/ Rhizome ArtBase
A central case study for my own archival work on The Wrong Biennale is the Rhizome ArtBase. As a leading platform for displaying and preserving net art, Rhizome provides a critical framework for understanding the institutional practices of digital preservation. My analysis aims to extract principles that can be translated into my independent archive of projects from The Wrong Biennale (2022-2023). With limited resources, I seek to establish a sustainable model for preservation, making the ArtBase's operational logic, its balance of accessibility, metadata structure, and explicit acknowledgement of digital decay, a vital reference point.
The interview I conducted with Mark Beasley specifically addressed the user experience of such archives:
Tatiana Isaeva: When I first started looking at net art projects, the websites were confusing. I'm interested: to what extent, when building a website, are you thinking about the end user? Or to what extent is it more about playing based on artistic practice?
Mark Beasley: Yeah, I think there's a lack of scholarship and accessible historicizing around internet art. That's something we're trying to rectify. If you've explored ArtBase, it's a very wonky technical interface. We are working to make it more accessible and friendly... We want to make it clearer when there are links to living work or just a record of something ephemeral. Consistency in interface is important so it's clear if you're clicking to view net art or a record of something that doesn't exist anymore (Beasley, 2024, personal communication). 4
Making the interface a transparent part of the archival design is a key insight I aim to implement. This chapter provides a detailed analysis of Rhizome's archival practices, highlighting observations applicable to my own project.
Mark Tribe, the creator of the digital platform Rhizome.org, maintained a sustained interest in online art, particularly in what he termed “the idea of direct access to the public” (Tribe in Cook, 2001, p. 138). 1 Launched initially as a mailing list in February 1996, Rhizome emerged while Tribe was residing in Berlin and regularly attending festivals such as the Dutch Electronic Art Festival, ISEA, and Ars Electronica. He observed that conversations within these events remained largely confined to small local communities, lacking a broader international discursive space. The prototype for Rhizome was the Nettime platform, which, as Tribe noted, was “focused on critical discourse and network practice, but not really focused on artistic practice” (Tribe in Cook, 2001, p. 139). 2 Through Rhizome, Tribe sought to overcome these limitations by establishing a digital platform where net artists could exhibit work and exchange ideas, enabling participation from anywhere in the world. For Tribe, the ambition extended beyond mere information exchange; it was, “more importantly, to construct and develop a critical dialogue” (Tribe in Cook, 2001, p. 140). 3
Shortly after its inception, Rhizome bifurcated into Rhizome Raw, which hosted a high volume of posts, and Rhizome Digest, a curated selection of the most relevant articles. The Digest quickly proved more popular, amassing over five thousand subscribers. During the summer of 1996, Tribe, alongside Alex Galloway, then director of content and technology, began archiving selected posts from Raw into an online database named “Textbase.” By 1999, this curatorial impulse evolved into the systematic archiving of net art works themselves, giving rise to the ‘ArtBase’. This initiative responded to a practical necessity: numerous links to artworks and critical texts were succumbing to link rot, disappearing from the live web. In a deliberate departure from traditional exclusive curatorial models, the archival policy was inclusive; any artist could submit their project to the platform for potential preservation. As Tribe explained during a 2001 talk, “I guess my main thought about Rhizome in connection with curatorial practice is this notion of filtering rather than curating, of having multiple layers from totally open to progressively filtered in different ways. […] We are trying to create a many-to-many communication environment” (Tribe in Cook, 2001, p. 140). 4 This statement presciently outlines a shift towards more engaged and participatory curatorial models.
Today, Rhizome’s ArtBase stands as one of the most substantial archives of internet art, a vast database documenting the use of digital space, and the organisation has developed its own software, such as Webrecorder, which enables users to create personal archives of web pages and artworks. This aligns with the concept of ‘public curation’ discussed by Christiane Paul in Curating Immateriality, where the database functions as an open-source platform and the curator assumes the role of a ‘filter feeder’, selecting and categorising submitted artworks. Rhizome’s long-standing relationship with the New Museum in New York, formalised in 2003, has provided institutional support, leading to physical exhibitions such as Artbase 101 (2005) and Net Art Anthology (2019). These exhibitions materialise the archive, presenting historically significant net art within a gallery context. Initiatives like the Seven on Seven conference, which pairs artists with technologists, further reflect Rhizome’s commitment to collaborative and interdisciplinary practice, echoing earlier experiments like Billy Klüver and Robert Rauschenberg’s Experiments in Art and Technology (E.A.T.). This approach resonates with Paul’s model of multiple curatorial perspectives, advancing new forms of collaboration in internet art exhibitions. Consequently, examining Rhizome’s enduring pursuit of physical exhibition spaces helps illuminate the broader motivations behind internet art’s status as a spatial nomad over the past decade.
The development of the ArtBase itself reflects an evolving philosophy of digital preservation. Initially, artists could submit works as either “cloned” or “linked” objects—a hybrid model where cloned works were stored on Rhizome’s servers, while linked works remained hosted by the artist, with Rhizome providing only metadata and a redirect. As founder Mark Tribe noted, “The idea was that we would attempt to include everything that was within the boundaries of new media art, as we then understood it” (Tribe, 2018). This strategy offered flexibility but proved unsustainable; linked objects were highly vulnerable to decay, prompting a shift in policy. The accession process has transitioned through three distinct phases: an initial period of open submission (1999–2010), a subsequent phase of filtered submission (2010–2015), and a current model of acquisition by invitation only (2015–present). This evolution was driven by the practical challenges of preserving complex digital artefacts. As former digital conservator Ben Fino-Radin observed, early attempts to impose order via detailed questionnaires were ultimately less effective than qualitative, interview-based research for capturing the information necessary for long-term preservation (Fino-Radin, 2018). 1
Faced with the immense challenge of scale, Rhizome’s focus has shifted from being a comprehensive sharing platform towards a more retrospective, research-oriented archive dedicated to preservation, restoration, and reperformance. This re-conceptualisation is supported by ongoing technical development. A significant overhaul of the ArtBase infrastructure, supported by a grant from the National Endowment for the Humanities, has implemented Linked Open Data (LOD) standards, making the archive’s metadata structured, machine-readable, and interoperable. The relaunched ArtBase now operates a hybrid accession format, combining targeted open calls with invitation-only submissions, thereby balancing community engagement with the rigorous demands of sustainable digital preservation. This continuous adaptation underscores Rhizome’s central role in defining the methodologies and philosophies of preserving born-digital heritage.
The principles of how metadata is structured and categorised are central to what I aim to implement in The Wrong Archives. I am creating backend datasets with information I find relevant regarding the 148 curatorial spaces shown in The Wrong Biennale 2022-2023, focusing on how they can be categorised, preserved, and, most importantly, made readable for machines. Using a platform like Wix and its dynamic pages, rather than manually creating a unique page for each space, is a direct application of the scalable, resource-conscious logic observed in institutional practices. It saves a tremendous amount of time and makes the project manageable for a single curator. By converting screen recordings and research into machine-readable data and implementing the machine as a co-curator of sorts, the archival process becomes more systematic and sustainable, directly informed by the evolving model pioneered by Rhizome.
Browsing the Archive
The ArtBase archive consists of two distinct modes of browsing, each reflecting a different way of access and user engagement. The first is a standard search function, allowing users to query the archive by artist name or year of creation. This approach presents results in a simple list structure with clickable links for both queries. The evolution of this interface is itself a narrative of changing web conventions and archival strategies. Early iterations of the ArtBase, from 1999 onwards, reflected the text-heavy, slower-speed characteristics of the early web, consisting primarily of alphabetised lists navigated via vertical scroll and pagination. Browsing was facilitated by lists sortable by artist, title, or keywords, with date-based sorting introduced in 2007.
A significant shift occurred around 2011, by which time the "Web 2.0" era was well established. Rhizome introduced a more visually-led interface, where image-based grids supplanted text-based lists as the core navigational element. These paginated grids, displaying approximately twenty-five works per page, could be sorted alphabetically by artist, title, or tag. Each artwork page featured tags that generated lists of related works, and additional discovery mechanisms included featured sections, a dynamic timeline widget, and member-curated exhibitions. A subsequent update introduced a landing page with a magazine-style carousel and staff-curated exhibitions, creating multiple entry points into an archive that by then hosted over two thousand artworks. The 2015 platform overhaul further updated the grid's aesthetics, integrating it with the overall site design but reducing immediate accessibility by hiding artwork titles until the user hovers over the image thumbnail. As Zachary Kaplan explained, this design choice was a deliberate rhetorical gesture: it aimed to "de-emphasize the claim to credibility" of the entire ArtBase, acknowledging the disparity between various artefacts and the organisation's unresolved strategies at the time. The current (2025) landing page presents a minimalist interface, offering two hyperlinked options—'browse by artist name' and 'browse by date'—alongside a small, randomly selected sample of twenty-seven artworks from 1991 to 2021. The artworks are presented as cards featuring an image, the artwork name, its production year, and the artist's name, tagged as a ‘person’. The hover effect on the image signals its interactivity.
The second, more complex mode of access is through SPARQL queries, a functionality directly connected to the underlying structure of the ArtBase. The archive's initial data model was a "basic web model," structured around a custom taxonomy devised by Rhizome's founding staff. As Mark Tribe recalled, "We had to come up with a whole taxonomy. What do you put in those fields? Basically, Alex [Galloway] and Jennifer [Crowe] and I just made them up" (Tribe, 2018). This ad-hoc approach, influenced by limited resources and a lack of established institutional models for born-digital art, had long-term consequences for the archive's backend. A move towards standardisation began after the shift from open submission, when Rhizome had the resources to hire dedicated preservation staff. An effort led by the Director of Technology, Nick Hasty, along with David Nolen and Mushon Zer-Aviv, aimed to elevate the ArtBase from a basic web model to an authoritative records system, with plans to implement metadata schemas based on the Dublin Core standard. This culminated in a transfer of data to Collective Access, an open-source collections management software. However, the subsequent preservation director, Dragan Espenschied, opted for a different direction, seeking a system better suited to the heterogeneity of internet art than traditional museum databases. He initiated a transfer of the ArtBase data into Wikibase, an open-source software for managing structured data. Espenschied noted that Wikibase's flexible schema of items, properties, and qualifiers offered far more adaptability than classic database systems, which rely on fixed categories ill-suited to describing internet art's variable and collaborative nature.
From the user's perspective, the implementation of Wikibase enables a more technical and creative mode of inquiry. The archive provides a SPARQL endpoint with a graphical user interface. SPARQL (SPARQL Protocol and RDF Query Language) allows users to query and visualise the data in real time through tables, image grids, charts, and maps. This transforms the act of searching from a simple hyperlink navigation into an active process of data exploration, facilitating unexpected discoveries about artworks, artists, and software, and embodying the linked open data principles that now underpin the archive's structure.
Artwork Interface
The artist pages within the ArtBase consist of several interactive and static elements designed to present the artwork and its associated data. The primary visual component is a representation of the artwork itself, which may take the form of a screenshot, an animated image, or a video recording. Beneath this, two key containers organise the informational content: one for a descriptive text about the work, and a second for holding its metadata. This metadata container is particularly significant, as it can potentially present data to the viewer in several different forms, reflecting the complex nature of digital preservation.
The first and most straightforward type is ‘descriptive metadata’, which includes fundamental details such as the artist’s name, the artwork’s title, and its year of creation. The challenge of structuring metadata was aptly outlined by Richard Rinehart, who identified the core types necessary for each artwork: descriptive, administrative, and technical (Rinehart, 2018, p. 45). 1 However, defining what constitutes sufficient metadata within each category remains a complex issue. While descriptive metadata has been collected for most ArtBase works, its completeness varies significantly. Categories like artist name, title, and date are often less straightforward than they appear for net art, which may involve collectives, changing titles, or processual creation periods. Administrative metadata, relating to acquisition dates and catalogue IDs, has been partially preserved, but data on copyright licensing is considered unreliable. For a long time, the default submission form field was set to “Creative Commons,” leading to potential inaccuracies if artists did not fully understand the implications. The redesigned archive interface presents an opportunity for Rhizome to clarify licensing on a case-by-case basis. The least complete data in the ArtBase is technical metadata. Submission form fields were frequently left blank, and not all collected data proved useful for restoration. The critical question remains: what constitutes the minimum necessary technical metadata for effective preservation? As Rinehart observed, artists should select only the minimally necessary technology choices to run the work, but he did not elaborate in detail on what this minimum should encompass (Rinehart, 2018, p. 58). 2
The ArtBase presents artworks to the user through six distinct archival formats, each representing a different preservation strategy. The most basic is the ‘outside link’, which typically directs users to a version of the artwork maintained by the artist. The ‘file copy’ and ‘server copy’ categories involve hosting a copy of the artwork’s files on Rhizome’s infrastructure, functioning as a file repository, an approach similar to the ‘video’ type. The final two formats are more innovative and directly address the preservation challenges discussed in previous chapters. The ‘emulated instance’ allows a user to replicate the server environment of an artwork that no longer exists on the live web on their own device. This strategy aligns with Rinehart’s recommendation of emulation over migration for preserving the aesthetic and historical characteristics of original formats (Rinehart, 2018, p. 102). 3 For Rhizome, emulation has been facilitated through projects like the bwFLA Emulation-as-a-Service, which uses a cloud-based architecture to allow users to interact with original software environments directly within their browsers, eliminating the need to download additional software. However, as Rinehart noted, a limitation of emulation is that it focuses on standalone computers rather than the network itself; it cannot replicate the entire internet environment, including live data streams, that some net artworks integrally depend upon (Rinehart, 2018, p. 105). 4
This is where the final and most comprehensive format, the ‘webrecording’, becomes essential. For works that link to external data sources or require user interaction, a version preserved as a WARC (Web ARChive) file is often the most appropriate solution. Tools developed by Rhizome, such as Webrecorder and its current service iteration, Conifer (2025), address this need directly. These tools create an interactive copy of a webpage by recording the server-client traffic in real-time as a user browses, capturing content revealed through interactions like playing media, scrolling, and clicking. This method can preserve the complex, behavioural aspects of net art that static copies or emulations cannot. For example, a web-archived version of a work can redirect broken links to archived versions of those external resources, reconstructing the artwork’s original networked context. This approach is particularly vital for preserving artworks that exist across third-party platforms or were initially submitted only as ‘linked objects’. By utilising these advanced tools, Rhizome can accession and restore works as complete archival experiences, effectively mitigating the condition that pure emulation cannot capture the full, interactive life of the network as it existed at the moment of the artwork’s creation.
My analysis of Rhizome’s ArtBase reveals two key insights that directly inform my independent archival project, The Wrong Archives. First, it demonstrates that an archive is not a static repository but a living system that must evolve with web conventions and preservation technologies. Second, it highlights a spectrum of preservation strategies, from complex, resource-intensive methods like emulation and webrecording to more accessible documentation practices. For my project, focused on documenting the 148 curatorial spaces from The Wrong Biennale (2022-2023), the institutional practices of full emulation or webrecording are beyond the scope of an independent curator with limited technical resources and data storage. Instead, I adapt the principle of documentation as preservation. My primary method is creating screen recordings of each pavilion, hosted on YouTube. My archival focus anlike Artbase net artworks precvation shifted from preserving individual artworks to mapping the metadata of curatorial spaces. My tagging system does not describe artworks but catalogues the architectural, interactive, and conceptual features of each pavilion. This structured data allows for machine-readable analysis to identify patterns, such as the most common platform choices or interaction models. For instance, evidence from the length of my screen recordings shows that 3D environments take longer to explore, which directly influences my own curatorial choices for the final exhibition, Alternative Realities Illustrated.