Collective intelligence (CI) is a phenomenon studied in multiple disciplines such as decision-making, economics, internet science, computer and web science, sociology, political science, psychology, and even biology [1
]. CI is defined as a form of universal, distributed intelligence, which arises from the collaboration and competition of many individuals [2
], or the general ability of a group to perform a wide variety of tasks [3
]. The idea of CI entails the utilisation of interactions in communities to create solutions, concepts, and ideas; selecting between the created alternatives, refining, implementation, evaluation, and gaining feedback. All this is based on collective work and cooperation, so the studies on CI mainly focus on the question: How groups of people collaborate together in a smart way. The most promising examples of recent CI projects are combining humans and machines, organisations, and networks [4
]. The concept of CI is therefore closely connected with the development of Web 2.0 and the interactions between users in online communities to create “added value” based on collective work and cooperation. It enables synthesis of knowledge, visualisation of arguments, open innovation, a reduction in the number of errors, coopetition and collective decision making. Collective intelligence frequently manifests itself when collaboration, competition, or reciprocal observation gives rise to totally new solutions to the problems that a community faces or leads to an increase in the capability to solve complex issues. Empirical studies and theoretic simulations prove that a diverse collective can, under certain conditions, achieve better results in problem solving than a narrow group of experts [5
]. Furthermore, contemporary cognitive psychology shows that by analysing the way of thinking of a single person, we discover that this person’s reasoning is largely based on the use of community of knowledge resources, and not just that person’s mind. Either external resources of clustered knowledge available from experts in a given field or collective work on solutions by means of deliberation are used [8
However, it is not difficult to notice that crowds, especially those gathered in online communities, can be both smart and inventive while also being irrational, biased, or overconfident. For this reason, studies on CI attempt to answer the question—in what situations and in relation to which issues can we create conditions conducive to emerging group intelligence [3
]. Previous studies have also shown that collective work and cooperation is not only an effective method of work, but also a certain organisational culture that releases the potential of individual community members and creates relationships of interests and ties bonding the group [1
The issue of CI in the public sphere, which we deal with in this article, concerns the kind of civic debate on public issues that helps in solving the problems that exist in this sphere. The public sphere, as evidenced by numerous studies [13
], cannot function well in atomised, broken, distrustful, and insecure societies—societies where social capital is lacking. Social capital is a term specific to social sciences that describes intergroup ties. A high quality of social capital can mean mutual trust, reciprocity, and effective cooperation, which translates into social stability and a sense of community of interests that is necessary to raise the level of debate on public matters [15
The research described below is focused on the possibilities of emerging CI in response to the social challenges specific to Central and Eastern European communities. This region of the world is currently affected by several social problems—both the new ones, resulting from the contemporary geopolitical conditions: The political crisis in Ukraine and the resulting migration, as well as old ones, derived from the legacy of post-communism: Low trust in political institutions and the loosening of social bonds, weakened norms of consensus and agreement, the low quality of public debate [16
], and as many authors believe—the problematic quality of social capital [17
For this reason, the study of the possible impact of internet-based CI initiatives on the public sphere in this region, specifically the possibility of creating new communication and cooperation platforms, engaging people with a low level of social activity, is a matter of vital importance. Both Poland and Ukraine, as a result of the experience of communism and the uncertainty associated with the political transformation implemented since the 1990s, struggled with a serious crisis of involvement in social life. Limited social activity was, to a certain degree, caused by the earlier compulsory aspect of membership in formal associations, additionally influenced by significant disappointment with the consequences of the transformation. This situation was more visible in Ukraine [18
], but also observed in Poland [17
]. However, the dynamic socio-political situation of recent years has created new conditions. Ukraine has experienced anti-government protests (“Euromaidan”) and social turmoil related to the war in Crimea and Donbas. Poland has also found itself in a new social situation due to high immigration (especially from Ukraine) and emigration (toWestern European countries), which in a short period has led to the creation of a multi-ethnic society.
This new reality also creates opportunities for the development of a new type of social capital. Due to the increasing coexistence of people of different ethnicities and lifestyles, new types of relationships and new levels of cooperation are created; people interact with each other, interdependently building their relationships, identities, and common interests. The increase in cooperation opportunities for diverse groups living side by side can therefore lead to an increase in their bridging social capital [20
]. A new participative model of social life based on internet projects, with a relatively low entry barrier, space for creativity, and the widespread use of ICT technologies, can provide the new ways of debating civic engagement, and collective action. This possibility becomes particularly valuable if we consider the erosion of the traditional public sphere, as observed by Jürgen Habermas [23
]. In order to prepare for the implementation of this type of projects, and to determine whether they will meet the CI paradigm, we need the most appropriate methods for tracking the various features of CI in existing and future Polish-Ukrainian online initiatives. We need methods of monitoring the factors conditioning group collaboration, and the circumstances of their occurrence, as well as the methods of sensing participants’ incentives and attitudes. This need has been expressed in this study through the following hypothesis and research questions:
H: The existing CI research methods can help to identify the specific aspects of debate in online communities consisting of Poles and members of the Ukrainian minority in Poland, which will contribute to create social bridging capital between these groups.
RQ 1: What kind of model of civic debate will be suitable for use in Polish-Ukrainian internet projects?
RQ 2: How the existing CI research methods can be used to monitor particular aspects of the civic debate in online projects that contribute to building social bridging capital between Polish and Ukrainian communities?
The methods used to conduct the study included: In the first stage, a multidisciplinary literature review; in the second stage qualitative in-depth interviews (IDIs). The literature review covered the widely present topics in scientific literature that we considered particularly important to our research. These topics are: Debate models existing in the public sphere, methods of researching CI compatible with these debate models and suitable in identifying their features in online projects, bridging social capital, and factors affecting its level in internet projects. As a technique of literature review, following H. Snyder guidelines [24
], we selected an integrative review that allowed us to critique and synthesise knowledge and was intended to create a theoretical framework to be used in later qualitative research [25
The authors focused on the results published in the well-known data bases (Web of Science, Scopus, Science Direct, and Google Scholar). A cursory analysis was performed by reading through the titles and abstracts of each topic. The following criteria were used to select the articles for review: (1) the article discusses the selected topic; (2) the publication in which the article is published is listed in at least one of the above mentioned databases; and (3) the article makes a non-trivial contribution to the debate, meaning it involves more than a couple of references to the term. As for the subject of civic debate models and measuring deliberation, we also based our research on literature review performed by Steenbergen et al. [26
], Stromer-Galley [27
], Mouffe [28
], and Paxton [29
The analysis of the articles selected in that way allowed us to assess, to critically review, and to potentially reconceptualise the literature on the research topics to enable a new conceptual theoretical framework to emerge [24
]. The review allowed: (1) determining the most important models of debate in the public sphere and their aspects, especially features important for conducting online debates; (2) selection of such methods of researching collective intelligence in internet projects that would be adequate to identify specific aspects of online debate; and (3) definition of such criteria for increasing the level of social bridging capital that would be possible, thanks to the implementation of internet projects that meet previously adopted assumptions. This work led us to prepare a theoretical framework that summarises the methods of researching CI projects and the corresponding properties of debate in the public sphere, adopting an additional criterion of the expected result, which is the impact on the increase in the level of bridging capital in the studied projects.
The second stage of our work was to subject the theoretical model to qualitative research. The sampling for the personal interviews was conducted through e-mails and telephone calls. As a result of the sampling, 12 respondents met our inclusion criteria. We conducted 12 semi-structured interviews using the IDIs technique. In order to perform an in-depth analysis, face-to face interview was chosen as the main method. This method enabled us to gain a wide range of information resulting in evaluation of the whole contextual environment. The selection of the participants for this study ensured the best combination of experience and broad knowledge of Polish-Ukrainian relations, especially the situation regarding the Ukrainian minority in Poland. All of the interviewees were specialists in Polish-Ukrainian projects in NGOs, businesses, universities, and public organisations. Their work experience included past implementation of international projects involving Polish and Ukrainian people, working in mixed Polish-Ukrainian teams, and researching issues connected with this kind of cooperation. All participants were informed that anonymity of volunteers and their organisations would be guaranteed. The interviews lasted one hour on average, depending on the participant’s responses. The interview consisted of open-ended questions and the procedure was semi-structured. The participants were asked seven questions in total, each containing some sub-questions. Pilot tests were implemented with two researchers to ensure the clarity and relevance of the questions. The interviews were recorded, transcribed, and coded. The scope of the interview included the following issues:
What are the key problems related to the situation of the Ukrainian minority in Poland, its integration with the rest of society and the level of social capital?
What kind of online communication is suitable for debating common Polish-Ukrainian issues? Which of the debate models presented in the theoretical framework will be particularly relevant to Polish-Ukrainian internet projects?
Which of the presented CI research methods seem to be the most suitable for studying Polish-Ukrainian internet projects and why?
Which of the debate aspects examined by these methods are particularly relevant in raising the level of social bridging capital in Polish-Ukrainian communities?
Based on both the literature review and the opinions of the participants in our study, we conclude that the presented CI research methods and debate models can significantly help to develop internet communities that will create bridging capital between Poles and Ukrainians. The presented models of civic debate, both the deliberative model and the agonistic model are applicable to our case. The deliberative model, focused on consensus-building and rational argumentation, happen to be good at mitigating the contentious and emotional themes, increasing group cohesion, and increasing interest in the common good reduced to a universally acceptable outcome of the debate. This model is especially relevant for the debate on issues of local communities or labor law. On the other hand, the agonistic model can also be used, because preserving group identities and supporting competition draws the debate out of apathy, and even becomes a method of realising self-interest. Depending on the topic for debate, both consensual agreement (specific for the deliberative model) and preference ranking (specific for the agonistic model) can be a useful result in an online CI project.
In the existing literature, the issue of the public debate model in CI projects has been treated in one way only. Collective intelligence has been implicitly combined with the deliberative democracy model, supplemented by the cognitive diversity criterion [34
]. What is more, researchers on this topic emphasise the need for further work due to insufficient knowledge of the actual relationship between the deliberation process and the development of valuable results in internet projects [72
]. The conclusions of our research meet this need. Noting the possibility of two alternative debate models in CI projects, that have been identified in social sciences (deliberative and agonistic model), we open the possibility of further work that should be focused on verifying the adequacy of these models to specific public issues implemented in CI projects and the possibility of combining some of the features of these models in the various projects. The impact of cooperation and competition factors on collective intelligence processes has been noticed in some previous work [74
], but it has never been thoroughly analysed in relation to online debates on public affairs. The most interesting attempt in this direction was the recent work of V. Lowndes and M. Paxton presented in the text: Can agonism be institutionalised? Can institutions be agonised? Prospects for democratic design [38
], where practical possibilities of applying the agonistic model in the form of preference rankings in public online projects were pointed out. However, there is still a need for a dedicated, comparative study in this field.
The review of the selected CI research methods revealed that each of them is seen as potentially useful, although they can help identify different aspects of CI. So far, the results associated with the use of each of these methods provide examples of their effectiveness in relation to the specific features of online debates. Calculating the metrics of online communication and using it to extract issues, ideas, and arguments, as well as detecting users’ relations, decisions, and meaningful patterns, was assessed by J. Capella et al. as one of the most appropriate factors affecting intelligent decision-making in public deliberation [72
]. The outstanding examples of practical use of this method for monitoring the level of collective intelligence have been carried out by M. Klein and his collaborators. Studies conducted with the use of Deliberatorium software have included: Large-scale argumentation on the use of biofuels in Italy, conducted in the University of Naples, debate on the controversial questions about possible changes to Italy’s election laws [40
], and Intel conducted deliberation on the possible use of “open computing” [11
]. The results of these projects showed how software algorithms can help users allocate their effort in a large-scale debate, as well as how debate metrics can be used to generate personalised attention mediation suggestions [41
], although it is sometimes difficult for users to switch from the conversational attitude to formalised argument-based discussion. When used properly, argument mapping helps to organise and summarise information, but as with other formal knowledge representations, it does not work well for mediating dialog, nor does it support the movement from less structured exploratory dialogue to more structured knowledge representations [77
]. It was also confirmed in our research, that this method, on the one hand, is attractive because of its objectivity, but on the other is the most “impersonal”: The calculated metrics may not always show where people’s behavior comes from.
CI studies conducted so far with the use of a composite index, were primarily based on the CIPI that was developed by Skaržauskienė et al. [53
]. The major research project which used this method was “Social Technologies for Development Collective Intelligence in Networked Society”, where 15 online communities were studied. The application of this methodology primarily allowed for evaluation of the creativity, problem solving level, and civic empowerment in these communities. During the study, however, problems arose, due to the large number of accepted sub-indexes and monitored factors, it turned out that it was not possible to collect uniform data everywhere [53
]. Our research results pointed out, that assessment using indexes and scales with qualitative criteria allow for the performance of more flexible study than using purely numerical metrics. At the same time, however, it turned out, that an overall factor may not be enough to explain what is really happening and what we can do to increase group intelligence. Some elements of the index are more and others less applicable to the study of specific communities. In our opinion, therefore, individual sub-indexes should be selected to study specific aspects of community performance in CI projects.
The last examined method, that is creating qualitative CI assessments with the help of a CI genome framework, was used in several research projects, mainly by T. Malone et al. [6
] and Wise at al. [54
]. In these projects, respectively, approximately 250 online platforms [6
] and over 120 plaforms were studied. This work led to identification of different kind of CI “genes”, the conditions under which they are useful, and the constraints governing how they can be combined [6
]. The used framework, although certainly not showing the detailed behavior of online community members and not describing all the possibilities, allows an overall picture to be formed. Our study refreshed the possibility of using this method, by applying it to specific types of communities. Our proposal to include the contestation gene in the study of Polish-Ukrainian CI projects was appreciated by the study participants. Therefore, carrying out this type of research in the future would be a contribution to the development of the framework used in this method.
The conclusions of our study accommodates the hypothesis that the use of a combination of the described research methods in relation to online Polish-Ukrainian projects can bring valuable results. The optimum solution for future research would be to combine the examined methods to mutually complement their characteristics. The proposal to conduct a study using multiple approaches to capture various aspects of the phenomenon being studied is in our opinion the best approach to the study of collective intelligence in complex social situations, as is the case in Polish-Ukrainian relations. This proposal goes beyond the projects implemented so far. Each of these methods, as analysed in our research, was found suitable to examine certain aspects of communication in these kind of online projects. Especially combining the quantitative methods, processing numerical data from the server, with qualitative assessments clarifying the incentives, interests, and emotions related with communication processes, seems to be a promising approach. On the other hand, using composite indexes to valuate diverse aspects of CI in a graded scale of points may allow monitoring the real online communities, regardless of the laboratory experiments carried out.
The presented debate models and CI testing methods may also significantly contribute to an increase in bridging capital in online projects. This may occur especially by building mutual trust, which is a necessary component of social capital: By gaining knowledge about the intensity and density of networks; reciprocity; fulfilling community norms, as well as common CI genes (including “contestation” gene). Studies conducted so far prove that internet communication under certain conditions can have a positive impact on social capital [62
]. Relations between emerging collective intelligence and an increase of social capital have been indicated in several publications [53
], but have not been the subject of separate research in recent years. That is why we encourage researchers from various disciplines to take up this topic in their future work. Additionally, an issue that our participants pointed out and which should be analysed in more detail, is filtering out users intentionally distorting communication (internet trolls), which shouldhelp to maintain the community standards. This can be tracked by the analytics server, composite index method or by qualitative assessments.
The results of our work, both the literature review and qualitative studies, open up the possibility of conducting further research that would allow the validation of their accuracy using empirical methods. First, experiments should verify which of the debate models, as well as the debate topics will actually be appropriate for specific Polish-Ukrainian online communities: What kind of community features and the issues raised in the debate will work better with the deliberative model, based on consensus and dispute mitigation, and when will the agonistic model, based on competition and allowing group identities and emotions, be more appropriate? This can be tested by selecting several more or less controversial debate topics, recruiting groups with different user profiles, and setting the purpose of the debate either as a common, widely acceptable result, or as a preference ranking. Another possibility of further work would be evaluating the performance of existing communities to find out methods for detecting “internet trolls”, i.e., people, who purposely disrupt communication to provoke other participants. However, it should be also examined whether stigmatising such behavior leads to abuse by restricting freedom of expression and introducing censorship. Another question to be further explored is how does the choice of the language of the debate affect equal participation, communication intensity, and conflicts, and in which situations is bilingualism necessary and in which optional.
As for the presented CI research methods, the actual suitability of each of them should also be checked during empirical research. Specific issues that should be explored in the case of the calculating metrics method, relate to its ability to detect nuances associated with the actual motivations of the participants. In the case of the method based on the CI potential index, it should be particularly examined which of the existing sub-indexes have a specific impact on Polish-Ukrainian communities. Then, in the case of a CI genome study, it would be interesting to verify the existence and impact of the contestation gene on community activities. Another particularly interesting topic that appeared during the IDIs, and which should be examined during empirical research, is the question of behavioral patterns present among Poles and Ukrainians: Whether differences in “individualistic” and “collective” attitudes are related to cultural differences between people, or are actually related to age and social situation, regardless of ethnicity. We hope, therefore, that the work we have done will create the opportunity to carry out many interesting research projects in the future.