Model Solutions and Pragmatism in Developing ICT for Public Consultations

The focus of this article is on the process of developing the inDialogue software within the loop of communication between researchers and the clerks who organize public consultations in local governments. The software was created in response to problems diagnosed during studies on the quality of public consultations in Poland. The design supports transparent, thoughtful, and collaborative planning for public consultations in town or city halls, to create an environment conducive to informed and inclusive opinion formation among citizens. Within the project, a pragmatic approach means some degree of openness among researchers and designers in negotiating the features of the software with institutional users. Testing inDialogue in nine municipalities, we asked the following questions: (1) How do the clerks respond to the model of public consultations that inDialogue implements? and (2) How do they build a relationship with a project that intervenes in their routines? Analysis of the data from the evaluation questionnaires shows that although, overall, the clerks gave the highest rating for the software’s ability to introduce order into the process of public consultations, they often complained that the features behind the structuration effect were time consuming. Depending on the city or town hall, more weight was given to one or another aspect of the tradeoff. No less important for the institutionalization of deliberative public consultations is the controversy over registration and verification, and consequently, the recruitment of participants. In the article, this is discussed in the context of the ambiguities in the law, and the different values that the various local governments attribute to participation. Moreover, the study demonstrated that prior experience in public consultations combined with an openness to experimenting, a repertoire of skills in communication and data analysis, a motivation to join the project, the length and depth of collaboration with researchers, and the direct involvement of a decision maker all affected clerks’ comprehension and acceptance of the use of inDialogue.


Introduction
In this article, the author discusses the development of the inDialogue software, which supports the organizing of public consultations, and its testing in nine towns and cities in Poland. The aim of the creation of the platform was to intervene in the practice of local governments by introducing procedures consistent with the model of deliberative public consultations, and with general democratic standards such as transparency. InDialogue was designed within the process of an iterative collaboration that joined the researchers' actions with the public administration's feedback. The author presents the process of negotiating the software's features as well as the negotiated elements of the interface being problematized. The research questions that guided the inquiry are (1) How do the clerks respond to the model of public consultations that inDialogue implements? and (2) How do they build a relation with a project that intervenes in their routines?
The theories and practice of institutional design, and the adaptive approach of public institutions toward democratic innovation create a relevant context for the study. Deliberation in public consultations is a democratic innovation that the author focuses on for several reasons. First, public consultations are an institutionalized form of civic participation. Second, they are used in Polish municipalities in many areas of public policy. Third, deliberation increases the quality of public consultations. And, fourth, high-quality public consultations may be a powerful incentive for citizens to get involved in decisionmaking processes in their communities, as well as contributing to the legitimacy of public policy decisions. The application of information technologies in support of organizational change, that is, to implement the model of deliberation in public consultations, is an additional aspect that is pertinent to this article.
Considering the various factors that may affect the performance of public consultations, one cannot overlook the role of civil society. In Poland, the coalitions between non-governmental organizations (NGOs) and governmental institutions have drafted codes of conduct like the Seven Principles of Public Consultations (Ministerstwo Administracji i Cyfryzacji 2 2013) or The Canon of Public Consultations (Fundacja Inicjatyw Społeczno-Ekonomicznych 3 et al. 2015), which aim to impact the law and practice. It is possible to find references to these documents in at least some local resolutions on public consultations. The cities and towns, however, opposed a national regulation, which was discussed with the involvement of presidential office around 2011 that was to standardize the mode and methods of public consultations. Swianiewicz (2012) explains about the engagement of local governments in some democratic innovations that, in principle, signify a shift from local government to local governance using the flow of ideas from Western European to Eastern European countries due to exchange programs, cooperation, and the requirements for receiving EU funds. The spectacular diffusion of participatory budgeting in Poland, mainly in its plebiscitary form, directs us to another explanation for the new practice. Politicians may expect that their political capital will grow, as citizens receive swift returns-in the form of many small-scale projectsfrom the investment of their votes. Deliberation, which often demands the participants' verification and selection as well as the preparation of information materials and moderated meetings in small groups, is more challenging to local governments in respect to resources and qualified staff. Nevertheless, following Åström and Grönlund (2012), and acknowledging the spectrum of the various attitudes and approaches to democratic innovations in local governments, one may expect some examples of 'strong democratic intentions,' even if they do not lead to the immediate implementation of the model of deliberative democracy.
By the means of comparing various methods of deliberative public consultations (Gastil & Levine 2005;Gastil 2008), as well as in reference to legal provisions, one can list the following characteristics of public consultations, where points 6, 7, 8, and 10 refer to procedures conducive to deliberation: (1) Public consultations are the process by which governmental agencies collect information from citizens.
(2) They are usually initiated by governmental agencies, and sometimes by citizens, after meeting some formal requirements such as public support measured by the number of signatures of lay citizens. (3) They are regulated by the law and require a political decision to get started. (4) The general public should be informed at the beginning of the process as to what information the governmental agencies are seeking from them, and how they can contribute to the decision-making process.
(5) The information collected concerns the needs, opinions, and preferences of citizens regarding public goods and services. (6) The process should have formal elements that contribute to an increase in the level of citizens' knowledge before they express opinions, like sharing information materials containing the main facts and proposals for consideration, and access to experts. (7) The targeted or random selection of participants is possible, and they are often recommended or required. (8) Information is gathered using various methods and techniques. The elements of dialogue between representatives of the different social groups are, however, in focus. (9) Debates should be moderated to support the equal, respectful, and reciprocal exchange of opinions and arguments. (10) Information should be systematized, generalized, and presented to the public. (11) The public can comment on the reported results or even contribute to the formulation of conclusions and recommendations. (12) The public should receive feedback on how the results of public consultations contribute to the decision-making process; the contribution should be justified.
The statement by Coleman (2012: 389) that ' deliberative exercises generally take the form of social-scientific experiments (deliberative polls, citizens' juries, and consensus conferences) rather than institutional pillars of democracy', adequately describes the overall situation, although it requires ongoing verification in reference to the changing institutional practice of local governments. The survey data that we collected in 2011 in 270 Polish local governments showed, as in other democracies, that public hearings, written comments to plans and documents, and questionnaires (polls) were most commonly used in public consultations. Moreover, the data analysis revealed the problem of a low level of transparency in the local governments' performance. For example, citizens had access to information on the subject of public consultations in 54% of events, the reports presenting results in 60% of cases, and the results connected to decisions in 58% of documents. The findings from the survey, as well as subsequent qualitative research, laid the ground for the inDialogue project.

Developing Information and Communication Technology for Public Consultations
Information and communication technologies (ICTs) have modified the conduct of public consultations by making available new channels for collecting the opinions and preferences of citizens. The use of ICT by local governments broadens the scope of choices for citizens who want to share their opinions on issues of public interest. It may serve to strengthen the public participation of groups with special needs (Davies et al. 2009).
In defining e-consultations, Shane (2012: 161-162) writes 'We use the term online consultations to refer to Internetbased discussion forums that represent government-run or at least government-endorsed solicitations of public input with regard to policy-making'. In the Polish context, the statement that public consultations involve a decisionmaking agency in the activities for collecting data from citizens is central. Following the trend of regulating public consultations, I would add information on their formal status to the definition. I would also broaden the definition of e-consultations by including nondeliberative forms of data collection like polls, although the use of deliberation is recommended.
As in traditional public consultations, the methods and techniques for data gathering in online consultations may vary. It is important to note that to strengthen public consultations, the online forms of participation should be given the same formal status as traditional forms. To make the most of the available channels of communication would require the standardization and parallel use of both the online and offline forms of participation in public consultations to increase the social impact (Åström & Grönlund 2012).
In Poland, on the official Websites of towns and cities, one can find text forums, forms to send a question to political representatives or comment on the drafts of local policies, as well as polls. Some larger cities have online platforms dedicated to public consultations. Noncommercial applications like I Have an Opinion, Citizens Decide, or Station-Consultations have been developed by NGOs, sometimes in cooperation with units of the national government, for use by public authorities. Nevertheless, the opportunity to participate in dialogue on moderated, threaded text forums during public consultations had been created on only 1% of Websites across all the Polish towns and cities we analyzed in 2014.
Researchers studying the relevance of ICT to deliberation and public consultations approach the subject from different perspectives. It seems that their main focus is on online discursive spaces and the processes that occur there naturally or are a result of an experiment. An analysis may cover social inclusion and reciprocity, as well as the knowledge gains and changes in opinion among participants of the interactions. Issues such as agendasetting mechanisms, synchronous or asynchronous modes of communication, moderation, rules for content acceptance, and framing using a particular form of organization of the discursive space, for example, threads, are also covered (Wright & Street 2007;Zhang, Cao & Tran 2013). Alongside these studies there are projects that have resulted in platforms being developed involving different, formal solutions to encourage argument-based discussions; some examples of these are Deliberatorium, 4 designed in the United States; Onlinedeliberation, created in Singapore; and D-BAS, 5 which originated in Germany; among others. Finally, applied research in online deliberation involves collaboration between scholars and local communities or authorities to facilitate ICT-supported public communication, collaboration, and transparent decision-making (Davies et al. 2009;Stromer-Galley, Webb & Muhlberger 2012). At this stage of the research maturity, all authors, independent of their focus, take into consideration the social and political shaping of technology next to the impact of ICT on the interactions and behaviors of social and political actors.
As mentioned before, the inDialogue software has been developed to respond to the problems revealed during a study on the quality of public consultations. It is embedded in the particular context of institutionalized processes of civic participation in local governments, as with other software like Decidim, 6 created in Spain, or the online platforms referred to earlier in the text and designed in Poland. In what follows, I characterize its main features and functions to create a background for the analysis of the data collected during the inDialogue project.
InDialogue's distinguishing characteristic is that it supports the entire process of public consultations, from the moment when the local government decides to organize them, to the publication of the final results. Its design supports transparent, thoughtful, and collaborative planning for public consultations in a town or city hall to create an environment that is conducive to the development of informed and inclusive opinions among citizens. Figure 1 shows the actions that clerks and citizens can take using inDialogue during public consultations in connection with the norms of deliberation, which informed the design of the interface.
The clerk's actions are executed through the clerk's panel, which helps to: (1) prepare a plan for public consultations, including a goal, lead questions, the identification of social groups affected by the policy, the criteria for participants' selection, a method for verifying participants, an anonymity mode (of citizens to each other), the division of tasks within the local government (e.g., moderation, internal experts), the selection of external experts from a database, the selection of online and traditional tools for data collection, and setting the schedule; (2) draft briefing materials for participants; (3) participation in the discussion on the plan for consultations and briefing materials within the group of clerks and invited experts; (4) receive feedback on briefing materials, respond to comments, and publish the final version; (5) moderate discussions using tools for text or voice group debates; (6) collect responses to questionnaires; (7) integrate data collected online and in traditional communication (face-to-face or on paper); (8) analyze data, also using an argument map for text debates, and draft a report; (9) receive feedback from participants on the report, respond to comments, and publish the final version.
In turn, the citizen's panel invites citizens to: 1) comment on briefing materials; 2) select a channel in which to participate in a debate and be part of the discussion; 3) fill in questionnaires before and after discussions (if relevant); 4) comment on reports. 7 Citizens can choose to participate in one of four forms of moderated debate: face-to-face, online text, online voice, and an online debate on the consultation document, for example, an act of local law. As moderation is an important form of intervention in the discussion, it may also be relevant to shed some light on this topic. For voice  Concerning the text forum, it is worth noting that the space for discussion is threaded by questions that organizers formulate during the planning of the debates. Participants are required to use arguments when they present a proposal or respond to somebody else's proposals or arguments. The division of the input template into categories (proposal, question, argument for, and an argument against) makes it possible for a text debate to be automatically transformed into graphical form. A moderator can move participants' contributions to the debate from one section of the forum to another, and edit or remove posts. All posts that have been removed are saved and can be viewed along with the moderator's comments, but are not part of the discussion itself. Each intervention has to be justified.
InDialogue strengthens the transparency of procedures and processes at each stage of public consultations by allowing users to access the plan, methodology, information materials, recordings from meetings, possible transcripts, and reports; as well as by assigning roles and tasks in the city or town hall. The archive of public consultations, including datasets, can be browsed according to predefined criteria. Apart from transparency, four norms of deliberation were addressed in the design of inDialogue: openness, inclusion, reflexivity, and reciprocity.
The software implements the norm of openness in public consultations, at the level of the institution, by enabling access to the plan of consultations so that all clerks can take on the role of an internal expert, comment on the methodology (e.g., by adding questions to questionnaires), or contribute to briefing materials. All units in the city or town hall can use a shared database of external experts, including NGOs, to broaden the knowledge on the issue under consideration. Media who are alerted by the inDialogue platform can observe the process, help to reach out to people concerned by the issue, and scale-up deliberations. In reference to members of the local community, inDialogue implements openness by allowing everyone to register and join public consultations as long as the person meets the criteria of participation in a particular project. The level of participant verification should be justified by the aim of the consultations. A system of alerts keeps citizens informed about upcoming events. Commenting on briefing materials, which can also be printed in hard copy, is open to all members of the community. Furthermore, citizens can choose a form of involvement and sign up for an online or face-to-face debate, according to their communication habits and preferences. The software was developed in line with accessibility standards and involved people with disabilities who tested its openness.
The inDialogue platform has been designed to support inclusive, reflexive, and reciprocal interactions. These occur at the planning, conducting, and summarizing stages of public consultations. They involve clerks who discuss the plan and briefing materials, as well as citizens who comment on the briefing materials, then participate in a discussion on the subject of the public consultations, and finally, comment on the report.
At the level of the institution, I understand inclusion, in deliberative public consultations using inDialogue, to mean equal opportunities for all clerks to engage in interactions with citizens regarding the plan for consultations, as well as their collaborative work on information materials. All clerks, unless they play the role of a lead organizer, have at their disposal the same tools to impact the form or content of public consultations. In regards to participants, access to briefing materials helps them to strengthen their arguments during discussions, and increase their impact on decisions, irrespective of their education or general knowledge. While preparing the information input, clerks are reminded by a system of instructions that the content should not exclude any citizen due to any lack of language skills or civic literacy. In online debates, participants can be anonymous in regard to each other or not, depending on the subject and other circumstances, which requires consideration in relation to the level of participation and the quality of input. Uninhibited, equal, reciprocal contributions during an exchange of opinions can be strengthened by moderation. Small, moderated group discussions are favored by the setting options on the clerks' panel.
InDialogue is expected to have a positive impact on the reflexivity of public consultations, at the level of the institution, which is due to the software's educational function. The system of templates and instructions help clerks to follow a procedure and prepare the process of data collection thoughtfully in connection with its goal and the social impact of its results on a decision. The exchange of comments between clerks who represent different departments should facilitate the competent preparation of the plan and briefing materials. At the level of group discussions, reflexivity is about the exchange of arguments based on reliable information. A template for briefing materials divides them into sections that present basic facts as well as legal, budgetary, and other limitations to the policy. A system of pop-ups reminds users of their availability. The possibility for clerks and participants to use multimedia content should increase the appeal of the information. Both moderation and a text forum oriented toward the exchange of arguments are additional factors that can have an impact on reflexive participation. The rating of posts in text debates is possible only after all proposals are published, which should eliminate the priming effect. A graph of arguments can be useful for participants in forming their opinions, and for clerks to prepare reports that embrace all proposals along with their pros and cons. The fact that briefing materials and reports can be commented on should increase their impartiality (for more information see Przybylska 2017).
In my view, the norm of reciprocity is reflected in the considerations about inclusion and reflexivity. Interactions via alternative channels of communication that facilitate different forms of expression focused on merit, and conducive to the exchange of arguments, should encourage the respectful and mutual exchange of opinions. This hypothesis should be tested in another study.

Methodological Approach
Connecting research with practice, and sometimes also with the evaluation of intervention, is well-represented in studies on civic participation, including online participation, deliberation, and collaboration (see, e.g., Davies et al. 2009;Fishkin 2018;Macintosh & Whyte 2008;Noveck 2009;Stromer-Galley et al. 2012). The inDialogue project focuses on designing an ICT system with local authorities to change their practice in public consultations at each stage of the process by the inclusion of procedures conducive to deliberation. It contributes to existing knowledge by seeking the answers to the following research questions: (1) How do clerks respond to the model of public consultations that inDialogue implements? and (2) How do they build a relation with a project that intervenes in their routines?
Action research seemed the most appropriate methodology to apply to a project that was aimed at the implementation of new procedures for public consultations in local governments. The action research project's subsequent stages (a research-based problem definition, the plan to modify unsatisfactory practices, intervention, evaluation) are often represented in the form of a spiral, as they may be repeated until the problem is solved. The pragmatic approach of the project means researchers and designers are open to negotiating features of the software with institutional users, as long as the model that inDialogue implements remains intact.
The procedure for designing the model for the inDialogue software was distinct from the action research in one important way. The model was based on the results of studies that diagnosed the quality of public consultations, including a survey conducted in 270 town and city halls, followed by group discussions, participatory observations, and interviews. As the criteria for this evaluation referred to the standards described in the literature on deliberative democracy and democratic institutions, the impact on the research tools by the researchers and by city or town hall representatives was unequal. The elements of dialogue, however, had an essential place in data gathering and subsequently in the inDialogue Project.
The presentation of the model and prototypes of the inDialogue software were followed by evaluations by the staff of the town and city halls. We started with a questionnaire that verified (1) the problems that we had identified during the diagnosis and (2) the adequacy of the solutions presented in the model for the inDialogue software. The first prototype was tested in laboratory conditions and discussed during a face-to-face meeting with clerks from nine municipalities. The second prototype was tested in two municipalities of different sizes. And the third was used in pilot consultations in eight municipalities. During the pilot studies, and shortly after, we collected information through participatory observations, interviews, and using questionnaires filled in by clerks. Figure 2 shows the cycle of action and evaluation in the inDialogue project.
Next to iterative design, another characteristic of action research is the cooperation between all the parties involved (researchers and practitioners of different kinds) (Reason & Bradbury 2008) to develop practical knowledge about the needs of the institutional and social partners  and their preferred course of change. During the long process of research, coding, and feedback, the first critical moment was when we were preparing for the pilot use of inDialogue and some towns and cities decided to leave the project. Several reasons were behind these decisions, including: (1) coinciding political change after elections and the loss of support for participatory processes beyond participatory budgeting and surveys, (2) the intention to build their own software using their own grant money and experience from the project, (3) the workload the clerks became aware of at the moment the prototype was presented, and (4) the anticipated challenges regarding the implementation of the software in city and town halls. We continued the project with the remaining municipalities and those who left were replaced by others. The problem for the team was that the newcomers had not contributed to the discussion on the model behind inDialogue in the earlier stages of the project, and did not have any knowledge about the software and so had to be introduced to it.
The cities and towns that tested inDialogue during public consultations represent seven regions in Poland. Two cities are the capitals of regions, and two more were capitals of regions at the time of a former administrative division of the country. Their populations ranged from around 91,000 to 220,000 inhabitants. In addition, six towns with populations from 4,500 to 40,000 inhabitants participated in the pilot use of inDialogue. Four local governments joined the project in 2014 and six in 2016 (replacing the ones who left). Five municipalities, including all the cities, had a resolution regulating the mode and method of organization of public consultations when they accepted the invitation to join the project. Cities have a unit dedicated to public consultations, whereas in towns, public consultations are usually within the scope of the responsibilities of one person who often also has other tasks. Although all local governments have their Websites, and cities have platforms dedicated to public consultations, only one organizes online public debates. All towns and cities have their Facebook profiles and declare that they organize face-to-face meetings with citizens as well as online and traditional polls during public consultations.
While establishing a working relationship with municipalities, we took into consideration that public consultations are initiated by politicians, who also decide on how the information received from citizens is used, whereas it is the clerks within the public administration who are tasked with managing the process. They occupy a central role in designing the methodology of information collection and processing. This is why the clerks' knowledge and competences, as well as their conceptualization of public consultations, as set in the broader legal and institutional context, have important practical implications.
To answer the research questions posed in this article, I will analyze data that came, first of all, from a mostly open-ended evaluation questionnaire that was filled-in by clerks in the city and town halls that used inDialogue in public consultations in 2016 (tests) or 2017 (pilots), and evaluation interviews. Occasionally I will refer to a questionnaire filled in by clerks in 2014, and the results of the participatory observation as a method of data gathering. The research is qualitative, and its results are not universal for all Polish municipalities, although with their use, I try to explain some of the mechanisms behind public consultations in Poland. I have referred to the distribution of opinions among clerks to add to the explanatory value of the data.

Evaluation of New Tools and Procedures for Public Consultations
In this part of the article, I have addressed the research question concerning how the clerks respond to the model of public consultations that inDialogue implements. Specifically, this has been done by analyzing which ICT tools they identified as useful and which they wanted to modify or abandon, and why. As I attempted to demonstrate earlier, we aimed to design the software in connection with the standards of deliberation. To create a context for the analysis, first, I have shed light on the problems that clerks reported in referring to their general practice of organizing public consultations, then discussed their choice of tools for the performance of tasks in the project.
The concerns about how public consultations are conducted, which clerks listed in response to researchers' questions, 8 can be divided into two groups: one involving communities and the other municipalities. A problem noted by seven out of nine clerks, and which had the highest position in the ranking, referred to the low level of citizen involvement. An open question allowed the clerks to express nuanced opinions. These presented the above problem as 'general passivity,' 'the low return of questionnaires and the low level of participation in meetings,' or 'resistance to engage, especially in consultations regarding strategic documents.' The low quality of citizens' input due to a 'lack of deliberation,' 'mostly negative comments,' and 'ignorance of the regulations' was also mentioned. In addition, three clerks pointed out the difficulty of reaching inhabitants with information about the public consultations. From further studies, we learned that this problem could mean different things in various localities: while some clerks complained about the limited cooperation of local media, who had allegedly expected financial compensation for publishing news; others meant the different levels of attentiveness of citizens to news about public consultations, and the effects that this could have, especially for underrepresented groups.
Next, the internal problems of city or town halls with organizing public consultations, listed by three respondents, concerned (1) the range of methods of data gathering, which were usually limited to public hearings and comments on documents; (2) uncertainty as to which method of collecting opinions should be chosen to draw meaningful conclusions; and (3) the self-selection of participants. The remaining problem brought up by one clerk involved a low budget for performing tasks. Even more self-critical opinions-if we take into account the fact that respondents represented the institutions whose activities they described-came from the large or middle-sized city halls that had left the project before the pilot study. These included (1) a poor understanding of the purpose of public consultations, or even, an inability to distinguish between them, and the provision of information via an official bulletin; (2) a deficiency in skilled staff who undertake tasks and handle them competently; (3) insufficient support from the clerks responsible for the subject areas in which consultations are initiated, and their lack of cooperation with people who have the knowhow; (4) the lack of a coordinated effort between different units of the city halls in conducting public consultations; and (5) the risk of ' distorted results' due to a lack of knowledge about the methodology. The clerks responsible for public consultations in these city halls were aware of the problems, but they were either unable to introduce profound changes, or they anticipated the costs of the change, including the costs to themselves.
The other element that helped to interpret clerks' feedback on the software was the actual use of inDialogue in public consultations. All clerks planned public consultations as well as created briefing materials on the platform, whereas only one city and one town also consulted citizens on the information input. Although the plans for public consultations indicated the involvement of internal experts in public consultations, they did not participate in the exchange of comments using the software. Significantly, only 6 out of 9 municipalities ran an online text debate, and only in 3 were there some exchange of opinions, while the highest number of participants was 12. A further, 3 local governments proposed an online poll with an equally poor return. In all towns and cities, there was at least one face-to-face debate, with the number of participants ranging from 3 to 45.
Whenever researchers moderated debates, the meetings had the form of discussions; others were organized in the form of a public hearing. Paper questionnaires, which were distributed in 2 towns, brought up to 61 responses. Finally, all cities and towns published reports on the platform, but they either did not ask for comments or did not receive them.
Below I have analyzed the answers that clerks gave to our questions after the completion of public consultations using inDialogue. They have contributed to our knowledge about how representatives of local governments reflected on their experience, and what their concerns were. We asked nine of the partner municipalities about their opinion on the practical value of the inDialogue software's functionalities.
Although no municipality had a negative opinion about the application, only one expressed unconditional satisfaction with it. Eight out of nine municipalities were satisfied with it, but had some objections. 9 Table 1 shows the list of advantages and drawbacks of the software presented by respondents who filled in the evaluation questionnaire.
Clerks particularly appreciated the features of the inDialogue software that facilitated proper planning, but, simultaneously, some disliked the application because it involved them in new activities that required more of their time. This tension manifested itself in the following comment from a clerk from Town 1: 10 Certainly, it forces a clerk to thoroughly prepare it [the public consultation] as it should look like. So, certainly, when it comes to the tool, it is obviously a plus, because it forces someone [to do things]. But, on the other hand, the same is a minus, because it is time-consuming.

Advantages
Supports consistency in consultation planning; requires users to follow a plan 4 It is easy and intuitive to use 3 Accessible to inhabitants, who can share their opinions from home 3 Provides easy and instant access to citizens' opinions 2 Supports the preparation of briefing materials, organizes them logically, which contributes to the reliability of information 2

Supports transparency 1
Supports dialogue at each stage of consultations 1

Drawbacks
It is new, and citizens are accustomed to other forms of participation 4 Does not allow for the omission of some procedures 3 Requires clerks' time and effort 2 Requires registration, which discourages participation 2 Citizens who are not logged in cannot be observers 1 Does not allow the automatic transfer of all data to the final report 1 Does not allow results to be received instantly 1 It seems that the idea of parallel online and face-toface debates that included various groups of citizens did not appeal to everyone. Already, at the early stages of evaluation, some clerks had raised the argument of additional work due to the larger volume of empirical material. Also at that time, using ICT to plan face-to-face debates, among others, was new to some clerks and was difficult to understand. During the evaluation interviews some respondents presented their particular view of the citizens' preferred forms for sharing their opinion with city and town halls. By doing so, they compared them indirectly with inDialogue. They highlighted the convenience of face-toface meetings, polls, and Facebook, which is illustrated by the following statement by a clerk representing Town 2: I think that a barrier to such consultations [i.e., using inDialogue] is that you have to sit down and write something that would be easier to say; or some people have resistance to longer statements. Writing two sentences on Facebook somewhere; or ticking 'yes', 'no', 'I do not know;' or 'I really want', 'I do not care'-the simpler the form, the easier it is to gather a larger number of responses. Well, of course, the in-depth comment will bring more substantial knowledge.
It appears that most clerks focused on the quantity of input from citizens rather than its quality. The level of participation was their main concern before the pilot study, and, apparently, it remained so after the new experience. Also, clerks were rather unwilling to negotiate their own practices in return for potentially higher participation (especially in view of the postponed effects) whenever it meant more work (with some exceptions). The expected higher quality of input might not have been an incentive for all.
The expectation that citizens would be capable of acquainting themselves with the new habit of using online tools for public consultations is better represented in the city and town halls where clerks were open to learning new skills, albeit with some hesitation, which is shown by the opinion expressed from City 2: It is interesting that people are happy to use social networking sites, and to get involved in this type of venture [online public consultations] it is more difficult for them. Well, maybe they have to learn it as something new. Yes, it probably looks like that.
Their explanation for the low participation of citizens included the citizens' daily duties, and, in connection with this, a preference for forms of communication that were low in engagement, as well as the topics that could have a direct influence on their quality of life; and a generalized lack of civic culture. In turn, the research team, which focused its observation on city and town hall routines, would add that citizens were not always notified about upcoming consultations through all the communication channels that local governments have at their disposal. The information might have been misleading or incomplete. Also, there was some resistance to engaging, for example, social partners to reach out to residents. On one occasion, when a representative of the local community expressed an eagerness to involve other members in a particular issue, it was withdrawn from the agenda of public consultations. We may speculate that the citizens' motivation for participating could have been further decreased by announcements like the one in Town 4 about the experimental use of inDialogue, or as in City 1 about the software's use outside of the legal framework for public consultations (meaning less impact on the decision). We may doubt that all city and town halls welcomed higher activity from citizens in public consultations that they first wanted to test and evaluate for themselves.
The anxiety about a decrease in citizens' input might have been one reason why, during the pilot use of inDialogue, none of the municipalities used the software's features for the verification of citizens. This was in contradiction to the argument put forward by representatives of local governments that the usual group of active participants was homogenous, and that inDialogue's features would enable intervention. Certainly, clerks were concerned by the lack of adequate legal provisions for local governments to use personal data to conduct public consultations. Local governments, however, vary in their practices and quite a few (some also participating in the project) verify the identity of citizens who vote in participatory budgeting. From the technical point of view, the problem should be solved by the country's public administration introducing Trusted Profile.
In the project, we have also learned that any form of de-anonymization of participants may meet with fervent opposition by local activists, and clerks attempt to avoid open conflicts. We presented the project team's opinion that using the software for registration, verification, and (usually) de-anonymization among participants should be encouraged; but we have left the choice to public consultations' organizers as to whether to use the last two functions. The following statement from a clerk representing City 3 is interesting not only because it is in support of the de-anonymization of participants, but also because it proposes a specific solution, that is, the status of an observer, which was later introduced by designers: People should register, give their names, should say who they really represent in this process; then everything is easier [because it is transparent]. In turn, those who do not want to actively participate in consultations by speaking or commenting on what appears, should not be forced to provide any of their data, but should be able to observe … because maybe the fact, that they are reading some discussion in which they cannot participate, will cause them to state: I want to join in because they speak wisely, I have to support what they say; or, it doesn't make sense what they say; I have to oppose it; then they will register.
Another amendment to the software that the clerks expected concerned a simplification of its functions. For example, respondents requested more flexibility regarding the schedule. We tried to be responsive to the clerks' needs, but we could implement only those changes to the software that were compatible with the model that we support. Thus, we could not allow the preparation time for public consultations to be shortened, even if some clerks argued that 2 weeks for the whole process, including recruitment of citizens and reporting of results, should be enough. We understood that this demand mirrors the expectations of politicians, who often build the agenda for public consultations disregarding timeconsuming duties. We have made one adjustment in the registration settings to enable citizens who are late to sign up for text debates, enabling them to join upon an invitation from a clerk. We think that registration is necessary for the proper planning of small group debates. The change we have, however, made will not disturb the reflexivity of asynchronous communication as the number of moderators is usually fixed and the briefing materials can be read at any time during the debate.
In their comments, our respondents were mainly strengthening the existing practices regarding public consultations in city and town halls, rather than challenging it. For this reason, we have not accepted the proposed changes to templates for briefing materials and a report. We may consider some modifications in the near future, but will not eliminate core elements. After reflecting on the features of inDialogue, their own daily practice, and the public expectations they had, however, generalized, some clerks proposed new tools, like a debate on documents, which designers have added in response to this request. In addition to this, we have implemented many small fixes to make the inDialogue software more versatile and easier to use. In particular, we have been very attentive to all aspects of the use of the software that concerned local government legal obligations (e.g., regulations for personal data protection-RODO, accessibility for people with disabilities-WCAG 2.0), or had to comply with administrative tasks like the circulation of documents, and the use of geographic information system (GIS) maps or other software.
It is useful to reflect on the ICT tools that were not used during the pilot public consultations, although they can, and even should be, applied in every process. This is the case with features that allow comments to be made by invited municipality staff on plans for public consultations and briefing materials. This is inconsistent with the earlier opinions of the clerks responsible for public consultations in some municipalities that communication and cooperation are weak between them and clerks representing the various departments of city or town halls, and that whenever other units collect data from citizens, they do not keep up to standards. This is the case even in City 2, recognized as one of the most open toward public participation: We are trying to plan consultations, but not all departments plan them. They sometimes are organized hurriedly. So [the inDialogue software] would paralyze us. The fear paralyzes, at first, but later it seems that, well, it may be easier. If someone is systematic, then really this tool can help us to channel the ideas somehow.
The clerks responsible for public consultations did not use the opportunity to challenge other clerks by trying to introduce some elements of change into the practice of their local governments with the help of inDialogue. It might be that there were too many new procedural elements, and that internal interactions can take place at a later stage of inDialogue's use, as was recommended in City 3.
Nevertheless, our respondents also rarely sought feedback on the inDialogue software from other clerks, representatives of NGOs, or citizens. This did not prohibit them from making generalized assumptions concerning their opinions and preferences. These were most likely grounded in prior experiences that did not relate directly to the use of inDialogue in public consultations, and usually supported an argument against the implementation of both the software and changes in public consultation procedures.
Based on data collected during the entire project, it appears that the clerks responsible for public consultations tend to be cautious and selective regarding the profound transformation of local public consultation procedures. Especially among new towns, not all elements of the model were noticed or correctly understood, which was confirmed by some counterfactual remarks regarding the tools. Also, there were some conflicting remarks, especially regarding the verification of participants, which was rejected as it could lower the level of participation, and was expected to prevent the participation of residents from other towns and cities, or homogenous groups of their own residents.
To sum up our findings in reference to the model of deliberative public consultations that inDialogue supports: first, I have to admit that, even though at least some clerks reflected on the pros and cons of the implementation of the software by public administration and its use by citizens, not even one city or town hall followed all elements of the procedure.
At the level of the institution, openness for launching collaboration among clerks as well as between clerks and external experts on the platform was not apparent. It was also not confirmed during interviews. Two to three main organizers and an invited group of internal experts were included in the reflection on the plan and briefing materials prepared using inDialogue; however, they communicated using conventional channels instead of inDialogue, which was not expected. We do not know how reciprocal this exchange of information and opinions was. For some towns, the preparation of a report that is publicly accessible was a new experience. The content of this documents was, in its main points, influenced by the use of inDialogue (e.g., the automated transfer of information on the methodology of public consultations), and in this way we might have intervened in the practice of local governments, making it more reflexive. Only two city halls, however, listed replies to proposals brought up by citizens, mostly following the standards adopted by their local governments.
It is difficult to draw substantial conclusions on the impact of inDialogue at the community level. First, city and town halls did not much change their usual pattern of interaction with citizens during the experimental public consultations using inDialogue, and did not use the available tools to their full potential. The openness of local governments to participation by citizens requires the activities of city and town halls to be further investigated in connection with the unsatisfactory level of participation. The platform may support direct recruitment if there is a database of users. Based on available data, we can confirm that residents who participated in online or face-to-face discussions received briefing materials prepared using the software's tools. For most of the local governments it was a new practice to develop the information to this level of detail, and in a language that is understandable to lay citizens. The analysis of an online text debate in City 1, where it was most vivid, shows that it differed from the interactions held casually on the city hall's Website. The number of participants was lower, but they showed respect for each other, and arguments were substantial (Sierocki 2017). This could be explained by the features of inDialogue, or by the effect of novelty, or the involvement of the university as a co-organizer, or by the social composition of the group of discussants. As this project was mainly focused on the performance of clerks, the performance of citizens during discussions using inDialogue needs a separate in-depth analysis.

Local Governments in the Loop of Communication with Researchers
By proposing the inDialogue software and the procedures of public consultations that it supports, we created a space for new patterns of interactions between local governments and citizens based on norms of deliberation. In the previous section of this article, I reported some learning effects among clerks and researchers resulting from discussions about specific features of inDialogue. In the following paragraphs I have analyzed how clerks build a relation with a project that intervenes in their routines. Specifically, I investigate the learning effects of the project for clerks and the roles they adopt in situations demanding their being positioned toward change.
In the project, the research team was engaged in multiple activities that required both ongoing face-to-face and online interaction with local governments. We had to consider the diverse institutional contexts of potential change, and plan for different support and interventions during the pilot study. In town halls, more support was required because of the absence of public consultations units and the lower skills and time resources of the person responsible for organizing the process. The strong involvement of the mayor in one of these, however, demonstrated the importance of the right attitude in task completion. While in another city hall, considerable ongoing intervention was needed due to repeated changes in the person in the position responsible for the input to public consultations and task management; in two other cities, these problems did not occur, and our support did not exceed training and feedback.
Beyond collecting data from clerks, the team conducted training for municipality staff involved in public consultations, in whatever role; training for NGOs; and people with disabilities. Furthermore, members of the research team provided ongoing help with activities regarding the use of the inDialogue software during public consultations (e.g., planning of methodology, preparation of briefing materials and reports, as well as, during situations where clerks were inactive, the dissemination of information on public consultations through different communication channels). Undoubtedly, our involvement in the public consultation process, especially in the preparation of documents, was higher than we expected, and we asked ourselves a question about its potential impact on the educational effects of the project. On the one hand, we became aware of the missing patterns for what we consider to be proper briefing material containing impartial information comprising basic facts and references to reliable sources of information, pros and cons of alternative actions, or a proper report with all of the arguments that appeared during discussions included; on the other hand, we realized that our standards may be illustrative, but the probability that they will be fully adopted is not very high.
Clerks treated the research team who were involved in this system as a carrier of new ideas, which, in some instances, were better understood (e.g., planning, preparing briefing materials), whereas other ideas were miscomprehended (e.g., simultaneous use of the same questionnaires to collect information online and offline). We were cast in the role of innovators whose work had to be looked at through the prism of real practice. 'Real' meaning, coming out of first-hand experience in the office and not the research, even if its results included many observations collected over a long period of time. Understanding the rationale behind proposals does not necessarily mean their adoption in the future, as some are treated by a section of partner institutions as unrealistic (small group discussions, online or even offline, briefing materials containing the main facts, and information about the limitations of solutions). For example, inadequate resources-as the clerks declared-might preclude the use of a tool in regular practice.
At times the distinction between the roles of a clerk and a researcher was introduced by representatives of municipalities to explain the differences in approach to some tasks in public consultations, like reporting. The clerk from Town 1 acknowledged: If I were to focus on such a comprehensive report … we are interested in the result, without such a detailed analysis. Case analyses, what would happen, if it were this or that way, what would be the results. We cannot do such detailed reporting. Of course, we are grateful to you, as you put in so much work and showed what it should look like, but in reality it looks a bit different. To do such reporting, as you did, someone needs to be trained because such a detailed manner of reporting requires knowledge and time; so, simplification, possibly.
In the project, the principal difficulty was to find, with local governments, a common definition for standards for public consultations, and an agreement regarding the feasibility of change toward the application of norms of deliberation in practice. Although it may seem that members of the project team and the clerks shared the goal of increasing the quality of public consultations, due to the various definitions of the situation and the motivations for joining the project, this may be achievable in varying degrees. The liminal moments for the research team concerned uncovering the motives underlying the organization of public consultations by local governments, understanding their rationales for particular methodology choices, in some instances connected with legal barriers or lack of clarity, as well as a measure of success. In conversations with clerks around the software's use, we were able to reconstruct the model for public consultations that they support.
Six out of nine municipalities admitted in the evaluation questionnaire that the use of the inDialogue software in public consultations had an impact on their knowledge about public consultations. Only two municipalities indicated that they gained new knowledge about the procedure of public consultations. This included public consultation planning and the use of preconsultations for commenting on briefing materials. The same issues were pointed out in the answers to a question on changes in attitude to public consultations. To the group of new issues that were not indicated by clerks but were observed by researchers during interviews, and supported by viewing the platform, one should add research questions that operationalize the public consultation aim and demonstrate what information the municipality seeks, as well as, to a lesser extent, group discussion questionnaires that are not usually used, as public hearings dominate face-to-face consultation meetings. The use of briefing materials was new to the majority of clerks, although only one noted it in the questionnaire.
In the postpilot evaluation questionnaire, we also asked clerks about their opinion on the cooperation with the research team. All the clerks noted the research team's ongoing commitment, and almost all observed its competent support. Among the qualities that clerks expected from the research team, flexibility toward the design of the software took a dominant position. Two out of the nine municipalities had critical remarks regarding cooperation with the project's team. These included the imposition of the format for consultations (some elements are mandatory such as the time for participant registration, when briefing materials are made available to them) as well as 'meticulousness' in respect to subsequent parts of the pilot.
The group of municipalities changed during our project, and each stage demanded different activities. Those with the longest record of collaboration with us, as well as City 3, which was most open to learning, were most satisfied with the project. We may hypothesize that the impact of the project on their routines for the organization of public consultations, whether with the use of the inDialogue software or without it, will also be the most considerable. Four out of nine municipalities were dissatisfied with the pilot study. All belonged to towns that had joined the project late, and-judging by their answers to questions on inDialogue's functionalities-evaluated it mainly from the perspective of the number of users, and largely ignored features of deliberative public consultations.
In the relation to the research team and the project, the clerks adopted different roles. I propose their typology taking into consideration the motivation to join the project, the approach to tasks (their timely implementation, the need for external support, reflexivity), skills, as well as the openness to learning and change.
Clerks who could be characterized as engaged contractors (City 2, Town 1) fulfilled their tasks in a timely fashion out of duty. Both joined the project at the beginning, developed a good working relationship with the research team, acquainted themselves with the new knowledge, and, with some exceptions, comprehended the functioning of the software, but are hesitant to implement the model at present. Both expressed satisfaction with the project. In another local government (Town 6), this attitude to tasks, combined with the very active approach of the mayor, is promising in regards to implementation. Still, there was a false innovator, City 1, which joined the project to build its reputation as part of a public relations repertoire. It was the only city in the project that had a text forum, and although it was searching for new solutions and had a good general understanding of the deliberative model, was lacking the motivation to introduce a profound change. Among the latecomers, there was a group of participants out of necessity (Town 2, Town 3, Town 4, Town 5), that joined the project in search of knowledge and skills but communicated their disappointment with the number of tasks. It seems that they did not have enough time to understand all aspects of the software's use. Finally, City 3, which also joined the project late, could be called a hesitant challenger. The proactive approach of the team of clerks, and especially its leader; the best understanding of inDialogue's functions out of all the municipalities; the negotiation of proposals; the readiness to embrace solutions that other clerks considered risky, were all accompanied by a moderate optimism.

Conclusions
Deliberative methods of public consultations have been designed and experimented to support reflexive dialogue between politicians and citizens in the context of decisionmaking. Their institutionalization can be an important element for constructing deliberative systems (Fishkin 2018). It is particularly relevant in countries, regions, and local communities where public consultations are made in various areas of public policies and have an impact on the quality of life of local communities. The process of designing institutions, which, however, increases and deepens citizen participation in political decisions, challenges existing democratic institutions (Nabatchi & Blomgren Amsler 2014;Stromer-Galley, Webb & Muhlberger 2012).
In Poland, due to the mostly decentralized character of regulations on public consultations, local politics has a considerable impact on how public consultations are organized. By looking at Polish local governments from a broader perspective, one can observe a tendency toward merging the models of representative and participatory (public consultations) democracy. Municipalities learn from their own experience and inspire one another regarding the content of regulations for public consultations as well as new venues and channels of communication with citizens. Over the past few years, Polish local governments have also tested some methods of deliberative public consultations in collaboration with either NGOs or universities. Now, these methods are sometimes mentioned in regulations.
A closer examination of standards for public consultations reveals differences in performance between municipalities. The local governments have at their disposal unequal resources for conducting public consultations, with the involvement of methodology and tools requiring substantial investment. Studies, however, demonstrate there are procedural problems in local governments of different sizes and unequal budgets. We have identified three areas of concern: inadequacies in the methodology of public consultations, limited public access to information at the input and output ends of the process, as well as the weak internal collaboration within city and town halls.
In response to the problems in organizing public consultations, we have developed the inDialogue software to intervene in local governments' performance, and to implement norms of deliberation. In this article, I have studied the negotiation of inDialogue's design within the loop of communication with clerks, as well as its pilots in nine Polish municipalities of different sizes. The software guides an institutional user through the process of planning online and face-to-face consultations and integrates their results. It also supports communication and collaboration on the plan and briefing materials in city and town halls before deliberations with citizens. The focus of the analysis has been on the clerks' openness to the model of public consultations that inDialogue implements, and their relation with a project intervening in their routines.
In discussing choices for ICT functions and features, we were referred continuously, by clerks, to a broader system of norms, procedures, and institutions in local government in Poland. The institutional context that clerks referred to while experimenting with the use of the inDialogue software in public consultations included: (1) legal aspects at the national and local levels (rules that confuse or limit the choice of methodology), (2) a participatory turn in municipalities motivated by political gain (quantitative measures of success of the participatory processes), and (3) resources at the disposal of administration (inadequate knowledge, competences, and skills, but also time and money). In evaluating the inDialogue software, the clerks made connections with citizens (and their hypothesized preferences regarding forms of participation), with other clerks (usually to demonstrate inadequate support), and rarely with politicians, media, or NGOs.
In their feedback on the inDialogue software, the clerks were unanimously concerned about the need to include citizens. Their definition of openness, however, typically differed from the one presented in the model of deliberation. This was a political conception of popular suffrage, where the first and most important rule is that every citizen has the right to participate in public consultations. The success of the process is mostly measured by the number of replies to questionnaires, the number of votes in participatory budgeting, or the number of people attending during a public hearing. In the model of deliberation, the social composition of participants who represent the community, or that part most affected by the problem, requires the attention of public consultation organizers.
The random or targeted selection of residents could help face this problem. Still, as long as the mechanism for participants' verification, selection, or even registration, limits the number of local community members involved, they are at best treated with caution. It seems that behind this approach is apprehension about political risk, and the anticipated cost of introducing procedures that may be protested against by social activists and picked up by the media, particularly given legal ambiguities. Moreover, proposed changes in procedures result in a higher workload, so as long as they are not required, they are rather avoided (even if the better quality of processes may lead to the higher participation rate). We were surprised by the marginal concern for the quality of participation and the tools aimed at strengthening it.
The action research approach, along with the collaboration with city and town halls, allowed us to adapt at least some elements of the design for the inDialogue software to the expectations of the clerks. Facing contradictory recommendations regarding, for example, the verification of participants in public consultations, we left the choice of its level to the clerks, who would then decide what is justified in particular circumstances. We only advised the selection of some choices over others through the instructions. The pilot study's results raise doubts about whether they will be used frequently, even though clerks described some problems resulting from a lack of authentication from the beginning of the project.
We attempted to be flexible in the design and keep the entry level low where possible (Towne & Herbsleb 2012), but we also left some elements of the public consultations procedure as mandatory. This may discourage some municipalities from using the software, but the purpose has never been to replicate the existing practice. The openness to learn and use inDialogue, which can help tackle the challenge of organizing public consultations based on impartial information, inclusion, and dialogue, differed among city and town halls. The most important factor differentiating them was the motivation to improve performance by clerks learning new procedures and tools, followed by the direct support of a decision-maker. The length and depth of collaboration with researchers, as well as the repertoire of skills in communication and data analysis, which is often connected with the size of a municipality and clerks' tasks, were also crucial as they helped to understand the proposed model of action; but they were not decisive.
The ambiguous role of the inDialogue software, which, on the one hand, supports clerks in their tasks, and on the other, controls their actions, became apparent in municipalities along with the project's development. An experiment at the level of the local public sphere that involves media and social partners may be needed to test the clerks' view on the costs and gains of the adoption of democratic innovation (i.e., the model of deliberative public consultations). The feedback received, at least from some municipalities (clerks and decision-makers), has assured us that we can plan for the implementation of the inDialogue software on a broader scale, although we cannot be sure how municipalities will cope with the tasks without ongoing support from the project team.
In the process of research and reflection in collaboration with clerks on the model for public consultations implemented by the inDialogue software, we have drawn some conclusions about factors that may increase the success of systemic change in the performance of local governments. Some of them we took into account before the project started, and these were just confirmed by the new empirical material, such as the legal aspects and the organizational context in local governments that may differ considerably, not only between municipalities of various sizes, but also between municipalities of the same size; the support from an umbrella organization like the Association of Polish Cities, which connects researchers with public administration units; and combining research, action, and evaluation. New lessons include working hand in hand with clerks, and political representatives in local governments; considering the selection of local governments who have strong democratic intentions to lead the systemic change; and defining the right incentives for local governments to get involved and not to avoid challenges. The sustainability of inDialogue's use may need to include support from regional governments, as in the case of Mazovia where the software is being implemented in several dozen municipalities. based-argumentation-system/. 6 See https://decidim.org/. 7 Links to video demonstrations of inDialogue's functions are available at wdialogu.uw.edu.pl/en. 8 In 2014, we asked in the questionnaire that clerks filled in Q1: 'Please, list the most important problems regarding the organization of consultations with residents in your city.' Clerks from municipalities new to the project were asked in the interview in 2017 Q1(IV): 'What is the biggest challenge for the implementing of consultations with residents?' 9 The questions on the questionnaire that clerks filled-in in 2017 included, among others, the following questions Q1: Does the inDialogue software help in the organization of consultations with residents? Q2: What are the most significant advantages of organizing consultations with residents using the inDialogue software? Please, mention one to three positive features of the software and briefly describe how it helps in organizing consultations with residents. Q3: What are the most significant drawbacks of the inDialogue software in the context of its use in planning and implementing consultations with residents. Please mention one to three negative features of the software and briefly describe how they adversely affect the organization of consultations with residents. 10 I have anonymized cities and towns participating in the study by replacing their names with unique codes.