Lost Net Art Preservation — A Nordic Perspective on Disappearing Digital Heritage

When Adobe ended Flash support in December 2020, thousands of interactive artworks stopped loading. The crisis was real. But framing it as a Flash problem misses the point — systematic net art loss started decades earlier, through format obsolescence, dead platforms, and institutions that were never designed to preserve executable environments. Tools like Ruffle, EaaS, and Conifer recover some of what remains. Nordic archives — especially Sweden’s Kulturarw3 — document a paradox: formally archived, functionally inaccessible.

The Systemic Causes of Net Art Loss — Flash EOL and the Obsolescence Crisis

Flash EOL in December 2020 was a trigger, not a cause. The underlying problem is that net art has always been entangled with proprietary, short-lived runtime environments that no institution was mandated to maintain.

Adobe’s end-of-life deadline for Flash was 31 December 2020. After years of decline — mobile exclusion starting with the iPhone in 2007, Steve Jobs’ open letter in 2010, Adobe’s own discontinuation announcement in 2017 — the plug was finally pulled. Browser vendors followed. Flash Player disappeared from operating systems. Works that had loaded without friction for a decade simply stopped.

The SWF format — Shockwave Flash, the container for Flash animations and interactive pieces — was and remains a proprietary binary format. Adobe never published a complete, open specification for it. That single fact is worth sitting with. Thousands of artworks were created in a format whose technical definition was owned by a private company with no archival mandate. When the company moved on, the format became opaque. This is not a story unique to Flash. VRML, Shockwave (Director), Java Applets — the graveyard of browser-based interactive formats from the 1990s and early 2000s is large. Flash was just the biggest tenant.

The migration window existed — roughly 2010 to 2020, as HTML5 matured and became the realistic alternative. Most artists working alone, without institutional support, missed it entirely. The artworks that survived that decade intact are those that either had an institution behind them or happened to be simple enough to reconstruct. The rest joined the queue of losses that had been accumulating since the mid-1990s.

Format obsolescence is the structural argument. Every proprietary format is a ticking clock. Open formats — SVG, WebGL, JavaScript — have different failure modes, but their specifications are public and their implementations are distributed. A closed format fails when one company’s attention moves elsewhere. Institutional blind spots complete the picture: web-based interactive art arrived before most collecting institutions had developed policies for it, and those institutions that did collect it often preserved documentation rather than executability. A screenshot of a net art piece is not the piece.

The Rhizome Methodology — EaaS, Conifer, and the Net Art Anthology

Rhizome’s preservation infrastructure is the methodological baseline for the field. Its frame is New York, Walker Art Center, institutional budgets. The geography shapes what gets preserved.

Rhizome launched its ArtBase in 1999. The collection now holds more than 2,200 works — a net art archive built over two-and-a-half decades with real institutional depth. The Net Art Anthology project, running 2016 to 2019, went further: rather than passive archiving, it restaged and recontextualized one significant net art work per week, creating curatorial access to works that would otherwise exist only as broken URLs or PDF documentation.

The technical core of Rhizome’s preservation approach is Emulation as a Service — EaaS. The framework was developed at the University of Freiburg and has been deployed by Rhizome since at least 2020 as the primary mechanism for giving users access to legacy software environments. EaaS runs full operating-system emulations in the cloud: a piece created in Internet Explorer 6 on Windows 2000 can be accessed through a remote browser session that genuinely runs that software stack. The fidelity is high. The infrastructure cost is also high, which is why Rhizome’s preservation services [VERIFY before publish] charge between $500 and $1,500 for an initial assessment — a figure that immediately separates institutional clients from individual artists.

Conifer (formerly Webrecorder) is the other major tool in Rhizome’s portfolio. Where EaaS emulates the runtime environment, Conifer captures it: it records a live browser session — including dynamic JavaScript, embedded media, and user interactions — as a WARC (Web ARChive) file. A researcher can then replay that WARC and experience the archived site in an emulated browser context. Rhizome built Conifer with Mellon Foundation funding; it is open-source and free at a basic tier (5 GB), which puts it within reach of smaller institutions and individual artists in a way that EaaS is not.

The Rhizome frame is important to name honestly. Its institutional partners — Walker Art Center, Museum of the Moving Image, Vienna Museum of Science and Technology — are major collecting institutions with budgets and staff positions for digital conservation. That constellation of resources is not representative of Nordic or European media art institutions outside the major centres. Kulturarw3 is Stockholm. Nasjonalbiblioteket is Oslo. The infrastructure, the mandate, and the budget look different from there.

The Kulturarw3 Paradox — Archived But Not Playable

Kulturarw3 is the oldest national web archive in the world. The archive is impressive in scope. It is also broken for interactive net art.

Sweden’s Kungliga biblioteket (KB) — the national library — launched Kulturarw3 on 24 March 1997 [VERIFY before publish]. It was the first complete national web archive in the world — not just an experimental crawl but a systematic, ongoing programme to capture the Swedish web. Since that start, KB has collected more than 500 million Swedish web pages, totalling over 5 billion files and approximately 350 TB of data. Those figures come directly from KB’s own collection descriptions, updated as recently as August 2025.

The access terms tell you something important. To look at Kulturarw3, you must travel to KB’s reading room in Stockholm. There is no remote access, no API, no web interface that works from outside the building. You enter a URL — specifically a URL, not a search term — and you see whether KB has snapshots of that address and from which dates. This is not a technical limitation that got overlooked; it reflects a legal framework around how national archives can make copyright-protected material available.

Here is where the paradox sits. Kulturarw3 is formally one of the most comprehensive records of the Swedish web that exists. It captured the years when Flash-based net art was at its most active — the late 1990s through the 2000s. Many Swedish artist and institution websites from that period are present in the archive as URL snapshots. But archiving a URL is not the same as archiving an executable environment. A Flash piece is not a document — it is a program that requires a runtime to run. KB’s crawlers captured the HTML containers, the SWF files, the surrounding pages. What KB did not build — and what, from public documentation, there is no evidence it has built — is any emulation layer that could execute those SWF files for a researcher sitting at the Kulturarw3 terminal in Stockholm.

Whether Kulturarw3 has made any structural provision for Flash emulation, or whether this is simply a known gap in the archive’s functionality, is not answerable from publicly available KB documentation. This is a hypothesis that requires direct contact with KB’s digital collections team or on-site research. The question matters enormously for anyone trying to understand what actually survives of Swedish net art from that decade. [Research Note: hypothesis — verify via direct KB inquiry or on-site visit before citing as confirmed.]

The pattern is not unique to Sweden. Nasjonalbiblioteket in Norway has been crawling the .no domain annually since 2001 and is a member of the International Internet Preservation Consortium (IIPC) [VERIFY before publish]. The same structural question applies: national library crawls capture files, not environments. Archiving and accessibility are two different problems, and the field has been better at the first than the second.

The KB Webbpublicerat collection — the e-plikt stream, active from 2013 — adds another layer to this. Unlike Kulturarw3, which crawls whole sites, e-plikt covers individual published items. The KB documentation states clearly: the two programmes collect different things. Kulturarw3 captures site structure. E-plikt captures individual publications. Neither was designed with interactive Flash environments as the primary use case. The archive covers the Swedish web. It does not necessarily preserve Swedish net art as art.

Nordic Preservation Cases — What We Know and What Remains to Be Researched

This section requires a direct acknowledgement: no independently verified catalogue of named Nordic net artworks with confirmed preservation status — lost, extant, or partially accessible via Kulturarw3 or Nasjonalbiblioteket — was researchable from public sources at the time of writing. The KB archive is accessible only on-site in Stockholm. Monoskop’s Sweden entries [VERIFY before publish] list platforms and institutions but do not document individual works’ archival status. Moderna Museet’s digital collection documentation is not granular at the individual Flash-work level from external access.

What can be said with confidence:

The structural conditions for significant loss are present. Sweden had an active Flash-based art scene in the early 2000s — the Electrohype biennials in Malmö (2000, 2002, 2004, 2006, 2008, 2010) were explicitly focused on computer-based and web-based art during the peak Flash period. Works exhibited at those events were browser-based in many cases. Whether they survive in executable form is, at this point, an open research question. [Research Note: plausibly significant losses occurred; the absence of verified surviving examples from that specific context is itself significant and warrants formal study.]

Kristoffer Gansing’s research at Linnaeus University provides one verified anchor in the Nordic theoretical literature — his work on network-based art and techno-aesthetics of infrastructure addresses these questions at a structural level without cataloguing specific lost works. The Monoskop entry for him includes a 2023 publication, Homegrown, Outsourced, Organized, which may contain the closest thing to a Nordic institutional analysis currently in print.

For comparison: the RIXC Riga has built a Latvian Electronic Arts Archive that documents specific works and their preservation status. The Baltic model — smaller country, dedicated media arts organisation, explicit archival mandate — may be more instructive for the Nordic context than the Rhizome model. It suggests that institution-building around preservation is a choice, not a consequence of national wealth or technological capacity. [Research Note: RIXC Riga archive existence is confirmed; whether its documentation methodology is directly comparable to the Nordic situation requires verification.]

The honest framing for Nordic net art preservation is this: the archive infrastructure exists (KB, Nasjonalbiblioteket) but was not designed to preserve executability. The art scene was active during the Flash era. The intersection of those two facts — an active scene plus an archive that captures but cannot execute — is likely a substantial body of permanently inaccessible work. That hypothesis has not been formally tested. The research does not yet exist. This article is partly a call for it to be done.

E-Plikt and the European Rights Framework

Sweden’s E-Plikt has been in force since 2013 — but interactive net art largely falls outside what it covers, and the gap is structural, not accidental.

E-plikt is the Swedish term for legal deposit applied to electronic publications. KB began receiving mandatory electronic deposit in 2013, under legislation that extends the principle of physical legal deposit — Sweden has required deposit of printed works since 1661 — into the digital domain. The scope covers publicly accessible online publications with a Swedish connection: web-based articles, e-books, podcasts, films, government publications.

The coverage gap for net art is built into the definition of “publication.” E-plikt operates on items — discrete, identifiable digital objects that function as publications. An artist’s website built in Flash that operates as an interactive environment is not, under the current framework, a publication in the relevant sense. It has no ISBN or ISSN. It is not classifiable as an e-book or an article. The KB documentation notes explicitly that e-plikt covers “enskilda delar” — individual items — rather than whole websites. Kulturarw3 covers whole sites. Neither covers executable interactive art as a category.

The Norwegian legal deposit situation follows a comparable logic. Nasjonalbiblioteket’s mandate covers Norwegian online publications, and the national library has been an active IIPC member since the consortium’s formation — IIPC members share methodology and coordinate crawl infrastructure internationally. But IIPC membership and WARC-format crawling do not solve the executability problem any more than Kulturarw3 does. The format is captured; the runtime is not.

EU copyright law adds a further layer. Article 4 of the 2019 DSM Directive (Digital Single Market) creates an exception for text and data mining by research organisations, which is relevant for mass-analysis of archived web content but does not directly enable emulation of copyrighted software environments for preservation display. The specific question — whether a national library can legally deploy EaaS to execute Flash content from its archive for a research visitor without rights clearance from the work’s creators — remains unresolved in the Nordic jurisdictions from public legal analysis. Rights clearance is a practical barrier even when technical solutions exist. [Research Note: specific legal position in SE/NO/FI requires verification against current national copyright law; this is the current author’s reading of the framework, not a legal opinion.]

The argument for E-Plikt revision is direct: the law was designed for a publication model, and net art is not a publication. A revised framework that treats interactive digital environments as a distinct category — with its own deposit requirements and access conditions — would need to grapple with executability, not just file capture. That is a more demanding brief than anything current Nordic legal deposit legislation contemplates.

Artist Self-Archiving — Practical Steps

Institutions cannot save what artists have not documented. Self-archiving is not a substitute for institutional mandate — but it is the only option most artists currently have.

The Dullaart-Sakrowski Method, named by Rhizome [VERIFY before publish] in their December 2020 Flash preservation guide, describes a low-tech documentation practice: record a guided screencast of the work being used, narrating what happens and why it matters. No servers. No emulation infrastructure. A video file that shows the piece working before the runtime disappears. For works where full emulation is not feasible, this is the fallback of last resort — and it is far better than nothing.

For works that can still be captured in a browser, Conifer (webrecorder.net) [VERIFY before publish] remains the most accessible tool. It is browser-based, free at the 5 GB tier, and produces WARC files that can be replayed in compatible viewers. For Flash-based work specifically, the window for live capture has closed — you cannot capture a Flash site in a current browser. But for JavaScript-heavy, WebGL, or platform-dependent works that currently function, Conifer is a practical first step for artists without institutional support.

Ruffle [VERIFY before publish] is the Flash-specific emulator — written in Rust and compiled to WebAssembly, which means it runs natively in modern browsers without a plugin. The Internet Archive integrated Ruffle into its Flash collection in November 2020, allowing browser-based playback of a significant subset of archived SWF files. Ruffle’s compatibility is not 100%: works that used advanced ActionScript 3 features, or that interacted with external services or browser elements in complex ways, may not play correctly. Self-contained, single-file Flash pieces are the best candidates.

For metadata, artists documenting their own work face a choice of frameworks. PREMIS (Preservation Metadata: Implementation Strategies) is the institutional standard, developed by the Library of Congress; it is comprehensive and maps well to institutional archives but has a steep learning curve for individual use. OAIS (Open Archival Information System, ISO 14721) is the reference model that institutional archives are built around — useful to understand as a framework, but not a tool an individual artist deploys directly. CodeMeta is the most practical option for artists whose work is software-based: it is a schema for software metadata, JSON-LD based, and designed to be human-writable. For a Flash piece with an .fla source file, documenting the work using CodeMeta — title, creator, programming language, date created, repository — creates a machine-readable record that can persist independently of any specific archive.

The practical priority list for an artist with a body of net art that still functions:

1. Record screencasts of every work — narrated, dated, showing full interaction.
2. Preserve source files (.fla, .swf, project directories) in at least two geographically separate locations.
3. For live JavaScript/HTML work: capture with Conifer now, before a platform dependency breaks.
4. Write a CodeMeta record for each work. One JSON file per piece. Store it with the source.
5. Contact KB ([email protected]) if your work has a .se domain and you want to verify it was captured.

This is not a long-term solution. It is what exists.

Outlook — Preservation as a Continuous Obligation

Nordic institutions need interoperable emulation infrastructure — not just URL crawls. Three positions, clearly held.

First: the current model of archiving URLs while ignoring executability is not preservation. It is filing. Kulturarw3 has 350 TB of filed Swedish web history. The portion of that which is interactive net art is formally present and practically inaccessible. Calling that an archive is accurate only in the most literal sense.

Second: artist self-archiving cannot substitute for institutional mandate. The Dullaart-Sakrowski Method is valuable. Individual artists should absolutely document their work. But placing the preservation burden on individual creators — who have no archival training, no storage infrastructure, and no institutional continuity — is a category error. Institutional archives exist because individual stewardship of cultural objects does not scale and does not survive personal circumstances. Net art is not different.

Third: E-Plikt revision is overdue for interactive digital art. The 2013 framework was a significant step. A 2026 revision that created a distinct deposit category for executable digital environments — with accompanying rights provisions that allow institutions to deploy emulation for legitimate research access — would be more significant. The IIPC has been developing shared methodology for exactly this kind of infrastructure; Nordic institutions are members and have the technical relationships to act on this if there is political will to do so.

The resource argument is real. Rhizome’s EaaS infrastructure costs money. Walker Art Center can fund a digital conservation programme. Most Nordic media art institutions cannot. The gap between what is technically possible and what is institutionally funded is where most of the losses happen. Smaller institutions need tools and shared infrastructure, not just inspiration from New York case studies.

The EU’s digital heritage initiatives — including the Europeana framework and ongoing work within the Digital Single Market — are the most plausible policy lever for this at scale. Whether they will move fast enough to matter for the works currently at risk is an open question.

Frequently Asked Questions