Opening science: towards an agenda of open science in academia and industry

The shift towards open innovation has substantially changed the academic and practical understanding of corporate innovation. While academic studies on open innovation are burgeoning, most research on the topic focuses on the later phases of the innovation process. So far, the impact and implications of the general tendency towards more openness in academic and industrial science at the very front-end of the innovation process have been mostly neglected. Our paper presents a conceptualization of this open science as a new research paradigm. Based on empirical data and current literature, we analyze the phenomenon and propose four perspectives of open science. Furthermore, we outline current trends and propose directions for future developments.


Introduction
For centuries, science has been based on an open process of creating and sharing knowledge. Over time however, the quantity, quality, and speed of scientific output have changed, as has the openness of science. In the days of Galileo, scientists had to use anagrams to hide from inquisition. Later, scientists used letters to distribute their knowledge amongst colleagues. When in 1665 'Philosophical Transactions' was founded, scientists began to send their insights to scientific journals. In the last century, the number of scientific journals exploded. At the same time knowledge diffusion slowed down. In some fields the peer review process takes several years from first submission to final publication (Björk and Solomon 2013). New IT-based submission and paper tracking platforms hardly improved review times, as reviewers remain to be the bottleneck. Today, more and more academic institutions open up science by employing open access journals, sharing research data, or including others into the research process. But also large firms like Siemens, IBM, or Tesla are part of the open science phenomenon. Instead of patenting knowledge, they publish large parts of their research in order to participate in the scientific community. In doing so they mark findings as state-of-the-art and thus prevent others from patenting them.
Despite recent trends, management scholars have mostly neglected the phenomenon of open science. That is surprising as there is a clear link between the activities in the field of open science and those in the field of open innovation. Science has the purpose of developing a knowledge domain by adding theoretical or empirical insights, whereas innovation has the purpose of developing and bringing to market new offerings such as products or services. The most widely used definition of open science stems from Nielsen (2011): ''Open science is the idea that scientific knowledge of all kinds should be openly shared as early as is practical in the discovery process.'' Open innovation by contrast is defined as ''the use of purposive inflows and outflows of knowledge to accelerate internal innovation, and expand the markets for external use of innovation, respectively. [This paradigm] assumes that firms can and should use external ideas as well as internal ideas, and internal and external paths to market, as they look to advance their technology'' (Chesbrough 2006, p. 1). Both definitions have the acceleration of a process by sharing knowledge in common. Many scientific findings are later turned into innovations. It is therefore desirable to understand the link between open science and open innovation as the definitions suggest that open science can lead to open innovation.
Current literature on open innovation is predominantly underlying a business-centric view. This view assumes a firm to be mainly motivated by profits. Numerous studies investigated how external ideas are utilized inside companies in order to develop new product offerings (Dahlander and Gann 2010). Additionally, scholars analyzed the possibilities of commercializing internally generated knowledge in form of intellectual property (IP) for profit generation outside the company boundaries (Chesbrough 2003a, b). Research is understood as an enabler to come to new knowledge (Koen et al. 2001). Yet, the very early stages of research and science have hardly been analyzed by the current open innovation literature. In other words, we currently experience a vibrant debate within scientific theory that focuses on open science (Bartling and Friesike 2014;Jong and Slavova 2014). Yet, this debate mostly neglects its interdependency to innovation.
This article provides a conventionalization of open science based on a literature review and semi-structured interviews with CTOs, research managers, open innovation directors, open access leaders, industrial researchers, and scientists. The paper is structured as follows: In the next section (Sect. 2), we review the literature on open innovation and open science to derive similarities and difference between both concepts. In Sect. 3, we describe our methodological approach before (in Sect. 4) we analyze and discuss current trends based on empirical insights. The paper concludes by providing implications and suggestions for future research (Sect. 5).

Research streams in open innovation
The failure of large industrial research labs to drive scientific advancements towards value generation in the early 1980s manifested an anomaly that changed the rules of innovation. Shortly after its foundation in 1984, Cisco started with its open R&D strategy, which ended up in outcompeting the world's largest R&D center, the AT&T's Bell labs. In the Kuhnian sense (1962), this marked a paradigm shift in innovation management. Since then, the practical and academic community called for more open models of innovation (e.g., Chesbrough 2003a, b;Christensen et al. 2005).
Within the last decade of academic research, several special issues on open innovation underpinned a fundamental change in the perception of innovation (see e.g., R&D Management 2006R&D Management , 2009R&D Management , 2010  (1) Integration of external cooperation partners along the value chain. Downstream the value chain, von Hippel's (1986) works on lead user integration highlight the virtue of user collaboration for radical innovation. Numerous studies investigated user characteristics and their impact on the degree of innovativeness, the modality of user integration, and user's motivation to collaborate (Bilgram et al. 2008;Franke et al. 2006;Luethje 2004). The phenomenon of free revealing and the fact that the user is the only external collaboration partner with use-experience makes the user a very valuable partner (Nambisan and Baron 2010;von Hippel and von Krogh 2006). Upstream the value chain, research emphasized the importance of supplier integration. The integration of suppliers into the development process at a very early stage can significantly increase innovation performance in most industries (Hagedoorn 1993). (2) Partnering and alliances. Strong specialization necessitated the need for many companies to collaborate with partner companies from the same or other industries (Hagedoorn 2002;Schildhauer 2011). Especially, the phenomena of cross industry innovation and innovating with non-suppliers was investigated (Howells 2008;Herstatt and Kalogerakis 2005). Also established engineering firms take the role of innovation intermediaries moderating open innovation activities between collaborators (Howells 2006). This indirect opening up of the innovation process is leveraging the cross-industry innovation process, not only in traditional R&D outsourcing modes but also in strategic innovation partnering. (3) Open innovation processes. Open innovation can be subdivided in three core processes: outside-in, inside-out, and coupled. This classification provides guidance on how to complement and extend the internal innovation process by an external Opening science 583 periphery (Gassmann and Enkel 2004). Most large companies such as Siemens and BASF started to developed detailed firm specific open innovation processes. In addition some companies such as Procter & Gamble and Siemens assigned process owners with special positions and titles for open innovation within their corporations. In both corporations these directors attract considerable attention within the company. (4) Open innovation tools. As a means to implement open innovation numerous tools emerged; most of them support the integration of external innovation sources (West and Lakhani 2008). Crowdsouring platforms like InnoCentive, 99design, Jovoto, Nine Sigma, or Atizo bring together solution seekers and problem solvers (Bullinger et al. 2010;Sieg et al. 2010;Dahlander et al. 2008). Thereby, they generate a virtual market place for innovative ideas and problem solutions. Toolkits for masscustomization allow an adaptation of design and product features according to customer preferences based on an iterative creation process (Piller and Walcher 2006). Community based innovation enables companies to use blogs and discussion forums to communicate with a mass of stakeholders outside the company. (5) Open trade of intellectual property. The times where IP was solely used as a means to secure the firm's freedom to operate are over. The more open approach towards IP changed its role and importance within the firm's value creation processes (Pisano 2006). The active use of IP for in-and out-licensing unfolded new business models, which are widely discussed in literature. New phenomena like patent funds, patent trolls and patent donations emerged in recent years and increasingly attracted scientific research (Reitzig et al. 2007;Ziegler et al. 2014). At the moment, there is an ongoing debate amongst policy makers in the European Union whether a financial market for IP should be created. Policy makers in favor of new modes of technology transfer as well as financial institutions interested in new product categories are predominantly driving that process. The overview on the existing research streams in the field of open innovation shows the strong application and commercialization focus of the present literature. But, detailed insights on collaboration and openness in the field of knowledge creation and science are lacking.

Perspectives of open science
In the context of academic and industrial science, the sharing and combination of information is regarded as the core process of knowledge creation (Thursby et al. 2009). As scientific problems are getting more specialized and complex at the same time, it is not surprising that collaboration in science and research expanded in various disciplines within the last decades. For example in sociology, the percentage of co-authored articles almost quintupled in the last 70 years (Hunter and Leahey 2008). Comparable trends were observed in political science (Fisher et al. 1998), physics (Braun et al. 1992), and economics (Maske et al. 2003). Studies even show that authors with a high h-index are those who collaborate widely with others, form strong alliances, and are less likely to be bonded to a certain in-group (Pike 2010;Tacke 2010). According to Merton (1973), the principle of openness has always been an integral part within the academic community. This openness is rooted in a reward system that the first person to contribute new findings to the scientific community receives in return various forms of recognition (Stephan 1996;McCain 1991;Hagström 1965 Here, open has to be seen in contrast to the prior status quo (e.g., a publication was available only to the subscriber of a journal and only after it was published) and not in contrast to 'closed science'. Contradictory, industrial scientists were perceived as being much more concerned about confidentiality as a means to secure future returns on R&D investments (Cohen et al. 2000). Recent studies however indicate that this disparity seems to diminish as increasingly cross-institutional bonds emerged (Murray 2006;Powell et al. 2005). For example, Haeussler (2011) found that for both academic and industrial scientists the likelihood of collaboration and exchange depends on the competitive value of the requested information and on the degree to which the researcher's community conforms to the 'norm of open science' (Rhoten and Powell 2007). Differences between academic and industrial research become blurred (Vallas and Kleinman 2008). Thus, academic and industrial science moved from a ''binary system of public vs. proprietary science to […] arrangements which combine elements of both'' (Rhoten and Powell 2007, p. 346). The convergence of academic and industrial science and the increasing importance of collaboration and openness drive the need to gain more insight in how open science is characterized.
Actors in open science include institutions such as universities and corporations as well as individual researchers. From a value chain perspective, open science includes the very front-end activities of basic science, applied science, and applied research. Despite the contextual backgrounds of academia and industry, research is rather driven by curiosity, reputation, and acknowledgement than by profit and applied oriented thinking. We differentiate four perspectives of open science: Opening science 585 (1) Philanthropic perspective. Doing research requires infrastructural and content-related elements whose access has been predominantly restricted. Current trends foster a democratization of science and research in the sense of distributing scientific content, tools, and infrastructures freely. Many universities started to offer public lectures or courses with the goal in mind to bring science and research closer to society and to market scientific findings. Most of the public lectures are streamed online and thus are globally available (Tacke 2010). Additionally, this trend includes the rise of open access journals that provide users with the non-restricted right to read, download, copy, distribute, print, search, or link to the full texts of articles. As most traditional journals generate revenues based on subscriptions, the majority of open access journals are funded by the authors through publication fees. Within the last years, the visibility and prominence of open access journals significantly increased due to the growing numbers and the establishment of the Directory of Open Access Journals.
(2) Reflationary perspective. Currently, we witness a trend towards making scientific results freely available during pre-publication. Knowledge is shared in a very early stage within the research process. Motives to do so are manifold. Researchers are able to reflect first thoughts, to promulgate preliminary scientific results, and to promote new ideas within the scientific community. Thereby, they signal tacit knowledge and reputation that might attract other researchers and institutions (Hicks 1995). Furthermore, they are capable of actively influencing future research directions and starting new scientific discussions. Colleagues and amateurs are invited to give feedback and to join in for collaborative knowledge creation.
External involvement diminishes problems with respect to local search bias and groupthink many closed scientific research teams suffer from. At the same time the journals and publishers have a self-interest in pre-publications: Papers which have been published before being printed will get cited more often and therefore increase the citation impact and thus attractiveness of the journal. Moreover, the memory and transparency of the World Wide Web allows tracing thoughts and knowledge creation. This minimizes the risk of lost authorship. Comments and evaluations of peers might give guidance in research phases of high uncertainty. (3) Constructivistic perspective. The opening of science and research enables new collaborative forms of knowledge creation. This knowledge creation does not only bring new knowledge into being but also new opportunities for new user models and new businesses. Crowdsourcing is one prominent example: Problem seekers pull for new scientific solutions by broadcasting problems to an unknown mass of potential problem solvers. Virtual rooms are used as an exchange platform where problem seekers and solvers can interact. Small groups form virtual exchange platforms for loose or moderated exchange with the goal of knowledge creation. Open platforms typically address several fields in a more interdisciplinary manner than the typical disciplinary mainstream journals. The integration of more than one scientific discipline under one roof fosters cross-fertilization of researchers and scientists. This interdisciplinary approach enhances technology fusions and the generation of innovative solutions (Kodama 1992). (4) Exploitative perspective. Most researchers are oriented towards the generation of novel scientific findings neglecting real life application. The active sharing and promoting of scientific knowledge enables researchers to close this gap towards application-oriented knowledge exploitation faster. In cooperation with practitioners a common shared construction of new artifacts based on the latest scientific findings is possible.
The Following table provides an overview on open science initiatives with respect to the four just discussed perspectives (Table 1).

Methodology
Given the young nature of the phenomenon, our empirical research mainly relies on a qualitative exploratory research approach based on interviews, Internet research, and document analysis. This triangulated qualitative approach is an appropriate means to navigate unclear boundaries between phenomenon and context in the early stages of research. Our data generally relies on the primary source of semi-structured expert interviews and secondary source of company press releases and Internet research. Between 2008 and 2011, we conducted 38 interviews with different actors in technology intensive industry and academic research, namely CTOs, R&D managers, open innovation directors, senior industry and academic researchers, directors of research institutes, editors and referees of academic journals, and university presidents. This kind of triangulation allows us to minimize the bias of personal perspective and enhance the validity of the information.
At the outset of our interviewee sampling, we scanned our personal contacts, websites of research institutions, research databases, and the public press to identify experts in the field of open science that were most promising in revealing new insights. Our goal was to generate a heterogeneous sample that allowed us to analyze open science from various different viewpoints and validate our results. After each interview, we asked our respondents to name further colleagues who may reveal more insights applying a snowball sampling. No further interviews were conducted when we achieved theoretical saturation and the interviewees did not reveal new empirical insights.
The framework described in chapter 2.3 guided our data acquisition and data analysis. Based on the framework we developed an interview guideline, which we adapted according to new empirical insights that emerged during the research process. Thus, our field data was collected and analyzed iteratively based on reflection of scientific literature. This led to alternations between inductive and deductive procedures (Eisenhardt 1989). To combine the advantages of unstructured and semi-structured interview methods, we started with open-ended questions, followed by a structured questionnaire protocol. Besides asking formal questions regarding the institutions' motivations and barriers of opening up science, the interviewees were also strongly encouraged to provide related examples from their daily business, including current research projects. The intention of the interviews was to identify drivers, inhibitors and current trends in open science and research.
The data was primarily collected by personal face-to-face or telephone interviews, which lasted between 40 and 120 min on average. Each interview was transcribed. The transcripts were transferred into an excel tool to break the interviews down into single statements. Each statement was paraphrased and grouped into categories. The identified trends emerged out of the categories. Two researchers independently analyzed the data allowing cross comparisons that increased the validity of the results.
Academic research representatives of the following institutes were interviewed: In early 2013 we conducted a second field phase and carried out 22 further interviews with experts in the realm of open science. These interviews were mainly conducted to gain insights into how individual researchers deal with the challenges in open science (e.g., open access journals make publications widely available but a researcher's career is not based on availability of research results but rather on the ranking of publication outlets). We used these interviews to verify the present research results 2 years after the initial interviews. All interviews were transcribed verbatim and coded using the software NVivo. Results discussing individual factors concerning open science were published by Scheliga and Friesike (2014).

Emerging trends of open science
The open science paradigm has just paved the way towards a new division of tasks and a new role understanding within scientific research. New links and forms of collaboration emerged within the science community itself but also between academic research and more application-oriented institutions. The times when research institutes demonstrated intellectual fortresses following the goal of Humboldt's knowledge creation as an end in itself seem to be over in most areas. The complexity of scientific problems and the required investments (time, expertise, and materials) to solve them dramatically increased within the last decades and necessitated the breaking of new grounds in external collaboration (Bozeman and Corley 2004). We identified several trends, which underpin the opening of science: ( the Swiss CTI the establishment of a knowledge distribution platform in form of published results is very helpful in order to receive grants. In most EU funded projects the open distribution of knowledge is already a prerequisite to get funding. (2) Role of research institutes: from ivory towers to knowledge brokers. Traditionally, there was a gap between research driven universities and application driven private companies. This gap is diminishing, as the distribution of tasks between academia and industry changes. The tremendous rise of technology transfer fostered by many universities and private companies further closes the gap between science and practice. For instance, the ETH Zurich and IBM jointly operate the Binnig and Rohrer Nanotechnology Center in Zurich. The center provides a common collaboration platform for researchers of both institutions. As equal collaboration partners, both institutions have the right to publish and to commercialize the jointly created IP. This dual relationship increases the pressure on both partners to timely find applications for the scientific findings generated and to commercialize research results. The local consolidation of many highly dedicated innovation teams proofed to accelerate knowledge creation and opens up fast ways for the commercialization of current results. Additionally, mutual career paths in the ETH and IBM emerged that manifest a liaison management between both entities and create spill-over effects especially with respect to the transfer of tacit knowledge. In recent years, the self-conception of many universities and research institutes changed. Many public institutes moved from being a provider of basic research, towards more application centric research. To enable multiplication on a global scale and to merge competences, various research institutes formed networks that aim at providing direct solutions for business problems. The Auto-ID Labs are a prominent example. They represent a leading global network of academic research laboratories in the field of networked RFID. The labs consist of seven renowned research institutesincluding the MIT Lab, the ETH Zurich Lab, the Cambridge Lab, the Fudan Lab, and the Keio Lab-located on four different continents. The goal of the Auto-ID Labs is to architect the 'Internet of things' and to provide an efficient infrastructure, which facilitates new business models and applications on the basis of the RFID technology.
(3) Outsourcing research: from make to buy. The industrial trend of reducing the value chain activities to focus on core competencies has also affected the relationship between private, application oriented businesses and research institutes. Following this trend, many companies cut their expenses in corporate basic research. As a consequence, numerous firms started to outsource research activities: The elevator company Schindler works together with the Institute of Applied Mathematics at the University of Cologne. On the basis of precise requirements, Schindler outsourced the development of genetic algorithms for its latest elevator control systems. In this regard, the Institute of Applied Mathematics became a knowledge and technology supplier at the very front end of Schindler's innovation process. Daimler outsourced many of the telematics research to several research institutes and universities. ABB outsourced its research on inspection robotics for their installations to a joint venture with the ETH Zurich. SAP has set up several decentralized research labs on campuses of universities, e.g., TU Darmstadt, ETH Zurich, and St.Gallen. Novartis is more and more relying on start-up firms and research institutions to fill the technology pipeline in research and preclinical development. Additionally, the outsourcing of research activities offers SMEs new possibilities to overcome the 'liabilities of smallness' (Gassmann and Keupp 2007). Earlier, due to resource constrains, many SMEs were not able to conduct basic research on their own. Thus, outsourcing scientific problems to research institutions allows them to increase their competitive position. (4) Financing of research: from single-source to multiple-source funding. In recent years, the increasing cost pressure on many public households, led to declining budgets in many public research institutions. Formerly being largely financed by public money, many universities are forced to find additional ways for financing research activities and thus progressively seek third party financing. Various universities increased their activities in technology transfer and in IP commercialization. For example, 25 Bavarian academic institutes formed a patent exploitation network, which is coordinated by a patent bureau. Under the roof of the Fraunhofer Institute, BayernPatent is responsible for the IP commercialization. Additionally, it assists inventors with the filing of patents. Thereby, it works closely with local patent attorneys and offices. BayernPatent covers 100 % of the patent filing and maintenance costs and thus minimizes the risk for the academic institutes. Revenues are split equally between the inventor (25 %), the faculty (25 %), the university (25 %), and BayernPatent (25 %). Whereas many universities moved from public to more private funding, numerous corporations made an opposite shift. In the 80's roughly 80 % of Siemens' Corporate Technology was financed by uncommitted corporate funds. Today, more than 70 % have to be financed by Corporate Technology on its own responsibility via third party money or business units. This trend is also reflected by several other large firms such as ABB, Daimler, and Philips, which are forced to collaborate with universities and spend seed money in basic research. This supports the universities in their research. (5) Research culture: from closed disciplinary to open interdisciplinary thinking. For decades, science was predominantly driven by disciplinary research. Within the scientific community, research streams were influenced by few dedicated and topic specific journals. A narrow and disciplinary framing of research articles increased the probability of getting accepted. Additionally, the dogma of 'publish or perish' forced researchers to keep their work secret-at least in the early stage of competition-until submitting it to a scientific journal. This development led to scientific progress but also to disciplinary silos. Within the last decades, the number of interdisciplinary journals grew constantly and new forms of Internet-based collaborations emerged. Offering new ways of publication and collaboration, this development caused a change in thinking towards more open and interdisciplinary research. New scientific cross-links between various research fields offer novel platforms for publication. With this the entire scientific landscape is in constant flux, new research areas emerge and others vanish. (6) Focus of research: from broad universities to specific institutes. Within the scientific landscape the specialization of research institutes in the public but also private sector increased. The requirement to be more cost efficient forced research activities to be closer related to the core competencies-responsible for value creation and profit generation-of the executing institutes. Within the public sector, specialized researches institutes were formed that attract new researchers as well as private companies. Numerous examples can be found in the sector of environmental technologies in Germany. In the private sector, companies deliberately invest in basic research in strategic fields of major importance. For example, Sulzer Innotec became a specialist for computational fluid dynamics. Later, their know-how in simulation software was applied in the development of a wide variety of products. An open science research system does not only improve academic research but holds tremendous potential for industrial applications. Here, we need to learn more on how this interaction can be supported and managed. At large, current scientific contributions are still fragmented and are far away from presenting a holistic picture of open science. Many knowledge gaps within various fields are evident. Clear policy implications are needed to address the issue of individual incentives for scientists. The promising idea of open science comes with a multitude of possible challenges, which need to be addressed in order to make this research paradigm work. The field of open science is still at an early stage and offers a wide field for future research. We invite researchers to contribute to this fascinating area and to help answering the many remaining questions.

Limitations
Given our research design it is helpful to point out several limitations. Firstly, as typical for qualitative research, our study describes a present phenomenon. But we are unable to quantify it. We are unable to provide insights on how 'open' research actually is and how many researchers are engaged in open science. Secondly, this study provides a general overview on this rather young occurrence. We do not compare disciplines or highlight promoting and hindering factors that would explain why certain disciplines are engaging in open science more vividly than others. Thirdly, we do not look at individual drivers. The idea of open science and its key argument (making research processes as transparent and open as possible) is good for society as a whole; yet it does not take into consideration the individual researcher. In many cases what is good for society is not in the best interest of the researcher himself. Career incentives and institutional policies might hinder open science. Fourthly, our data emphasizes the German speaking research and business community. To minimize local bias and ensure global validity, we interviewed managers and researchers with an international background. Practically, we selected researchers that previously published in international journals and manager that work in a global business environment to ensure that the findings are valid beyond the German-speaking world.
Open Access This article is distributed under the terms of the Creative Commons Attribution License which permits any use, distribution, and reproduction in any medium, provided the original author(s) and the source are credited.