Introduction

Media art is sustained by institutions whose practices continually evolve to meet the requirements posed by artworks under their care. A capacious category, media art encompasses video art, artist moving image, new media art, and digital art, such as software- and browser-based works, also including the myriad physical and performative manifestations of these forms—artworks whose dynamic and variable constitution exceed more traditional institutional matrices, urging their adjustment. This need for institutional reform reflects the pressures and opportunities posed by the organizations’ milieu. New infrastructures and interfaces, policy frameworks and business models, as well as emergent modes of cultural production and research practices—all shape the socio-technical environment of knowledge production, calling for heritage institutions to reimagine the channels through which their collections, archives, and expertise can enter the currents of this ecology. The digital reconstitution of the public sphere urges reconsideration of what it means to perform as a public institution. Thus, mobilized, cultural institutions, as both custodians and mediators of social memory, undergo a transformation with respect to their internal organization, intra- and intersectoral relationships, and public-facing activity. Part of this effort, Media Art on Wikipedia is a project launched jointly by Amsterdam’s platform for Media Arts LIMA, Wikimedia Nederland, Van Abbemuseum, Stedelijk Museum Amsterdam, the Cultural Heritage Agency of the Netherlands, NDE, DEN and ZKM. Initiated in 2021, the three-year-long program aims to increase the visibility of media artists, their works, and the collections they are part of by bringing information about them into the Wikimedia ecosystem.

Supported by Wikimedia Nederland and the Wikipedian in Residence program, the project was focused on publishing five hundred Wikipedia articles profiling works and artists from the collections, developed through collaboration with universities and edit-a-thons hosted by LIMA and partner organizations. However, as the project was unfolding, allowing for exploration of Wikipedia’s underlying structure, Wikidata, the crowdsourced database behind Wikipedia, became an increasingly appealing opportunity to expose the information contained in the catalogues as linked open data (LOD)—data structured and published according to principles and standards supporting interconnection of individual datasets and their elements in a context-rich, machine- and human-readable way. The scope of the project was expanded and the creation of articles has been augmented by the effort to structure and represent collection data on Wikidata. To date, the project has generated Wikidata entries for two thousand media artists.

This article aims to situate the organizations’ use of Wikidata within the particular institutional context of media art and the broader turn towards open, participatory, and distributed reconfigurations of galleries, libraries, archives, museums, and other cultural institutions—the so-called GLAM sector. By providing an overview of the challenges faced by media art institutions and tracing some of the directions for institutional change, this article seeks to develop a more nuanced perspective from which to understand the role Wikidata and other LOD initiatives can play in supporting cultural heritage. In doing so, the text aims to address concerns over the GLAM sector’s ‘LOD fetishism’ (Winesmith 2017) by identifying how the Wikimedia ecosystem can respond to the unique demands of the field and contribute to the reform of its knowledge infrastructure.

The Content And Context of Media Art

Media artworks are “interactive, live-generated, and [...] interconnected [in] networks,” incorporating multimedia components and existing across time and space in multiple, parallel versions (Wijers 2019, 147). Their variable constitution requires institutions to continuously reinvent their methods of documentation, preservation, and historicization. The cross-disciplinary histories of media art call for their caretakers to adopt a similarly cross-disciplinary approach (Graham 2014, 11). The need for a variety of perspectives and knowledges, often falling beyond the scope of an institution, is further stressed by the technical complexity of the works (Mucelli and Gobira 2017, 348). As such, media artworks “rely on collaboration in stewardship among a diverse group of people” (Haidvogl and White 2020, 3). Key to comprehending their histories is documentation—traces which are “scattered and distributed,” amounting to “conjunctures that only attain meaning by their connection” (Dekker 2018b). Many of these traces exist in smaller archives, raising questions about how to locate and include these materials in art historical record (Wijers and Blome 2010, 57). Those needs translate in turn to demands on information management infrastructures, essential to effective stewardship of the works (Mucelli and Gobira 2017, 347). Thus, systems facilitating knowledge production, transfer, and retrieval emerge from the background of institutional practice and come to the fore as the area in which to address the demands of media art.

In recognition of these challenges, the field of media art has developed a vibrant discourse and a wealth of research and tools, generated by a number of inter-organizational projects (Giannachi 2016, 24). However, similarly to the artworks they address, these knowledges and structures of support are also threatened by disappearance, often incapable of moving beyond the start-up phase due to unsustainable funding structures or technical and organizational setups (Dekker et al. 2020; Saaze, Rasterhoff, and Archey 2020; Declaration 2011). Funding policies that prioritize short-term goals and the resulting project-based infrastructures (Saaze, Rasterhoff and Archey 2020) do not account for the necessary long-term development and maintenance. The mission of media art caretakers is thus undermined by a “throwaway culture [with] no interest and no funding for the preservation of works that need a lot of care, such as digital or code-based works” (Sollfrank and Stalder 2021). Most vulnerable to fluctuating revenue streams are smaller organizations, whose documentation materials and collections are a crucial part of media art history (Jones 2010, 46)¹. The oft-repeated claim about the ephemerality of media art can be seen as referring not to its intrinsic properties, but to the precarity of their institutional infrastructure.

Networked, communal, and crowd-oriented modes of knowledge production have thus been proposed as both a strategic approach to navigating this precarious condition as well as a methodology most attuned to the nature of artworks it addresses. However, while the need for “global networked collaboration in Media Art research” has been long stressed, emphasizing the “use of networked collaboration to archive key data in a cooperative process of knowledge transfer” (Declaration 2011), the organizational model through which these goals could be achieved has been conceptualized in more than one way.

Open Museum, Interactive and Networks of Care

Formulations of emergent knowledge infrastructures and practices depart from the “standard institutional approach,” in which a single organization is appointed to “lead in creating a monolithic central union database, [...] accessible to the public online as a central repository” (Rinehart and Ippolito 2014, 112). The main disadvantages recognized in such centralist arrangements concern their feasibility; the amount of resources needed for its infrastructure to be developed and maintained (Rinehart and Ippolito 2014, 112); the necessary selectiveness regarding such projects’ institutional partners; and the representational bias such exclusion would produce (Jones 2010, 45; Rinehart and Ippolito 2014, 112). In rejection of the traditional institutional strategy, concerns over scale and viability blend with a critique of institutionally legitimated canons.

Instead, the sphere of media art lends itself more to distributed topologies, expected to foster more inclusive and reciprocal modes of knowledge production. This commitment is not only a response to the unique exigencies of media art but also a part of a broader shift observable in the GLAM sector, demonstrated by the proliferation of the rhetorical figures of open (Rinehart and Ippolito 2014), participatory (Simon 2010), or distributed institutions (Rodley 2020). Marianne Ping-Huang characterizes this paradigm shift as a “new knowledge economy” in which the “vision for the universal archive [...] is transgressed in favour of distributed platforms” (2016, 57). Departure from monolithic structures cascades through various organizational scales of the cultural heritage sector—from intra-institutional, where a pursuit of a “catch-all system” accommodating “every piece of art information” has been replaced by attempts to integrate the organization's various subsystems into a cross-departmental mesh (Haidvogl and White 2020); to the intra- and inter-sectoral, as formulated for example through national frameworks of Netwerk Digitaal Erfgoed (Dutch Digital Heritage Network), eschewing homogenizing dependence on traffic-driven large internet platforms, stimulating instead an archipelago of LOD projects expected to facilitate greater diversity and independence of cultural heritage (Netwerk Digitaal Erfgoed 2021, 6).

Echoing this shift, Richard Rinehart and Jon Ippolito (2014) proposed the Open Museum as a speculative framework for rethinking the institutional infrastructure of digital heritage. The proposal imagines a replicable model of a memory institution as a locale to which artists deposit their work and from which “the source code and files can be copied and downloaded by other users” (Jones 2010, 47). Open Museum envisions the Museum Catalog Matrix, a standardized data system incorporating controlled vocabularies and folksonomies (Rinehart and Ippolito 2014, 108). Built upon the Matrix is the Interarchive, a distributed mutually linked database, searchable through general-purpose search engines and possible custom-built portals. Through such a setup, the model is imagined as a franchise, allowing institutions to share the burden of cataloguing and description as well as “distributing technical storage and backup,” while linking “formal and informal memory into a tougher historical mesh” (Rinehart and Ippolito 2014, 111–112). Rinehart’s and Ippolito’s proposal shifts the focus from the level of an organizational entity to the order of a protocol, emphasizing better use of already existing technologies and services.

As the authors themselves admit, the vision of the Open Museum and the Interarchive serves more as a thought experiment, than an implementable roadmap (Rinehart and Ippolito 2014, 113–114)—a conceptual tool for “rethinking memory institutions as repositories of the commons” (Sollfrank and Stalder 2021). However, it is precisely because of its speculative character that the Open Museum can demonstrate how media art’s institutional imagination anticipated some of the structures and mechanisms pursued today through GLAM’s growing investment in LOD. Some of the aspects of this framework remain outside the scope of what is attainable through current LOD projects (i.e. an open repository of not only the metadata but the digital artworks themselves). Nevertheless, as a conceptual tool for “thinking through requirements” (Rinehart and Ippolito 2014, 114) of cultural heritage networks and their systems, the Open Museum offers a valuable point of reference for reflecting on how LOD can answer to media art’s demands.

While the Open Museum serves as a utopian vantage from which to imagine institutional change, the concept of networks of care, proposed by Annet Dekker, has been developed by examining the concrete and concerted efforts undertaken by institutional and non-institutional caretakers towards the preservation of digital culture. A conceptualization of emergent practices, rather than an abstract model, networks of care denote “community-driven strategies” (Dekker 2016, 153) involving a “combination of professionals and non-experts,” individuals and organizations, each contributing knowledge from different backgrounds (Dekker 2018a, 89–91) and together forming a multiscalar and heterogeneous network. Networks of care are thus forms of knowledge communities, in which “everyone knows something, but no one knows everything” (Dekker 2018a, 89),² connecting various expertise, skillsets, and perspectives into a transdisciplinary mesh. Organized around a shared concern, rather than from a single locus of authority, the networks are bound by relations of shared and self-appointed responsibility. As such, rather than to speak of their institution, it is more useful to identify ways in which to facilitate their emergence and amplify their capacities.

Identifying conditions determining the efficacy of networks of care, Dekker points out the need for open systems whose users can “add, edit and manage information and track changes” (2018b). Such a system should also afford continuous accessibility to the network’s content and information, staying functional and reliant even when contributors decide to leave the project. Equally important is the network’s “common mode of sharing, where everyone in the group has access to all the documents or archives” (Dekker 2018a, 91). In other words, the multiplicity the network invites requires a norm structuring the relations of its various parts. Simultaneously, however, a system should avoid rigid standardization, ensuring instead “variability rather than creating a freeze state” (Dekker 2018a, 92). Networks of care are thus a site of tension and negotiation between communication’s requirement for a common protocol and a variability affording the inclusion of “pockets of disorder, patterns of fragmentation, and local practices of knowledge production” (Ping-Huang 2016, 61).

Similarly under negotiation are networks’ possible topologies. While networks of care are said to be able to “operate without the structures of centralized archives and authorized custodians,” such distributed and non-hierarchical arrangements cannot circumvent issues of responsibility and coordination (Dekker 2018a, 90–91). The concerns are echoed by GLAM professionals, who, while embracing the potentials of more horizontal, open, and reciprocal structures, simultaneously acknowledge the importance of central points at which these networks could cohere—a level of centralization can allow a network to “clarify and distribute responsibilities” (Dekker 2010, 7) while having “one central point of information” as well as “one strong organization hosting [an] initiative” can make it easier for dispersed actors to “work together on a practical level” (Dekker et al. 2020). Following these perspectives, certain forms of centralization do not have to be antithetical to the desire for more distributed structures that mobilizes media art institutions, offering instead practical advantages towards this goal.

This overview of media art’s institutional discourse, while modest and highly selective, provides at least a general sense of the challenges faced by the field and proposed solutions to them. Together, the models and structures discussed above articulate the parameters through which to understand and strategize memory institutions’ employment of digital systems and their involvement in digital networks. The institutional paradigms and conditions outlined above provide a context in which to examine the increasing role Wikimedia and the LOD ecosystem play in the practice of media art’s caretakers, as illuminated by Media Art on Wikipedia and other related initiatives.

The Emergent Topology of Wikibase Ecosystem

Media Art on Wikipedia’s use of Wikidata involves the creation of database entries for information about the collections of the project’s partner institutions, including statements about artists’ biographical data, institutional affiliations, and select artworks. The information is sourced from mediakunst.net—a portal unifying media art collections of the Netherlands’s cultural institutions and a precursor of Media Art on Wikipedia. The project is part of a broader effort to strengthen and coordinate Wiki-GLAM partnerships, intended to make Wikimedia projects “a growing part of the heritage professional toolkit” (Stinson, Wyatt and Fauconnier 2018, 17). The most recent addition to this suite, Wikibase, as well as its public instance Wikidata, has gradually gained attention among media art caretakers as a supplement or an alternative to institutional infrastructure’s conventional databases. Developed by the Wikimedia Foundation and the developer community, Wikibase is an open-source platform for creating, managing, publishing and linking structured data—an “umbrella of services” (Rossenova 2022) supporting the development of LOD. One of its earliest adopters is the New York-based digital art organization Rhizome, where Wikibase has been used since 2015 as the backend and frontend of its redesigned archive of born-digital artworks, Artbase (Espenschied, Moulds, and Rossenova 2018).³ Other organizations followed, proliferating experiments in the institutional use of Wikimedia tools for managing and publishing semantically structured data.⁴ Wikibase allows its users to structure data according to Resource Description Framework (RDF), a data structure whose triple-based assertions (object–predicate–subject) help expand the “mono-hierarchical tree structure, which has thus far shaped archival description practice” into “a multidimensional network.” Such semantic expressiveness not only lends itself to the description of media art's structural complexities but also opens up the “institutional data silos” (Messner 2020, 10), allowing for more flexible mapping of records across organizations.

As a platform for managing, linking, and exposing structured data, Wikibase “dramatically lowers the barrier of entry to Linked Open Data,” making institutional records interoperable even if they do not “strictly follow a specific metadata standard” (Rossenova 2020, 66). This flexibility allows for the development of classification ontologies that can “grow organically” (Rossenova 2021), building upon previous metadata schemas but evolving in accordance with objects of their description. Several projects currently develop data models for different types of complex artworks, using Wikidata ontology as their basis. For example, a wikiproject stemming from Rhizome's pilot case study of Artbase provides a space for collaborators to structure a model representing digital and performative artworks for the purposes of archiving, documentation, or preservation.⁵ A related project has been initiated as part of Media Art on Wikipedia, developing a model that could accommodate other expressions of media art, such as video art. The benefit of structuring these schemas through LOD is that the commonalities between the two can easily be merged or mapped. As such, LOD provides a promising framework in which to negotiate the tensions between the common and the particular, as they have been identified in networks of care—contributors to the ecosystem can deviate from established norms without becoming isolated in their singularity.

Furthermore, by drawing on existing schemas the models can foster links with other knowledge domains, supporting multidisciplinary approaches needed, for example, for preservation of the artworks. Wikibase and Wikidata furthermore provide for the elaboration of semantic statements through additional qualifiers and source references, allowing “for well-cited ambiguity and disagreement” (Crompton et al. 2020)—an affordance particularly valuable for media art, whose processual and variable nature prevents the artworks from being fixed in a stable, canonical form. In addition, references supporting the assertions can help in tracking the parallel provenance of media works whose histories rarely follow a linear path, instead branching through multiple institutional (and non-institutional) contexts. Although Wikidata is in itself an instance of Wikibase, it is important to highlight the differences and relations between the former and other, independently managed installations. Unlike Artbase, which runs on its own dedicated instance of Wikibase, Media Art on Wikipedia publishes collection records directly to Wikidata. While hosting a Wikibase installation allows an organization to define its ontology from scratch (Rossenova 2022), retaining “sole authorship and scoping rights” (Stinson, Wyatt and Fauconnier 2018, 26), the process of its configuration and maintenance require greater time investment and resources. In contrast, structuring and publishing metadata directly through Wikidata, while arguably less straining, requires institutions to cede some of their authority and embrace the positive and negative aspects of the platform's more collaborative and participatory mechanics, as well as its policies and ontology. On Wikidata, any user can add or edit the statements, allowing for crowdsourced contribution and refinement of data, but raising the risk of inaccuracies. Importantly, the “no original research” policy means that “Wikidata is not a database that stores facts about the world, but a secondary knowledge base that collects and links to references to such knowledge” (Rossenova 2022). As such, the platform is inherently dependent on external sources, including other Wikibase instances which, free to host original research data, can provide primary sources to be represented in Wikidata. The reference- and schema-based links between Wikidata and Wikibase instances (as well as between the instances themselves) decentralize the Wikibase ecosystem, in line with the Wikimedia movement's long-term strategy and vision (Rossenova 2022). Such development, however, simultaneously relies on the ecosystem's centralist aspects.

The emergent topology of the Wikibase ecosystem unfolds through the proposed differentiation of roles played by Wikidata and other Wikibase instances, relegating domain-specific knowledge to the latter while using the former as a repository of more general information. As an example of such higher-level identifiers, one can consider the statements created through Media art on Wikipedia. The statements representing artists, artworks, collections, and relations between them can serve as resources to which other, more granular information can be linked—potentially from other Wikibase instances. Offloading more detailed data to other instances may be necessary, given Wikidata's finite infrastructure resources. Furthermore, non-domain-specific knowledge is less likely to be erroneously overwritten by Wikidata contributors, thus increasing the accuracy of the data. The successful implementation of this model relies heavily on strategically fostering community engagement and streamlining the setup of Wikibase instances, achieved through the development and release of easily deployable installation templates. Following this scenario, Wikidata becomes an “umbrella service, linking together a family of independently maintained Wikibases” (Duchesne 2022), searchable in a coherent way through SPARQL—the SPARQL Protocol and RDF Query Language enabling federated (cross-database) retrieval of RDF-structured data using its encoded semantics. Aligned along this general-granular axis, Wikidata and other Wikibase instances together can create an ecosystem whose topology combines both centralized and distributed approaches (Rossenova 2022). The proposed hybrid approach addresses the issues of scale while simultaneously allowing knowledge communities and institutions to maintain the integrity of their data without the need to constantly monitor alterations made to their contributions.

Such federation largely delivers on the promise of the Interarchive while adding to it a level of flexibility allowing for it to span across different knowledge domains. Importantly, having Wikidata as a strong central hub, independent of media art’s funding cycles (Rossenova 2022), makes the Wikibase ecosystem a viable and promising infrastructural layer for media art institutions. However, there remain concerns regarding the sustainability of dedicated Wikibase instances, as they can be susceptible to the same issues undermining many of the previous infrastructural projects. Perhaps, following Rossenova's suggestion, Wikidata may serve as a safety net, ingesting data from Wikibase instances once their maintenance becomes unfeasible (Rossenova 2022). Lastly, an important question remains regarding the scope of Wikidata—at what point does knowledge become sufficiently domain-specific to warrant deployment of its dedicated Wikibase instance (Stinson, Wyatt and Fauconnier 2018, 26)? These are not purely technical concerns, but rather matters of negotiation between the institutions and other stakeholders of the knowledge ecosystem they co-produce.

Conclusion

Media Art on Wikipedia is a collaborative exploration of new modes of knowledge production based on the principles of distribution and accessibility. Aimed at strengthening the presence of media art in Wikipedia and Wikidata, the project simultaneously seeks to mobilize institutional knowledge, bringing it into the dynamic and evolving structure of the Wikibase and LOD ecosystem. The dissipation of hierarchical modes of organization, observed at the level of data, is reflected by the shift in the structural models proposed for the field of media art—from monolithic power nodes to networks of care. In Wikidata, LIMA and its partners find a dissemination channel opening up their institutional corpus of knowledge to circuits in which it can be mobilized beyond the institutional silo and potentially across disciplines. Situating these efforts in the specific institutional context of media art and its discourse helps in providing a deeper understanding of how LOD can help the field address its challenges. What such examination reveals is the alignment between the Wikibase ecosystem and the exigencies, visions, and strategies of the field of media art, encouraging further experimentation and testing of its possibilities.