A tale of two ' opens ' : intersections between Free and Open Source Software and Open Scholarship

3


Introduction 4
Understanding notions of 'openness' in software and scholarship 5 A brief history of the Free Software and Open Source movements 6 The interwoven histories of Open Access and Open Scholarship 8 A resurgence in 'open research practices' 10 Contested understandings of Open Scholarship (and 'Open Science') 12 What aspects can be transferred between FOSS and Open Scholarship 18 FOSS and Open Scholarship as communities of practice around collaboration and sharing 18 Integrating FOSS practices into scholarly workflows 24 Integrating FOSS principles and values into scholarly communities 26 The role of commercial players 29

Systems of valuation in openness 31
What elements of scholarship are currently most-valued? 31 Introduction "Information wants to be free.

" -Stewart Brand
It is the best of times, it is the worst of times, it is the age of wisdom, it is the age of foolishness, it is the epoch of belief, it is the epoch of incredulity, it is the season of Light, it is the season of Darkness, it is the spring of hope, it is the winter of despair, we have everything before us, we have nothing before us. Departing from the words of Dickens, in short, when it comes to the state of scholarship, we are at a crossroad where existing and imagined infrastructures overlap, compete and displace one another. At the heart of that struggle lies our evaluation, valuation, use and operationalisation of "openness" in scholarship: what it was, what it is, what it can be. To each of those three, multiple legitimate answers exist. That multiplicity is the topic of the present article.
Since the 1980s, some of the key developments in academia have been based around Free and Open Source Software (FOSS) and Open Scholarship, as part of a wider and pluralistic Open Society movement (Şentürk 2001;David 2008). FOSS has been highly successful in a number of ways in society, and there is much that could be learned from this and applied more broadly to academia, particularly in context of an ongoing renewed interest in Open Scholarship. Contrasting FOSS and Open Scholarship together will provide reciprocal insight into these highly dynamic concepts and their relative successes or acceptance to date (Willinsky 2005;; Grubb and Easterbrook 2011;Levin and Leonelli 2017;Lahti et al. 2017).
Specifically, in this article we will address four key themes in the four sections below: • First, we will discuss notions of 'openness' in FOSS and Open Scholarship, and how these constitute and concern one another. This section describes the history of FOSS over the last several decades, and the resurgence in 'open research practices' within modern academia.
We additionally explore the contested understandings of Open Scholarship or Open Science, and the inherent complexities this creates with respect to FOSS.
• Second, we ask which aspects of FOSS and Open Scholarship can be transferred between each other. This section primarily deals with three issues: first, shaping communities of practice between FOSS and Open Scholarship; second, the fundamental principles and values behind both, such as 'gift-giving'; and third, questions about the roles of commercial players in each space.
• In the third section, we discuss what we can learn about systems of reward, value, and reputation by comparing the two 'opens'. Many problems within modern scholarship can be traced towards a defective system of reputation accrual and reward. What can Open Scholarship learn from FOSS in this regard?
• In the fourth section we ask how 'openness' intersects with issues around reproducibility and data sharing. Many fields of modern research are undergoing what is often termed a 'reproducibility crisis'. This section will look into how FOSS and open data intersect with Open Scholarship to address this critical issue.
Understanding notions of 'openness' in software and scholarship A large intersection exists between the principles and philosophies that underly FOSS and Open Scholarship. Due to their historical overlap, we can investigate the co-production that generated both in their current multitude shapes. We recognise movements of convergence as well as divergence in their shared history. For example, software and data accessibility (e.g., availability, inspection, and re-use) are now considered prerequisite for making full research environments transparent and rigorous enough to reproduce results (Ghosh et al. 2012;Sandve et al. 2013;Millman and Perez 2014;Hocquet and Wieber 2018). Many modern data collection and analysis workflows, including preparation and dissemination of publications, rely on a range of essential software or Web-based tools for which powerful FOSS solutions already exist (e.g., R , Python , Julia , GNU PSPP ). The combined ability to independently reproduce and refute research is one of the critical elements for most of the scientific enterprise (Popper 1959). Openness in research software and data promises to help close the digital divide and enhance the democratic globalisation of research participation (May 2006;Fuchs and Horak 2008;Packer 2009), and are critical to healthier socio-cultural and political discourse (Merton 1942). As the foundations behind FOSS tend to be more thoroughly studied and understood, understanding its history can help to elucidate modern concepts of scholarship and their future co-evolution (Bretthauer 2001;Weber 2004) (Figure 1). for such infrastructures (leading from Multics to Unix and the C language, in the late 1960s and early 1970s), created an environment in which source code for academic computing could be shared: repetitive work was left to machine and know-how could be quickly passed over. However, the Free Software movement itself was launched in 1983 (see the original announcement by Richard Stallman here ), followed by the launch of the GNU project based around the free operating system GNU (1984); most importantly this represented the first FOSS organized manifesto. The purpose here was to produce a novel software system by sharing the general code knowledge, which was until then primarily covered by intellectual property. The idea of intellectual property rights was incompatible alluded to in the widely known slogan from Stallman, "'free' as in 'free speech,' not as in 'free beer.'" This is equivalent to free as in 'libre' rather than free as in 'gratis', in many languages; or 'to be free' versus 'to be for free'. The acronyms FOSS or FLOSS (free/libre open source software) are able to encompass all aspects of the Open Source and Free Software movements. Now, Open Source, without the moral load associated with the word "free", is the standard term used to refer to this class of software, and is now ubiquitous in the global computing landscape -including within scholarly research.

The interwoven histories of Open Access and Open Scholarship
Scholarship, scholarly identities, and the politics and process of scholarly communication have a rich, non-linear, inconsistent, region-and discipline-specific history (Csiszar 2016;. The typically institutionalised forms that dominate modern scholarship, came about as part of a shift towards increasing 'citizen science' through the 19th century (Shapin 2011). This was followed by the explosion of global scholarship in the post-Second World War years, leading to increased growth of the higher education sector, diversification of academic roles and career paths, increasing international participation in research, and increasing dissemination of academic outputs through scholarly journals and books (Fyfe et al. 2017). Together, these factors led to an intricately evolving sociopolitical relationship between the 'public' and scholarship (Merton 1973;Geiger 2017;Neylon 2018). Scholarly practices of sharing precede the origins of scholarly journals, based around early forms of scholarly communication, such as witnessing experiments, and later, the exchange of letters between scientists, as the first form of so-called 'virtual witnessing' (Shapin 1984). These early forms of sharing were highly exclusive and catered only to a small and local elite. Over the course of the centuries following the scientific revolution, the 'public' character of science changed. This was primarily due to a shift in who is considered to be included in that 'public', informed by an evolution of the 'social contract of science' that increasingly placed science inside society, as opposed to esoterically hovering above it (Gibbons 1999). A consequence of this was that many more people became legitimate members of the public of science, whose voices were considered worthy of participation and inclusion (Gibbons et al. 1994).
Perhaps the biggest innovation in the history of scholarly research was that of the printed journal in the 17 th Century, which became the primary vehicle for communicating research and defining increasingly specialised scholarly communities (Potts et al. 2017;Hartley et al. 2019). This extended the physical college into a virtual college (Sztompka 2007;Csiszar 2018), coinciding with emerging scientific practices around working collaboratively to advance knowledge as a common good for society, rather than for personal or institutional gain (David 2004;Lahti et al. 2017 primarily focuses on providing unrestricted access to research articles (Guédon 2008;Suber 2012;Tennant et al. 2016) and helping to increase the global/public circulation of research information beyond professional scientists (Adcock and Fottrell 2008;Laakso et al. 2011;Bacevic and Muellerleile 2018). OA undoubtedly has its origins as a grassroots, community-led initiative as an alternative or complement to typically subscription-based proprietary publishing ventures (Willinsky 2005). Now, it has become widely adopted at various policy levels by universities, governments, and funding bodies; often aligned with the impact or 'excellence' agendas of these institutes (Finch et al. 2013;Vessuri, Guédon, and Cetto 2014;Moore et al. 2017;Katz, Allen, et al. 2018).
The early history of the Internet involved a number of academics based at either private or public research institutes (e.g., Tim Berners-Lee, Grace Hopper, Elizabeth Jake Feinler, Sally Floyd, Lawrence Lessig, and Linus Torvalds), suggesting that the ideals that underpin much of FOSS have their origins from within scholarship itself. Others suggest that "Open Science" and the idea of scientific knowledge as a common good has its origins as far back as the late sixteenth and early seventeenth centuries (David 2004;Shapin 2011;Martin 2019). While it is difficult to identify whether Open Source or Open Scholarship came 'first' due to their dynamic histories, both draw from the notions of common goods (and specifically digital common goods that are non-rivalrous) (Ostrom et al. 2002;Frischmann, Madison, and Strandburg 2014), and unrestricted participation and sharing based on the underlying values of equity and sustainability, among others. Compared to FOSS, Open Scholarship is practically more complex (Mirowski 2018); the former only relates to software development (including infrastructure, governance, legal and funding issues), whereas the latter regards the entire process of scholarship, including research articles, grant proposals, data, software, educational materials and methods, and research evaluation (Katz, McInnes, et al. 2018;Tennant, Beamer, et al. 2019); of which sharing the outputs from is now widely implemented and even mandated within many areas of scholarship (Vincent-Lamarre et al. 2016).
Many of the more-recent Web-oriented developments remain in direct tension with traditional methods of scholarly publishing. Here, commodification and privatisation of research has become the de facto norm for much of the industry, which relied on exclusionary business models to support itself. Because of this, some actors in the private sector, as well as a number of professional academic societies, have even taken active stances to subvert, co-opt, or even stop progress towards more inclusive and open systems (Posada and Chen 2018;Tennant 2018a). The hyper-competitive reward system only seems to have exacerbated this, instead promoting secrecy and individualism rather than any form of collective collaboration for knowledge generation (Merton 1968 activities of the academy, impacting its core missions of teaching and research, as well as the systems and processes that are critical to individual and institutional success." (Corrall and Pinfield 2014, p.293).
In principle, OA is highly distinct from traditional, subscription-based publication business models.
Subscription-based research articles have a strange duality in which they are published, but are largely inaccessible and restricted to the public due to financial, copyright, and licensing constraints.
Although the ideas behind OA certainly pre-date the advent of the Web, the potential of digital technologies and their associated workflows, relatively low cost and increased efficiency of copying, transmitting, and storing documents, catalysed the widespread development of OA. These all flew in the face of the subscription model, which, even in digital environments, relied on resource-based constraints by creating the illusion of scarcity in production and dissemination. Fundamental in the origins of OA were the unconditional free public availability for reading, as well as unconstrained re-use so long as original sources were attributed; often equated with the Creative Commons Attribution (CC-BY) license (Suber 2007b;Tennant et al. 2016 Accompanying OA, a range of 'open research practices' are now commonly called upon as potential methodological improvements to major issues around reproducibility, publication bias, questionable research practices, and more efficient or rigorous research workflows (Watson 2015;Levin et al. 2016;Crick, Hall, and Ishtiaq 2017;Masuzzo and Martens 2017;McKiernan 2017;Bowman and Keene 2018;Fraser et al. 2018). This includes a diverse range of practices such as pre-registration and registered reports (Nosek and Lakens 2014;Nosek et al. 2018), sharing of code and data (Barnes 2010;Levin 2015;Mons 2018), and opening up the peer review process in different ways (Morey et al. 2016;Ross-Hellauer 2017;Tennant et al. 2017). As research has become progressively computational, demands for FOSS have simultaneously increased as part of the complete scholarship environment. However, we must also be careful not to attribute too much power to openness.
Sociologist of science Harry Collins already demonstrated that sharing everything that can be shared is still insufficient to produce, for instance, replicable research (Obels et al. 2019). There is a lot to science that cannot be made explicit and thereby cannot be shared on paper or digitally; known as tacit knowledge and tacit expertise (Collins 1992;Derksen and Rietzschel 2013;Penders, Holbrook, and de Rijcke 2019).
Regional variations in uptake of open research practices are noteworthy. In Latin America (Arza, Fressoli, and Lopez 2017), "Open Access" (aka Ciencia Abierta/Acceso Abierto) seems to have become more widely adopted with the advent of SciELO in 1997 (Packer 2009) and Redalyc in 2002 (López et al. 2008). Other regions, including western Europe and North America, have a more diverse and fragmented history of making OA the norm, with correspondingly diverse economic and social tensions (Barić et al. 2017 Science is concerned with the details of how to lower or erase the technical, social, and cultural barriers to sharing" . This socio-cultural philosophy echoed the work of (Şentürk 2001), and was reflected in later works including Michael A. Peters' review of existing Open Science approaches (Peters 2010), and the now often-cited "Five Schools of Thought" of Open Science by (Fecher and Friesike 2014) . Concepts of 'freedom', akin to 'Free Software', have been rarely 2 discussed in the context of scholarly communication (Alperin et al. 2017).
During the last few years, Open Science has often come to be understood in a relatively narrow sense; see Lahti et al. (2017) for an exception. Earlier notions of Open Science emerged as an "umbrella term that encompasses almost any dispute about the future of knowledge creation and dissemination" (Fecher and Friesike 2014 suggest that open science is perhaps nested within the FOSS movement, as opposed to being subsidiary to it as implied by FOSTER. Irrespective of these ontological differences, the general consensus appears to be that Open Science is a commitment towards specific practices that lead to the public sharing of research outputs, and that this is a good thing leading to better science. Open Science then is about the process of knowledge production and then how that knowledge is shared; something overlapping with, and fundamental to, FOSS. Although not explicitly stated, the core concept around such practices creating more openly inclusive research environments aligns itself with the philosophies of earlier advocates of Open Science (Şentürk 2001;Peters and Roberts 2015).
However, even socio-cultural terms such as 'liberty' are still used in their practical, rather than philosophical, sense (Frankenhuis and Nettle 2018); and the wider humanist and political sentiments that underpinned early concepts of Open Science (e.g., based around fundamental freedoms), and indeed Free Software, appear to be almost absent.
Taking these described tendencies as a line of further evolution, more and more scholars To compound this taxonomic complexity even further, 'openness' in scholarship is often used at a number of different scales; for example, lab groups or teams, individuals, and institutions (Tennant, Beamer, et al. 2019). Thus, 'Open Scholarship' is often applied indiscriminately to mean very different things, but with the same blanket terminology applied. Linked to the diachronic shift in the meaning of Open Science away from the more philosophical underpinnings of 'openness' in academia as discussed by, for example, Peters and Roberts (2015), and towards a narrower and more pragmatic frame of reference, the widespread and often-synonymous usage of Open Science "openwashing" -derived from "greenwashing", where vendors rush to label their products "green" in environmental terms in order to increase sales -refers to practices considered deceptive insofar as they purport to be open but do so only to make themselves more attractive and in reality do not adhere to a majority of the principles of openness (Watters 2014;Weller 2014 There are two main competing approaches towards this lack of consensus. First, given the diversity of principles, practices, and outputs involved in Open Scholarship, a single, unified, comprehensive and widely-accepted consensus definition is probably not sufficient (or even desirable); unless such a definition readily embraces this diversity (e.g., as the Open Scholarship Initiative seems to do).
Second, there remains a need to rigorously define and enforce the philosophy, values, and principles of Open Scholarship, and explore how these underpin the practices, and to have consensus reached on this within the scholarly community. This would address the lack of common understanding, which has impeded the widespread adoption of the strategic direction and goals behind Open Scholarship, prevented it from becoming a true social 'movement', and separated researchers into disintegrated groups with differing, and often contested, definitions and levels of adoption of openness (Tennant, Beamer, et al. 2019). Rebecca Willen has also identified that there might be two, perhaps three, different sub-movements that intersect in different ways, involving 'open science', 'replicable science', and 'justice-oriented science' . A potential consequence of having multiple trend towards 'better' research overall. Instead, Open Scholarship, Open Research, and Open Science might best be thought of as overlapping/intersecting 'boundary objects' (Moore 2017) that represent this inherent diversity. The position of FOSS within any such schema depends on whether one regards it purely in a practical sense, or in a more fundamental and philosophical manner. Such might be preferential to avoid compounding these ontological issues further with legal and technical aspects associated with software, data, and other scholarly outputs. As Şentürk (2001) originally prescribed, the pluralism in concepts of 'openness' in scholarship (and/or science) might be where its power lies, in that different communities can understand and adopt it in different ways depending on their inherent social norms and research processes.

What aspects can be transferred between FOSS and
Open Scholarship FOSS and Open Scholarship as communities of practice around collaboration and sharing Communities form an essential part of FOSS (Katz, McInnes, et al. 2018). Here, software development and re-use is inherently communal and ideologically democratic based on its core methodologies, and self-organised either hierarchically or modularly. Version control systems allow developers to independently make changes to a copy of the code and share the changes with collaborators. Remarkable momentum towards more open collaboration models has been powered by Git (and GitHub). This modern, distributed version control system allows decentralized development of large-scale FOSS projects. It was initially developed by Linus Torvalds to facilitate Linux kernel development and has subsequently been widely adopted by the FOSS community. While harder to learn and use than earlier popular systems, such as SVN and CVS, Git similarly enables efficient collaborative workflows, and also adds additional support for decentralized development.
This process allows developers to copy the initial code repository, work independently on a 'fork' of it, and then share the changes or the full changed version on platforms such as GitHub, GitLab, Bitbucket, or Sourceforge. It has proved to be a highly effective participation system for software, leading to the rapid popularisation of such practices and development of FOSS projects.
Part of the success of FOSS was that the methods and products were proven viable alternatives, practically and commercially, to proprietary software. Another major part is that, in theory, anyone who shows an interest in a project is able to collaborate and contribute code, write documentation, add examples or tutorials, discuss, and provide feedback. This means that, even in expert-driven and software-intensive projects, there is a relatively low barrier to entry (particularly when using Web interfaces to the version control system) even for those with non-coding skills, as there is substantial value in, and appreciation for, generalist skills in documentation and communications. Increased participation also helps to divide up the labour required for larger projects, opening up new working environments and group formation and governance structures based around a shared sense of collective participation and authority (O'Mahony and Ferraro 2007;Palazzi et al. 2019). FOSS can also be thought of as a powerful method for social, community-driven or peer-to-peer dynamic learning, where the distinction between users, testers (or reviewers), and developers is more blurred than in a proprietary environment.
Collaboration in FOSS projects is essentially continuous and iterative as a progressive form of verification. The underlying assumptions for this type of process are very much idealised: public disclosure of bugs and issues that get resolved sequentially, usually by a core team plus external volunteers, until a piece of software is released, with the 'principle of many eyes' helping to increase the validity and verifiability of work. A consequence of this is a further assumption that this process also results in better software (Kelty 2001); or an alternative viewpoint that perhaps only 'better' software is worth being shared under a FOSS license (Pauliuk et al. 2015). Research projects are typically distinct from this, with 'releases' coming primarily in the form of research articles at discrete moments in time. These articles and their content are largely immutable, save for post-publication processes such as retraction or applying a corrigendum. 'Testing' as a form of quality control is usually conducted through peer review, which remains a highly-regarded yet curiously understood process at the heart of scholarship (Tennant and Ross-Hellauer 2019). These practices are slowly and tentatively beginning to change; for example, with preprint reviews in some disciplines like Meta-Psychology.
Much of this sharing and collaborative FOSS culture was driven by reciprocal motivations around sharing and repurposing, also sometimes called 'hacking' (Orr 2006;Coleman and Golub 2008).
When applied specifically to software, hacker culture leads naturally to FOSS as a community of practice, as the freedom to re-use and modify (i.e., 'hack') is built into the software through licensing requirements. Raymond (2001, p. 31) puts hackers at the heart of a gift-giving culture: "You become a hacker when other hackers call you a hacker. A "hacker" in this light is somebody who has shown (by contributing gifts) that he or she has technical ability and understands how the reputation game works. This judgment is mostly one of awareness and acculturation, and can only be delivered by those already well inside the culture." This 'gift-giving' culture strongly emphasizes the ideology that underpins the FOSS community, that sharing is intrinsically valued as part of wider hacker culture and in which reciprocity is key . It is essential that scholars understand how to apply these successful 4 collaborative practices from FOSS development, especially now that much scholarly research is becoming increasingly reliant on computation. This has the potential to improve the efficiency, diversity, productivity, and reliability of research processes, decoupled from any formalised journal-bound peer review system (Heller, The, and Bartling 2014;Frassl et al. 2018;Tennant, Bielczyk, et al. 2019) Here, however, it is important to clarify some important differences between participating in FOSS and Open Scholarship activities. These differences lie in the process more than the product. For products, granted the adoption of similar licenses (e.g., those that are Open Definition compliant), there is no substantial difference in the 'openness' for OA, Open Data, and FOSS; although in some jurisdiction, like the US, data cannot be subject to copyright. Users are typically able to access the products without restriction, and, according to the license, able to derive works from and redistribute these products. There are, however, residual issues in that just because something is 'open' does not make it accessible and immune from other inequities and ethical concerns that might influence its access and usability (Nicol et al. 2019). On the process side, it is the way that products are accessible for collaboration during development that differs. With FOSS, it is typical that a small, core group of people starts openly developing a software product; however, this depends on the scale of the project, and it can range from individual, networked, to large corporate or governmental ventures. Other people can participate in (and leave) the effort over time, contributing as they wish. Development iterates over multiple cycles and releases, and the product is available from the very beginning of the process. This method is very much idealised, and best practices can include aspects such as using open repositories, version control, and issue trackers; but such collaborative practices are neither strictly inherent to FOSS, nor are they limited to it. The production of knowledge goods that is based on sharing both resources and outputs, performed by loosely connected individuals who collaborate in a decentralized, modular and often non-hierarchical way has been identified as a new way of organizing production. Yochai Benkler coined the term "commons-based peer production" for this mode of development and identifies projects such as Wikipedia and OpenStreetMap as examples (Benkler 2006) . Through this lens, it becomes clear that the use of open licenses might be a necessary condition for the success of peer production endeavours such as FOSS, Wikipedia and OpenStreetMap, but it is not sufficient. Rather, a sustainable community of contributors is required for such efforts.
On the contrary, the process for Open Scholarship activities is mostly closed -as with more itself is more exposed, for example by using electronic laboratory notebooks (Myers et al. 1996), or registered reports (Nosek and Lakens 2014). These are still not as accessible to participation or collaboration as a typical FOSS workflow, but provide more of a window into the process of the research itself (Obels et al. 2019  Leaving the classic self-determination theory of intrinsic and extrinsic motivation aside, other studies have also shown how social identity and altruism might play a role in engaging with Open Scholarship, but also that tool support is important. An exploratory study of self-archiving practice and inhibitors among computer scientists (Graziotin 2014) found that authors often infringe the copyright transfer agreements of the publishers when self-archiving and stop self-archiving when they discover they are infringing. Lack of automation and time required were found to be the strongest factors inhibiting self-archiving of publications and sharing of datasets. This study called for tools from the FOSS community for facilitating author-based self-archiving that take authorship rights into account; we remain unaware of any such initiatives at the present, however. Similar research has identified widespread copyright infringement of scholarly articles (Green 2017;Jamali 2017), and there are a number of ongoing legal cases in this space (Chawla 2017;Manley 2019),.
Together these indicate that more awareness and support are needed for researchers regarding the legality and restrictions associated with licensing and copyright of scholarly works (Gadd, Oppenheim, and Probets 2003;Gadd and Troll Covey 2019).
Altogether, this suggests that there are complex barriers to enter an Open Scholarship mindset and to take up associated practices. Open research activities carry norms and culture that some individuals might find hard to understand even before engaging with them. This would seem to manifest itself in an apparent divergence between the attitudes towards (generally positive) and However, extrinsic and intrinsic motivations are complex and interrelated. For example, Ryan and Deci suggest that both types of motivation can be traced to basic human needs for autonomy, competence and relatedness (Ryan and Deci 2000). Such tensions between motivations and practices can manifest themselves between selective communication of knowledge, an inherent willingness to share, and the strange duality of scholarly research articles often being simultaneously public and not publicly accessible (Hilgartner 2012;Levin et al. 2016).
It is incumbent on the Open Scholarship community to develop and coordinate strategies for transitioning to OA, Open Data, and FOSS in academia that encompass social and cultural issues (Tennant, Beamer, et al. 2019 Integrating FOSS practices into scholarly workflows One key assumption embedded within open source communities and hacker culture is that the best software is the best because it works, and fulfils its intended purpose. Other aspects can also include These benefits could help to define a successful new community development model based around sharing, collaboration, and innovation, one which is already increasingly being adopted by the software industry who, among the many advantages, see it as an economic improvement. Other communities have already taken up parts of these practices as a way to improve research and help to make it more fundamentally reproducible, particularly in computational contexts (Sandve et al. 2013). These ideas are quickly spreading between academia and coding communities; for example, many software companies today are selling their expertise more than the code or the code subscription. Users can freely access the code and verify the robustness of the latter, improve it for specific cases when necessary and make sure all elements are in place for better company performances. One of the key aspects here that underlines the community is that research is a process that continuously evolves, and is therefore not captured and defined appropriately with just discrete outputs (i.e., peer reviewed journal articles). Research projects could contain numerous components and 'releases', akin to version control and FOSS, with an emphasis on the value of the process and documentation. This could create a tendency for projects to become more distributed, and interactive collaboration to be recognised as a fundamental and more widely-valued trait (Heller, The, and Bartling 2014).
Regardless of self-identified membership of the "Open Scholarship Community", one could argue that performing any (or a combination of) the multitude of practices associated within Open However, within the Free Software movement, the principles of freedom were much more deeply encoded from the outset than they appear to be for Open Scholarship or Open Science, which both appear to be more similar to Open Source in their current focus on pragmatism; with the exception of the OCSDNet and Vienna Principles. The result of this for software communities is that, by being based on essential freedoms, this helped to promote social aspects such as frictionless sharing and co-operation -factors that are increasingly important in a digital world. Often a moral basis is argued for OA around fundamental rights of access and re-use, akin to arguments made for democracy.
Similarly, a parallel argument is that Open Scholarship is a more efficient or effective way of performing and communicating research than traditional forms. Whether an increasing adoption of open practices is more due to the ideological or pragmatic arguments between software and scholarly camps remains elusive. In both, familiarity with open practices might be because they are both inherently better practices and lead to better outputs, rather than being because they align with the principles and values associated with a particular movement (i.e., Free Software).
Although FOSS and Open Scholarship might have similar foundations (i.e., around ideas of free sharing and common goods), the differences in the pace of technological and cultural changes between the two mean that 'openness' seems to have made more impact in the software world than in research. There are a number of key differences here between software and research that might explain this distinction in uptake rates between open practices, and the translation of values and principles into practices: Furthermore, adopting such practices allows individuals and communities to automatically adhere to the coupled principles and values of openness, and thus become included as part of the Open Scholarship community by default, irrespective of whether or not one openly identifies as such. If one considers science and research to be self-correcting, which often is assumed, then openness is necessary for others to repeat and verify, and thus, enable correcting to function at a system-wide scale (Ioannidis 2012). Open Scholarship as a process can potentially become relatively decentralised, distributed, and a peer-to-peer process of production, based on strong scientific principles and rigorous technological infrastructure; rather than a competition for spurious measures of success.
Thus, it appears there is a substantial place for refining the core values of Open Scholarship/Science, in particular around collaboration and participation; or perhaps even refining two separate 'camps' again based around different concepts. For example, Open Science already appears to be well understood to be more equivalent to Open Source in its pragmatic, process and output focus ( Principles seem to provide a strong basis from which to work on this, and to begin generating a more coherent discourse around open scholarly practices, challenging the issues with the present incentive system, and creating a fairer academic advancement system. Both the FOSS and scholarly communities are vast and heterogeneous, and there is no singular adherence to any particular set of values or organisation; and yet, the FOSS community has achieved great things. Ultimately, this foundation could be refined to be a more effective narrative based on the inherent motivators for different stakeholder groups; for example, how openness can catalyse technological innovation and economic growth (for industry and policymakers), or how openness can challenge unethical conduct while increasing the reliability of research through a more effective and collaborative creation method. In time, we would likely see such more effective communication, based around a combination of principles and practices, close the 'apathy gap' and resistance towards openness through appealing to a sense of common interests among stakeholder groups.

The role of commercial players
As we have discussed above, commercial players play an important role in both FOSS and Open Scholarship. In the latter, we have seen how they contribute to both our understanding and practices of openness, and maintaining the present system of evaluation and reputation. However, many of the commercial aspects of the wider scholarly communication ecosystem remain in direct conflict with any notions of openness. As a key example, traditionally many scholarly publishers operated around a subscription-based business model that has created an oligopolistic 'market' structure (Bergstrom et al. 2014;Armstrong 2015;Larivière, Haustein, and Mongeon 2015;Shu et al. 2018;Tennant and Brembs 2018), particularly based on explicitly anti-competitive practices such as non-substitutability, micro-monopolies, and non-disclosure clauses. This is clearly in violation of fundamental freedoms associated with openness, as many of these vendors actively prohibit sharing through mechanisms such as copyright and creating the illusion of artificial scarcity in order to preserve revenues. Such rights are mostly signed away by authors in exchange for publication, and for remuneration of these primary investors and value creators to be effectively nil, despite scholarly publishing being one of the most profitable industries that exists. A general lack of transparency has meant that the commercial elements of the system are able to maintain great power and control over the flows of knowledge and finances in scholarly communication (Björk 2016;Björk et al. 2018).
This tension has been one of the driving forces behind the entire modern push for Open Scholarship, Springer Nature (via Digital Science) and service providers like Clarivate Analytics. A major criticism of the latter is that the services they offer around data and analytics represent their ownership of key elements of scholarly infrastructure that create lock-ins for users and their workflows (Posada and Chen 2018;Tennant and Brembs 2018;Campfens 2019). The result of this has been that, despite the best intentions of Open Scholarship, much of the infrastructure it relies on has now been captured by commercial entities and subject to market control as a system of organised competitiveness (Odlyzko 2013;Anderson and Squires 2017). The scholarly publication process now primarily exists to reward and regulate a reputation economy (Hyland 2015;Lariviere and Sugimoto 2018), and thus poses one of the greatest single threats to modern scholarship (Brembs, Button, and Munafò 2013;Brembs 2018).

Similar to FOSS, commercial absence represents a key problem that key elements of open scholarly
infrastructure do not receive adequate sustainable funding and commitment for essential projects and services (Chang, Mills, and Newhouse 2007;Chengalur-Smith, Sidorova, and Daniel 2010), something with which Open Educational Resources also struggles (Wiley 2007). Most recently, in the Open Scholarship space, this struggle has been illustrated with discussions around the sustainability of communities and arXiv-like platforms based around the Center for Open Science. Initiatives like SCOSS, while invaluable, are currently far too small-scale to support the number of projects that require longer-term financial support.
Often it is the case that many of the people engaged in FOSS and maintaining its infrastructure operate from within large commercial organisations, rather than being committed to such projects full-time. Within the wider realm of scholarship, we know that there is 'enough money within the system' to support an almost total 'flip' to a more open system, especially regarding OA (Björk and Solomon 2014;Schimmer, Geschuhn, and Vogler 2015). However, this does not appear to have manifested in a way that is disruptive, or affordable to many researchers and institutes around the world (Barić et al. 2017;Green 2017;, and may now even be creating a second financial crisis, akin to the serials crisis, through hyperinflation of charges for OA and a lack of price sensitivity (Khoo 2019). Furthermore, this highlights the problem that such forms of 'closedness' are often more difficult to see because they are more about violating principles than any specific practices, with the unintended consequence that 'openness' can actually lead to new forms of enclosure (Tkacz 2012;Mirowski 2018). It might be the case that substantial intervention is now required from public administration bodies in parallel with the mobilisation and strategic coordination of research communities around these issues, seeing as this involves the transfer of billions of euros/dollars of public funds each year to commercial entities (Di Cosmo 2006;Tennant, Beamer, et al. 2019).

Systems of valuation in academia, and the role of open practices
In the preceding sections, we touched upon the issues of reward, evaluation, and assessment systems in modern research cultures. We also discussed how open research practices tend to lend themselves towards 'better' science, in that the research is more rigorous, accountable, and easier to build upon. This section will address the problem of how to make these open research practices aligned with the things that are valued (i.e., incentives) in order to make them more widely-accepted and adopted behaviours and practices. This presents a major challenge for academics (and academies) in how they can adapt and evolve into a system that is defined more by fundamentals of Open Scholarship (e.g., re-use, sharing, and collaboration). Indeed, such strategic implications around openness are only now just beginning to be explored within some communities (e.g., AI) (Bostrom 2017), and represent major challenges in defining 'value' in [open] scholarly research.
What elements of scholarship are currently most-valued?
At the present, few of the things which we might use to define 'better' scholarship (whether 'open' or not) are what we seem to value in terms of how research is evaluated or rewarded Schimanski and Alperin 2018;McKiernan et al. 2019). The former concept is based around things such as sharing of knowledge, ideas, and outputs, rigorous and reproducible research, and transparency; yet, current systems of reward rarely value these things, and are instead largely based on proxies for 'prestige' or 'quality', or arbitrarily based on things that can be measured, without considering the inherent value in what those things actually measure. At present, the primary reward system for professional advancement in academia in many STEM fields seems to be focused on peer-reviewed articles published in reputable scholarly journals and conference proceedings, author position/order, and perceived value of a publication venue. Current evidence suggests that citation-based metrics, including the h-index and journal impact factor (Bornmann and Daniel 2007; Lariviere and Sugimoto 2018), drive this system of evaluation (at least in North America and parts of Europe), along with factors such as the journal rank or brand that a particular piece of research is published in. This complex system of value transfer is often generalised under concepts of 'research impact' or 'excellence' (Vessuri, Guédon, and Cetto 2014;Moore et al. 2017).
What is often mistaken in this context is that, while such journals and other measures of success or reputation might be of comparative use (simply because they can be measured), this does not actually make them a useful or objective measure of reputation or prestige itself. What we know is that these current proxy measures of research quality, credibility or legitimacy not only perform poorly, but in fact often seem to correlate better to the opposite (Brembs, Button, and Munafò 2013;Brembs 2018). Journals, as the primary vehicle of communication for scholarly research in many disciplines, play a prominent role in this, forming the basis of the reward system, the distribution of prestige among communities around journals, where the attainment of having a piece of research published in such a peer reviewed venue is deemed to be of higher value than the inherent value or role of that piece of research for society. This intersection between journals as both vehicles of communication and of prestige lies at the heart of many problems in modern scholarly research.
At present, many elements of the peer review system are routinely exploited and (curiously) poorly understood (Tennant and Ross-Hellauer 2019). Peer review is often anonymous, unrewarded, and yet in high demand, and often in the control of commercial, third-party entities (Fyfe et al. 2017). This is further exacerbated by the fact that peer review, in its current journal-coupled implementation, is considered to be a necessity for virtually all forms of scholarship, irrespective of the diversity of processes that lead up to it, and it has been commonly criticised (Tennant et al. 2017;Ross-Hellauer 2017). Chief among all of these criticisms, are the fundamental issues that it remains virtually unknown how peer review operates as a quality control mechanism, whether it actually performs to the expectations with which it is regarded, or what actually even happens as part of the process. Some research has even shown that peer review barely changes the quality and content of research between the preprint and final published versions of articles (Klein et al. 2018;Carneiro et al. 2019); and yet, it remains almost universally highly regarded as the process which assigns most value to a piece of scholarly research.
Other knowledge generation systems, including FLOSS and Wikipedia, have more continuous 'review' systems that allow for greater differentiation of value, and on a more granular basis, than a binary state of 'peer reviewed' and 'non-peer reviewed' that continues to define research (Tennant et al. 2017). Non-peer reviewed work is often referred to as 'grey literature', and some researchers and communities explicitly state that they will not even read or use such works, and many journals forbid its citation in peer reviewed articles. This represents an unusual implication, as just because something has been 'peer reviewed' or not, does not mean that it should be treated any more or less critically as a responsible method of scholarship; especially when that review process remains concealed. Only recently has any progress been made in measuring the quality of peer review and reviewers themselves (Bianchi, Grimaldo, and Squazzoni 2019), aided by some publishers opening up their data on peer review (Bravo et al. 2019). However, given that critical information about peer review is typically kept private by publishers, it is hard to gauge reviewer and review quality, and yet curiously its use remains almost ubiquitous and held in high-regard (Callaham and Tercier 2007).
Any renewed understanding of 'impact' for Open Scholarship should encourage recognition of an increased diversity of research practices and outputs rather than those confined to peer reviewed research papers (Fecher and Friesike 2014). However, relevant and necessary changes in reward or 'incentive' structures are typically lagging behind this understanding at virtually all levels, including policy, institutional, and practical levels (Levin et al. 2016). The continued use of inappropriate metrics and proxies for value has contributed to a hypercompetitive culture widely known as 'publish or perish', and represents a vastly unscientific approach towards research evaluation. Social other essential components of research cycles, including software, materials, teaching, infrastructure, public engagement, and data, to be considered as 'second class' value objects and processes. This value distribution is now also becoming codified into policies as a new norm within Open Scholarship, where OA to research articles has undoubtedly had the biggest focus in the last 3 decades (Piwowar et al. 2018), but despite great progress has not been accompanied by concurrent changes in the evaluation system. This is peculiar as research articles could not exist without many of these other components, and yet they are consistently undervalued in terms of socio-cultural appreciation, evaluation, and recognition for funding ).
Alternative metrics: the alternative to what?
Present evaluation systems are dominated by metrics based on citation counts (Wagner 2016;McKiernan et al. 2019). Citation, in its rawest form, is usually an acknowledgement of an intellectual debt (Garfield 1979), and a hierarchical genealogy and network of ideas, dependent on a number of factors. Citation is an incredibly varied practice, and can represent an incredible array of different positive and negative relationships; yet, mostly because they are something that can be counted and are readily available from a number of sources, cumulative citation counts and a number of equally context-devoid metrics dominate valuation discourses and practices. Data regarding these metrics and indicators are typically maintained by commercial bibliometric databases, such as the Web of Science and Scopus, that, in turn, also contribute to setting the baseline for incentivisation.
Scholarly research is peculiar in that citations only between articles seem to have any form of currency, and not between different networks of objects, despite the principle and practice being essentially the same. For example, principles for software citation now also exist as a reflection of this (Smith, Katz, and Niemeyer 2016). Citation typically links recognition (as measured by citations) with reputation; this becomes problematic in the often dogmatic association of this reputation in one way or another to career and funding structure (i.e., success), which leads to cycles of self-reinforcement (more grants, better publications, thus more grants; i.e., the Matthew Effect) (R. K. Merton 1968;1988). Not only is this a misuse of the purpose of citations and a failure to understand what they are conceptually, it also misappropriates them in assuming that every citation is of equal value, as in a form of economic currency (Wouters 2014;Penders 2018).
The problem with using such a simplistic metric to measure impact and reputation is that it reinforces tendencies of risk aversion, lack of data transparency and many others, thus negatively affecting the Open Scholarship environment (Haustein, Bowman, and Costas 2015;Benedictus, Miedema, and Ferguson 2016;Wilsdon 2017;Wilsdon et al. 2017). This has led to the search for "Alternative Metrics" (popularly known as 'altmetrics') that can more accurately capture the diversity inherent within the wider publishing ecosystem beyond citations and peer reviewed articles (Priem et al. 2010). Examples of such metrics include mentions or shares on blogs, media reports, white papers and other indicators of a wider discussion or use of a given piece of research beyond traditional journal outlets.

Creating a new prestige economy based on open principles
If new metrics are employed and become prominent in assessing quality of research, it could potentially kick start a new set of biases within scholarly publishing. One way to circumvent this to an extent is to responsibly employ a combination of quantitative metrics and qualitative assessment (Hicks et al. 2015;Wilsdon et al. 2017), so the system becomes less dependent upon metrics; for example, similar to the Research Excellence Framework in the UK. The most important factor while choosing these metrics are the values which we want them to reflect (e.g., responsibility, trust, consistency), and how this is reflected in scholarly practice. To complicate matters further, Open Scholarship includes multiple contribution roles, such as authors, publishers, data analysers, software developers, reviewers, and editors, with a range of complex network interactions. To generate a robust framework for measuring impact and therefore reputation, different metrics and indices would be required to reflect this participant diversity (Nichols and Twidale 2017). Thus, any future appropriate valuation system has to capture the inherent diversity in dimensions between the practitioner, the practice, and the output.
As Cronin notes for the realm of Open Education, "complexity resides in determining and negotiating the value of open practice at an individual level, and structural and cultural barriers to openness persist within higher education" (Cronin 2018); but this surely also holds true for other dimensions of Open Scholarship. Thus, the question raised is how to measure reputation and impact within Open Scholarship. This critical factor has also received relatively little attention in the FOSS space (Hu, Zhao, and Cheng 2012), especially when compared to participation motivations. In FOSS, reputation is generally gathered based on one's ability to be a good coder, and the practices that come with that in an openly participatory environment (Kelty 2001). In this way, it is similar to academia in that prestige functions as its own sort of capital, and drives an 'economy' based on reputation. However, a study in 2016 demonstrated that additional factors such as past reputation based on positive evaluations, social network effects, and other demographic traits were important in defining reputation within an Open Source community (Cai and Zhu 2016). Furthermore, such positive reputations tended to lead to greater success and performance as a group, thus re-emphasising the collaborative basis for reputation building. Applying this to Open Scholarship, it seems that there is therefore a substantial amount of scope in realising it as a community-based movement focused on gift-giving (Feynman 1974), especially around the tools that allow others to create, akin to FOSS, rather than a commercially-governed system focused primarily on outputs. This would help to realign the focus of where the value lies most in research, exposed around the general process of collaborative creation, and not just purely focused on outputs.
This renewed view of reputation accrual through collaboration and sharing can also have a secondary benefit, in that issues to do with appropriate authorship and accreditation can be resolved by having 7 explicit and tangible guidelines for contributing to research projects. This is because the concept of authorship shifts radically from being something bound to an article, to something in which attribution is fundamentally transparent and easy to resolve, as well as dynamic based on ones' contributions, irrespective of what they might be. Thus, Open Scholarship workflow could seek to find new ways of quantifying or measuring contributions as different forms of 'impact' on research, including how the wider community interacts with processes and outputs. For example, these can be simple things borrowed from FOSS, such as project stars, forks, commits, and watchers (e.g., via GitHub). Each of these help to expose value in a different way, based on how content is perceived and digested by the wider community.
This would have several effects. First, credit could be obtained and granted for more than just authorship of research papers. Second, through valuation of a greater diversity of processes and outputs, these things become more widely incentivised. Third, the alignment of wider recognition would bring the reward system in line with that of Open Scholarship, and thus 'better' research or best research practices. It ultimately would mean that research that does not follow convention can be rewarded, thus fostering diversity, freedom of exploration, and creativity (Frankenhuis and Nettle 2018b). Prestige and reputation thus would return to principles of "good" scholarship, and not be outsourced to inappropriate and proprietary measures. Elements of such a system can already be seen in a range of digital platforms that leverage the power of networked communities to gain a deeper understanding of people and processes (Tennant 2018b). The bottom line is that the right measures of reputation across all levels of Open Scholarship could incentivize the right motivation.
For this to happen, a broad exploration of altmetrics must ensue to ensure a substantial shift in how impact and reputations are measured.
However, this focus on incentives, and in particular metrics, as a mechanism for cultural change towards openness could also have dangerous and unintended consequences. While there is little 7 CRediT -CASRAI.
doubt that rewards should be provided for more diverse forms of process and sharing beyond published articles, there is a risk that such evaluation procedures, and an increasing focus on 'alternative metrics', could lead to greater monitoring of researchers (and thus less freedom, innovation, and creativity), while still suffering from the same bureaucratic misuse as the impact factor does (Kansa 2014). One thing that OA, and Open Scholarship, has failed to do is to fully understand the significance and impact of such an entrenched paradigm that couples reputation, evaluation, and funding systems together, and how to alleviate or reform this. Goodhart's Law is commonly paraphrased as "When a measure becomes a target, it ceases to be a good measure." Through this, there is a further danger that some measures of openness can become used primarily as signalling mechanisms.

If the focus remains on individual researchers adopting the principles and practices of Open
Scholarship, the present evaluation system exerts a conservative pressure, invoking resistance to change. This is distinct from many parts of FOSS, where individuals could more easily form communities to help to fix elements of the system and solve problems, without explicit needs for substantial infrastructural or institutional support; indeed, there are many organisations that already provide financial support for individuals to create, contribute, and maintain FOSS. This creates a duality in which FOSS communities involve both individuals as part of their jobs, alongside groups of extrinsic contributors. Engagement with FOSS projects can be monitored and evaluated (e.g., through GitHub activity), which provides portfolio material for acknowledgement and credit as incentives for career progression. In this way, engagement with FOSS is built into the community through existing reward systems. Software development is more dynamic and independent of the institutional layer that the research community is rooted in.

Reproducibility and data sharing
This final section will look at the intersections between data sharing, FOSS, and issues around research reproducibility. What exactly constitutes replication, reproduction, replicability, or reproducibility, is often not very clear. This terminology is extremely complex, since many competing definitions are often used simultaneously. Here, we consider reproducibility to describe any measure that is taken to increase the verification of research results, which usually involves data sharing and thus falls under the banner of 'Open Science' (Baker 2016;Munafò et al. 2017;Wallach et al. 2018).
Reproducibility is perhaps one of the strongest pragmatic aspects that unites FOSS with Open Scholarship, being inherent to the efficiency and reliability of research (Sandve et al. 2013;Munafò et al. 2017;ter Riet, Storosum, and Zwinderman 2019).
Research that lacks openness to facilitate reproducibility can be compared to only making compiled source code available, where reverse engineering is one of the ways needed to reproduce software (at the mercy of the methods section). Open Scholarship is more equivalent to sharing the source code itself, allowing for immediate reproduction. This raises a key question over whether or not research can really be called "scientific" if it cannot be reproduced, computationally or otherwise. This is not equivalent to saying that research is 'wrong', just that it is unverified. In scholarly research, peer review is usually taken to signal verifiability, but in most cases does not actually indicate this as the process is concealed, and often focused on reviewing text rather than data and code. In software environments, computational reproducibility is now becoming more frequently considered throughout all stages of the research development pipeline, when executed efficiently; for example, with strong community management, good documentation, and using container technologies (Boettiger et al. 2015). FOSS can help to facilitate transparency and reproducibility of experiments within a specific environment, as the entire process is exposed and 'forkable'.
In the last decade in Europe much attention has been given to construction of research infrastructures (RIs) and e-infrastructures, initiatives where research communities can identify a governance body to plan policies, best practices, and digital services to be adopted by scientists to improve scientific workflows. Many researchers now perform their work in these "digital laboratories" (e.g., tools, services, standards, models), that are dependent on shared digital services for experimentation (e.g., virtual research environments, RIs, thematic services, workflow engines).
They use these to perform experiments that lead to the creation of new digital objects such as research data (e.g., time series, sensor data, maps) and research software (e.g. R scripts, executable workflows, scripts). RIs can support reproducibility and openness of research results. Computer scientists make use of digital resources to perform their experiments, ranging from their laptops, to the Internet itself, cloud resources, and shared on-line services or software as a service solutions.
These practices generate increasingly more data, a critical factor of scholarly research, yet not so much in pure software. RIs (and advanced virtual research environments) are devised exactly to enable a common (possibly online) environment where software can be executed in a shared, agreed on, trusted (certified?), platform (with again different degrees of complexity: plugged-in, embedded, web services, dynamic deployment, etc.). RIs take on board the cost of keeping a "common ground" of execution for scientists of a community with other benefits: economy of scale, big data close to processing facilities, encouraging common practices in software development, mitigating the obsolesce effect of software.
Here, software is the scientific product and runs in shared digital laboratories that are common to researchers. However, while FOSS might lead to better software and more efficient research, it alone does not necessarily lead directly to reproducible research. In other words, software alone is insufficient for reproducibility, but sharing software without the associated computational environment is a minimal requirement for reproducible research. Reproducibility is also affected by factors such as operating systems, inputs, parameter ranges and data availability, and experimental testing environments (Sandve et al. 2013;Millman and Perez 2014). For example, some widely-used R packages work differently in different versions of R. While FOSS, such as R or JASP, is not sufficient for open research and reproducibility, it certainly does promote these features. JASP, for example, provides users with the option to import/export and sync data with an Open Science Framework (OSF) account; an option that, at a minimum, facilitates and implicitly encourages such virtues and practices. Commercial alternatives like SPSS are more about the enclosure of services and communities. Within the scholarly ecosystem, reproducibility is focused on how the software is being used rather than being an intrinsic property of the software itself. Thus, reproducibility manifests in several different forms within the software-scholarship ecosystem.
Alongside FOSS, Open Data is now becoming an increasingly important and core aspect of Open Scholarship. This refers to the sharing of data, permitting others to analyse and reach the same conclusions for a piece of research -thus allowing research reproducibility -or to enable cumulative science by making data available; for example, for meta-analyses, and providing opportunities for new and originally unintended research. With this in mind, the fundamental elements underpinning research reproducibility are complete within an Open Scholarship system -articles, data, software, and all relevant research environment data are exposed and executable to help increase fundamental reproducibility.
The major tension here is that what is often considered to be best for research is mis-aligned with what is considered to be best for individuals or communities, as measured through evaluation criteria. Data sharing policies for journals and funders can be complex and difficult to enforce (Sholler et al. 2019). In the field of Industrial Ecology, for example, one recent paper stated "[editors of journals]...should consider asking authors for a major revision of manuscripts whose claims are not reproducible because of lack of documentation, data, or software. This would encourage authors to document their scientific claims with higher transparency and reproducibility." (Pauliuk et al. 2015). However, this focuses only on one element of reproducibility (i.e., results), and there are other concerns such as analytical, computational, and methodological reproducibility that should be considered. All of these are complex concepts in themselves, both philosophically and in terms of practical implementation, and they add additional dimensions or barriers to the adoption of 'open research practices'. For researchers then, it makes sense to consider these aspects prior to initiating a research project itself, rather than post hoc. This is, of course, bearing in mind the limitations associated with personal data or commercially sensitive data, such as those generated when collaboration between academia and industry is involved. These categories of data cannot be made open or public by default (unless anonymised or embargo periods have passed). Data sharing can in some cases be seen as a burden for researchers, who may be unprepared for the challenge of creating, treating and sharing data for external consumption rather than private use. Many funders and institutes now require research projects to have a data management plan, and these could be easily extended to a more holistic 'reproducibility plan', that outlines how the data, software, articles, and all other relevant methods and outputs will be documented and shared to increase reproducibility and potential for re-use.
As part of the Open Data movement, the increase in the implementation and promotion of the FAIR Similar to what has happened for OA, there is now a growing concern that Open Data too is being 'hijacked' by commercial organisations, who increasingly offer a range of research data services including repositories and management plans for researchers or research institutes as part of service packages (Posada and Chen 2018). This is part of a growing trend within the publishing sector, towards recognition of the commercial value behind such data. As such, this presents a new challenge for the public sector in that solutions and infrastructure around data sharing and storage might need to become more widely supported by non-commercial entities, including government-backed institutes. Therefore, these issues around data sharing also factor into core principles around reproducible research, research funding, as well as the incentive/reward system that governs how research is produced and communicated.
Computational reproducibility is not necessarily something simple to learn. One thing which is generally agreed upon is that computers and software are foundational to many aspects of modern research. Yet, despite this, computational reproducibility does not appear to receive due attention or credit in the training of researchers, or in the process of daily research. For undergraduate research teachings, where computational reproducibility is compromised in comparison to other software; for example, SPSS is seemingly often preferred to R, where code and environments can be more easily reproduced. Therefore, the uptake of code sharing seems to be slower in reaching some research fields. Generally, computational reproducibility and reliability are not considered perhaps as much as they should be, given the impact that they can have on comprehensibility and reliability of research results. The same can perhaps also be said often for data processing and management, with both data and software often being relegated to 'second class citizens' with little documentation, organisation, testing, version control. However, there are now many good examples of academic projects that try to improve this by promoting community standards, code review, and software such as Bioconductor and rOpenSci (Boettiger et al. 2015;Forero 2019).
A question that arises from this is whether FOSS and Open Data are prerequisites for reproducibility and Open Scholarship. This opens up additional issues by exposing research processes, for example allowing auditing of the methodologies, allowing attempts at reproducibility as part of the peer review process, and opening up participation in research processes to those beyond those with access to proprietary data/software (thus tending towards 'citizen science'). It also motivates or encourages different forms of 'talent' within scholarship, something which can only be exposed through transparency and inspection of the entire research process. Such transparency, in theory, also facilitates more productive criticism, evaluation, and diversity in forms of peer review. As such, this element of Open Scholarship has many parallels to the latter development of the FOSS movement (David 2008;Vermeir and Margócsy 2012;Lahti et al. 2017 being public at some stage, but only in the sense of not being secretive (Kelty 2001); however, from this it often ends up being privatised or obfuscated in ways that are not beneficial to wider society.
There is also likely an enormous 'file drawer' problem with many aspects of historical scholarship, and little in the way of understanding the scope of this at the present, or how to potentially resolve it. This is related to the wider issue of who scholarship is supposed to serve. If it is for the wider benefit of society, so that the maximum number of people can exploit and use it, then it has to be as rigorous and reliable as possible (Merton 1942). It clearly does also not meet this need if scholarly works are owned by private entities that actively prohibit its sharing and re-use. Thus, there is a direct tension here between Open Scholarship and organisations that maliciously violate fundamental freedoms underpinning this. However, without a solid principles-based foundation to Open Scholarship, and one that is widely accepted and enforced, then such violations will continue unchecked.

Conclusions
Open Scholarship might seem like a relatively new research term and paradigm, but it has deep historical roots in the very foundations of scholarship. Openness in one form or another has been part of the scientific ideal for centuries now. However, in the present way it is being discussed and often implemented, it could paradoxically lead to a more enclosed, monopolistic scholarly system dominated by 'siloes'. It seems that Open Scholarship, if based more on morality around freedom, would align itself more closely with Mertonian norms (Bowman and Keene 2018). In this way, Open Scholarship parallels FOSS in that it challenges proprietary systems as well as exclusive methods of scholarly knowledge production. However, two key distinctions exist between FOSS and Open Scholarship. The first is the inherent diversity and fragmented nature of the wider scholarly community, which makes any sort of monistic understanding of 'openness' difficult. The second is that the principles, values and respective 'freedoms' for Open Scholarship do not seem to have been as rigorously identified and implemented as in software communities. It is likely that the intersection between these two factors, combined with an overlaid homogenised system of research evaluation, is likely responsible for the relative slow growth of the Open Scholarship movement, which in turn has allowed the unintended capture of many elements of the space by proprietary entities.
If we consider Open Source to be more more about processes and methodologies, and Free Software more of a community-driven ideological movement (with some overlap), which of these best describes Open Scholarship? It seems that at its roots, Open Scholarship is a combination of both.
Often it is described as a more efficient, effective way of doing research, injecting more transparency and rigour/reproducibility into the process. Other times, it is described as an issue of equity and social justice. However, at least part of this divergence or ambiguity is due to the fact that Open Scholarship seems to encompass a diverse myriad of practices, outputs, and principles, which have not been connected together into a single, unified and unambiguous understanding. Thus, it appears that there is also no clear cut boundary -either in terms of principles/values, practices, or outputsbetween 'open' scholarship and 'closed' scholarship.
We can finalise this article by using two salient examples of how openness has failed to become normative in scholarly research. The first of these is the existence of the 'pirate' platform Sci-Hub, which provides illicit access to some 70 million paywalled research articles. Whether or not anyone agrees to the existence of this platform morally or legally is moot, given that it does exist, and exists purely for the reason that there is a fundamental access problem in scholarly research. The second example is the Wuhan Coronavirus, which became a global health emergency in the first months of Scholarship now readily exist, as do the moral and practical reasons for using such services, and it is simply the imperative of researchers to actually commit to them (Kupferschmidt 2020) .

The solution: A fully-open mandate
For the first few years of its existence, the Internet had the potential to form the technical and institutional basis for a new system of 'openness' in science. We believe that it has failed to achieve this for several reasons: 1. Evaluation metrics have not sufficiently changed from the pre-Internet world, and reputation and success are conflated; 2. Abuse of copyright (i.e., used counter to its initial intentions); 3. 'Closed' scientific practices, from the pre-Internet world, have been institutionalised.
Much of this can be ascribed to the increasing commercialisation of science, which often seems to stand in contrast to the agenda of science itself. Most metrics used are for short-term evaluations, whereas openness in science only offers a better diffusion and quality of science in the long-term.
We believe that the pragmatic way to reverse this problem is the formulation of new national and supra-national mandates, which could be based around the Foundations for Open Scholarship Strategy Development (Tennant, Beamer, et al. 2019). Governments and funding bodies should provide modern, sustainable and technically interoperable infrastructures based around existing established repositories, and all their associated functionalities (e.g., persistent identifiers, standardised metadata, research data repositories, usage metrics), with immediate, unrestricted, and full access to all research outputs. If implemented, this would simultaneously solve the major 9 outstanding issues with reliability, affordability, and functionality that global scholarship systems face and are in urgent need of fixing. We know that there is 'enough money in the system', currently being largely wasted on redundant services for commercial publishers, that would more than adequately finance this shift. The benefits of doing so would be immense, increasing the globalisation capacity of research, new services and infrastructures to be built on top of a huge knowledge database in order to, for example, help serve the UN Sustainable Development Goals. We also know that such services readily exist, as exemplified by case studies like the Coronavirus. This mandate would help publicly-funded research to speed up with respect to the relatively faster industry cycles, boost innovation and development in research, and enhance university-industry partnerships. Ownership and the value of research would be retained by the institutes that funded and created it, and dissemination and evaluation would no longer have to be out-sourced to commercial third-parties. The problems of 'predatory journals' would be diminished overnight as a side-effect, as there would be no incentives to publish with commercial third-parties anymore.
Commercial vendors would then actually have to compete fairly by providing real value-added services beyond the red herring of 'prestige'. To achieve this system, governments would need to simultaneously defund or depower the existing systems, including the highly profitable elements that exist within it, while constructing the new sustainable open scholarly infrastructures. But by having national funders working more closely with institutes (and research libraries), this would overcome the labour-intensive methods of mandating researchers to be more open, by more efficiently providing them with the tools and services to automatically do so. Critical infrastructure components including AmeliCA, OPERAS, SciELO, COAR and others are already running on limited finances and are highly effective. As another side effect of this mandate, it would prevent openness in scholarship from only being superficially addressed, or even corrupted by proprietary systems.
The capture of key elements of the open scholarship system by commercial players is not because they add any inherent value to the scholarship process, it is simply because they have the financial capacity to acquire such services. This amplifies the 'prestige' or 'validation' that these companies often distribute through their products in the eyes of public and academic institutes, while simultaneously alleviating the responsibility of public institutes and funders to fulfil the mandates given to them. Such ongoing privatisation of the research process is clearly not in the spirit of science and scholarship as fundamental to creation of learning and development of a commons around knowledge. Finally, by adopting a more commons approach to knowledge generation similar to FOSS, we expect that scholarship becomes an inherently more inclusive process, with equitable access and participation embedded in it as a core and fundamental value. Will the glass house of Open Scholarship be to ivory towers what the bazaar was to the cathedral in Open Source?
Author contributions JPT conceived of the idea for the project, which was collaboratively written as a MOOP. All co-authors provided text, edits, ideas, and discussion for this paper, which was written in the open and collaboratively using Google Docs. It received hundreds of contributions as part of an iterative and organic procedure.
constant source of inspiration. We also acknowledge that, because of the open writing process, there are some people who might have contributed thoughts, discussion, or comments, who have not been credited in the final author list. Thuds, we would like to extend our thanks to everyone who contributed to this in one way or another.

Disclaimer
Most of the authors of this manuscript grew up in a western context and are therefore potentially biased towards particular understandings of systems. Any additional information that helps us to fill any gaps in this manuscript are highly welcomed.