The problems of plagiarisms, data falsification, and research-related biases are well documented (e.g., Wicherts et al., 2016; Cook et al., 2018). However, replication studies that can combat such academic misconduct are scarcely practiced (Camerer et al., 2016; Mueller-Langer et al., 2019). Researchers (Fisher, 1935; Popper, 1959; Sackett, 1979; Dickersin, 1990) have already raised concerns regarding reproducibility several decades ago. However, the sheer number of publications being produced today made the problem as painful as ever. Other related issues are also under broad discussion. For example, the pressure to publish in academic employment and funding acquisitions (Nosek et al., 2012) and the reluctance to share the scientific data (Stodden, 2015). The practices of OA (Piwowar et al., 2018), Open Peer Review (Wolfram et al., 2020), Open Data and Source (King, 2011; Pasquetto et al., 2015), and OS Notebook (Bradley, 2016) have been developed to address this lack of reproducibility and openness. The academic narrative on OS also includes discussions on the ethical aspects of scientific research (Franzoni and Sauermann, 2014). A promising, new development addressing the ethical considerations is the FAIR standards referring to the need for data to be Findable, Accessible, Interoperable, and Reusable. Kramer and Bosman (2016) outlined that openness practices already made a positive influence on the research process and circulation of scientific knowledge.

Since 2006, European Commission invested heavily in the development of required infrastructures at the European level (European Commission, 2021b). Some notable examples include a series of OpenAIRE, Open Research Europe, and European OS Cloud projects serving as pilots for mainstreaming the openness infrastructure (European Commission, 2021b). In addition to the infrastructure, European Commission used the Framework programs (FP) for integrating openness elements into Research and Innovation (R&I) landscape. A gradual change in the way OA and data were approached in FP can be seen. Under the seventh FP (2007–2013) a pilot on OA was launched and in the eighth FP Horizon 2020 (2014–2020) OA to all peer-reviewed scientific publications became mandatory. The new FP (Horizon Europe) demands immediate OA, including both the right to read and reuse the materials (European Commission, 2021a). The outcomes of these policies are already apparent. According to the study “Monitoring the open access policy of Horizon 2020” conducted in 2021, “estimated level of compliance to the open access mandate for scientific publications under Horizon 2020 stood at 83%” (European Commission, 2021b, p. 10).

The OS movement extends beyond the EU. For the past decade various international bodies debated the OA policies (e.g., 2002 Budapest OA Initiative, 2016 Amsterdam Call for Action on OS and global initiatives) and developed the key principles in data and information sharing (e.g., OECD Principles and Guidelines for Access to Research Data from Public Funding). In 2016, the African OS Platform was launched (African OS Platform Strategy Workshop, 2018). Major open research data initiatives are underway in Australia, Canada, and China (UNESCO, 2021).

Despite positive changes in openness infrastructure, problematic areas of OS application prevail. For example, predatory publishers exploiting OA requirements (McCann and Polacsek, 2018), a limited number of high-quality OA journals, and the academic promotion and evaluation system, where OA is not of major importance. Research shows that researchers seem to be in favor of the general concepts of OA and open data (Ross-Hellauer et al., 2017) but are hesitant to publish OA and rank OA availability as low when considering where to publish (Blankstein and Wolff-Eisenberg, 2019).

Over the past few decades, academic research focused on organizational openness with implications of greater transparency and inclusion (Whittington et al., 2011; Hautz et al., 2017). Such relationships share one underlying albeit opaque principle of co-creation. In general, co-creation refers to the bi-directional, interactive development of new knowledge together with a diverse group of stakeholders. In the research and innovation context, the views seem to also be shifting toward opening the processes to citizens and other stakeholders. For example, the EU FP fosters open collaborations with non-academic partners. Gagliardi et al. (2016) and Greenhalgh et al. (2016) described participative approaches in science as Mode 2 learning where knowledge is created with those who are likely to use it, within the context of its use, and where boundaries between users and producers of knowledge are blurred. The research already includes qualitative (D'Este et al., 2018) and quantitative (Mascarenhas et al., 2018; Sjöö and Hellström, 2019) evidence based on the societal and the economic benefits of open collaborations.

There is a large literature on co-creative initiatives in science, for example, transdisciplinary research (OECD, 2020), university-driven interactions (D'este and Perkmann, 2011), citizen science (MacSweeney et al., 2019), and Triple Helix relations between universities, industry, and government (Etzkowitz and Leydesdorff, 2000). Currently, one of the most discussed open collaborative practices is citizen science. Here, the public participation is achieved through (1) crowdsourcing initiatives mobilizing voluntary contributions by non-researchers such as iNaturalist or Zooniverse and (2) initiatives aimed at citizen intervention and empowerment in the course of research itself, such as the definition of research problems or project coordination (MacSweeney et al., 2019). Citizen science approaches are growing among many research disciplines, most notably in environmental science, climate change, health, and biomedical research. Some applications are already apparent in social sciences and humanities (Tauginiene et al., 2020).

Despite the benefits of co-created research, it can face significant barriers. The academic community tends to collaborate with similar actors due to different institutional conditions (van Rijnsoever and Hessels, 2021). Co-creative initiatives require significant resources and continuous investments in project management, processes, and staff to overcome organizational and knowledge differences (Pinho et al., 2014; Whitmore and Mills, 2021). What complicates the work with non-academic partners even more is the lack of required skills in academia (European Commission, 2021a). However, training and guidance opportunities for researchers willing to engage with actors unfamiliar with scientific routines are limited (European Commission, 2021a). The academic literature is more concerned with collaborations with industry (European Commission, 2021a). In general, research on co-creation through OS initiatives focuses on conceptualization rather than on deconstructing the collaboration dynamics between researchers, civil society, government, and industry (Stier and Smit, 2021). Aguinis et al. (2020) used motivation theory to explain why there is a gap between the knowledge and practical application of OS and concluded that the perceived costs of openness are currently higher than the benefits. This suggests that a systemic change is needed for OS to gain traction.

The traditional innovation theories provide a linear view of one-directional flows from science to the commercial application (Arnkil et al., 2010). Recent academic thought increasingly acknowledges that complex knowledge is needed in addressing social and environmental challenges. Such knowledge cannot be generated within the boundaries of a single organization (Kazadi et al., 2016). Following this paradigm change a variety of new methods and theoretical approaches were defined deconstructing collaborative practices of new knowledge creation (Vargo and Lusch, 2016; Järvi and Kortelainen, 2017). In reviewing them, the notion of ecosystem revealed its importance with different qualifiers such as innovation ecosystem (e.g., Adner, 2006; de Vasconcelos Gomes et al., 2018), social innovation ecosystems (e.g., Domanski et al., 2020; Terstriep et al., 2020), open innovation (Chesbrough, 2003), ecosystems of shared value (Kramer and Pfitzer, 2016). Although functional purposes may vary, the concepts increasingly overlap and share certain inherent features. Mostly because collaborative approaches assume that within complex systems there is more capacity for new knowledge generation than as an individual. In most cases, the actors enhance their capacities by acting together (Jütting, 2020). What makes the concept of innovation ecosystems distinct from other forms of collaboration is the value of co-creation for all stakeholders involved and attention not only to structural conditions of innovation (i.e., funding and infrastructure) but also to intangible and qualitative interactions (Cai et al., 2020). Thomas and Walburn (2017) underlined the need to look beyond structural capital and consider human (i.e., people, skills, networks, and knowledge) and relational (i.e., trust, confidence, and shared vision behavior) capital in innovation ecosystems.

Based on the outlined consideration, the author argues that knowledge co-creation processes in OS should also be approached through the view of the ecosystem since it embraces a much wider socio-cultural system than pure dyadic relationships between research/industry or research/civic society. Traditional conceptualizations of OS focus on policies, infrastructures, and funding that support the openness paradigm similarly to other contexts of innovation. Hence, by broadening attention to the ecosystem more intangible and qualitative aspects affecting knowledge co-creation can be isolated. Such an approach provides a more holistic understanding including both the objective (i.e., quality improvement measures) and subjective goals (i.e., the social duty of science) of OS. Hence, the proposed conceptual framework (see Figure 1) suggests that the co-creative capacity is determined by a complex set of interactions grouped along two dimensions: framework conditions and ecosystem conditions.

The framework conditions focus on structural factors that are amendable through policy interventions. Structural factors provide an enabling environment for actors to engage in co-creation activities. According to Remøe et al. (2015), there is no such thing as generally optimal framework conditions because of varying conditions in different countries. However, certain characteristics can be seen as conducive to innovation. This includes policies and governance favoring OS approaches, the commitment of formal institutions and decision-makers, infrastructure for openness in terms of tools, spaces, and training available for actors willing to participate in collaborative research activities, consistent funding for openness initiatives, and socio-economic and cultural aspects in analyzed context. Implicitly, this means that the capacity to co-create depends on a wider economic and institutional environment.

The ecosystem conditions refer to the dynamics (linkages and networks) of the co-creation process. Although the OS researchers agree on the importance of co-creation as a new type of organizing, how to design them for co-creation to happen is researched to a much lesser extent. This is in part because the concept of value co-creation itself is elusive (Grönroos and Voima, 2013). However, some research is moving in this direction and will be discussed in the following section. Ketonen-Oksi and Valkokari (2019) looked at innovation ecosystems as structures for value co-creation and identified the key principles based on empirical findings: ensuring a clear vision and a shared value base on which the ecosystem activities can be built, facilitation for actors to make new connections and share their knowledge, and diversity among ecosystem actors. Pera et al. (2016) found that the enablers of multi-stakeholder value co-creation are trust, openness, and inclusiveness. Kramer and Pfitzer (2016) analyzed the ecosystems of shared value and defined five elements leading to collective impact and social change: common agenda, shared measurement system, mutually reinforcing activities, constant communication, and dedicated “backbone” support from one or more independent organizations.

Other complexity-based ecosystem studies focused only on particular elements of the co-creative process such as communication, heterogeneity, and roles of different actors. Ruoslahti (2018) argues that multi-stakeholder communication is a key process in networked activities which leads to the reduction of knowledge gaps and complexity. The heterogeneity of actors involved in ecosystems is increasingly recognized in collaborative innovation and refers to a wide variety of Quadruple Helix partners (Corsaro et al., 2012). However, few studies identify the exact number and diversity of stakeholders required (Reypens et al., 2016). Corsaro et al. (2012) based on previous literature identified six attributes of actors' heterogeneity that influence the development of collaborative innovation: goals, knowledge bases, capabilities and competencies, perceptions, power and position, and culture. This shows the importance of capacity evaluation of different stakeholder groups (i.e., can and how they can participate in ecosystem processes). Terstriep et al. (2015) distinguished four categories of actors involved in social innovation processes: developers, promoters, supporters, and knowledge providers. Terstriep et al. (2015) suggested that actors may fill a number of these roles and they are subject to change over time. There is also extensive literature on innovation intermediaries providing support for collaboration between two or more actors and bridging gaps in knowledge, competency, and capability (Edler and Yeow, 2016). Universities seem to play an essential role in innovation ecosystems as knowledge integrators (Cai et al., 2020; Tolstykh et al., 2021).

Some research already looked into the possible solutions for decreasing the fragmented implementation of OS. For example, Tabarés Gutiérrez et al. (2020) after analysis of an EU-funded project identified five pillars of openness implementation: (1) contextualization (institutional self-understanding that takes into consideration the structures, rules, and values of the target organization and institutional field); (2) ecosystem approach (institutional embeddedness, network relationships, and inter-dependencies of the target organization); (3) organizational theory (theoretically and empirically grounded framework for the organizational change); (4) metrics and indicators (qualitative and quantitative impact assessment); and (5) communication, culture, and trust (open communication to build and maintain trust). Dobers and Stier (2018) have listed recommendations for organizations that work with quadruple helix collaboration and co-creation in social sciences and humanities fields. The research has divided them into four categories: management, involvement, communication, and tools and spaces. Regeer and Bunders (2009) looked into interactions between science and society and defined success factors for interactive approaches at four levels: system-level (network is adaptive and provides learning opportunities), institutional level (organizational embedding, support for co-operation, possibility of changes in the process, stimuli, financing, and adequate funding criteria), project level (project leader/core and process management), and participants (committed to shared objective, open, listening, and skills for joint learning). Similarly, Ruoslahti (2018) looked at the co-creation of knowledge in EU-funded innovation projects based on the elements of complexity and emphasized self-organization, connectivity and interdependence, co-evolution, and the creation of new order. However, a more cohesive and holistic framework of evaluation has not been offered by the researchers.

Although the definitions of factors shaping co-creation processes in complex multi-agent ecosystems are varied, key properties can still be derived into the conceptual framework from the above discussion, i.e., heterogeneity in terms of actors involved (belonging to different social and technological networks), shared vision (to base the ecosystem activities), support system (institutional and non-institutional), feedback and measurement mechanisms (for continuous self-regulation), consistent and dynamic communication (between the actors of the ecosystem), and intermediaries facilitating the processes (of making new connections and resource integration).