and how the agenda is set), the Nepalese experience should be analysed on two levels: at the popular level, and at the elite level. At the popular level, the idea was to have widespread deliberations through public consultation mechanisms, questionnaires, and local level forums, that would ultimately be linked with the elite level. An example of these mechanisms would be the regional « federalism dialogues » supervised by the United Nations amongst local level leaders. The aim of these types of dialogues was twofold: first to provide information and to build a shared understanding of the salient issues, upon which subsequent deliberation could then be conducted, and secondly to create group deliberation over specific issues.

49At the elite level, the Constituent Assembly was composed of 601 members, of which most were political party representatives. As already mentioned, this Assembly can be considered quite inclusive because it comprised members originating from each of the various identity groups, and included over 33% women. This Assembly was the main deliberative body, and was divided into 11 Thematic Committees and 3 Procedural Committees. On the one hand, the Thematic Committees had the task of preparing preliminary drafts of particular aspects of the Constitution. On the other hand, the Procedural Committees were tasked with engaging citizens through preparation and dissemination of information campaigns, questionnaires, and collecting public opinions on the draft constitution once it would be ready [59].

50In terms of the level of information of the participants, at the elite level, members of the Constituent Assembly made contact with experts and studied international examples in order to inform their deliberations. At the popular level, the educative components were a focus of the process because they preceded most forms of debates and discussions that preceded agreement-making. However, it is clear that a significant disconnect existed between, on the one hand the elite (i.e. members of the Constituent Assembly), and on the other hand the popular level. For example, according to surveys in 2007, 93% of Assembly members supported federalism while a mere 42% of the general public supported federalism. A striking example of the increase in understanding and awareness of the Nepalese population can be seen in their gradually increasing grasp of federalism: according to an INGO survey in 2005, only 10% of respondents had heard of federalism. In 2007, 23% of people had heard from federalism, indicating a significant increase in basic understanding of the discussed concepts [60].

51These educative components did improve the understanding of federalism and political institutions, but this statement should be put into perspective since reports were made on people’s lack of access to mass media, difficulty in collecting opinions, and problems in the delivery of resources and reference materials.

52It should be noted that numerous international and local NGOs also participated by providing information and forums for deliberation. For example, CARITAS Nepal held meetings in the impoverished region of Karnali, and the United Nations held « democratic dialogues » which focused on giving marginalised individuals a voice. This strongly reinforces input legitimacy of the Nepalese process. Concerning agenda-setting, the agenda was determined by the members of the Constituent Assembly, which means that the agenda was open for the elites, but closed for the popular level [61].

53When it comes to throughput legitimacy, which as a reminder refers to the shape that the deliberative process takes (i.e. how individuals participate, presence or not of discussion moderators, and how the deliberation translates into a decision), several features should be highlighted, especially that popular level deliberations were linked with the elite level.

54In theory, concerning the Constituent Assembly, the aim was to reach decisions by consensus. However, if a consensus was not possible, a two-thirds majority vote would then determine the outcome. Discussions were based on member proposals, but members had limited speaking opportunities. In practice, it seems that in the Assembly itself, deliberation rarely took place, but instead happened in cross-party caucuses or in the Thematic Committees. Real decisions were often made in other places through bargaining instead of deliberation, strongly weakening throughput legitimacy of the Nepalese process. Moreover, journalists and reports indicate that political parties were often involved in closed door negotiations for important phases of the process, and in using political communication to further their own interests [62].

55On the other hand, at the popular level, according to numerous NGOs such as CARITAS or the Centre for Constitutional Dialogue, the deliberative process and outcomes were more apparent. Reviews indicate that moderation of vested interests and genuine deliberation took place. For example, in the « federalism dialogues », the events were organised in a 3-day program, in which the first day was allocated to background exploration of the issues and concepts, the second day consisted in a mix of presentations and discussions, and lastly the third day took the form of group work with the aim of arriving to common positions [63].

56Finally, concerning output legitimacy, which refers to the explanation and generalisation of decisions made in a small group to the wider public, it should be noted that local level deliberation seems to have fed the elite level discussion, ultimately leading to a narrowing of the possibilities that could be considered acceptable and reasonable. Nepal’s constitutional outcome still had to pass the test of time, but it seems that the deliberative processes that took place resulted in a somewhat reasonable accommodation of individuals’ preferences and concerns. Also, the Nepalese experience shows that deliberative democracy in divided societies and state-rebuilding situations is viable [64]. However, expectations should be managed properly due to Nepal’s traditionally hierarchical society, a fortiori when it is combined with a fatalist view of individuals’ life circumstances.

Conclusion

57Deliberative and participatory processes are now increasingly used in attempts to modernise constitutions throughout the world. This constitutional deliberative democracy turn raises however several key questions in terms of the legitimacy of these processes and therefore of the reforms that they might bring about. In order to assess their legitimacy, we have used a threefold framework of legitimacy that seeks to shed light on its input, throughput and output dimension. Such framework has been applied on recent attempts of constitutional reform – some of which led to reforms, others did not – in Iceland, Ireland, Luxembourg and Nepal.

58This short overview of four recent cases of – tentative – constitutional modernisation through deliberative democracy demonstrates quite clearly that there are tensions between different dimensions of legitimacy. Being high on the input side, for instance, does not necessarily mean being high on the output dimension, quite the contrary sometimes. A high-quality throughput is not automatically the result of high input legitimacy and does not for sure lead to decisions that are enacted politically and/or legally. This calls for a thorough assessment of the design of such constitutional modernisation procedures, in all their legitimacy dimensions. There is no magic solution, but rather home-grown solutions that fit the needs of a given polity.