Chapter 8 · 8 of 9
Where this sits in the field
vmctl doesn’t start from scratch. Conserving software-based art is an established discipline with mature tools and a deep literature — here is what is already solved, and the one stubbornly manual step this work automates.
Keeping software-based art alive is not a new field, and vmctl is not a replacement for any of it. A generation of conservators, archivists, and engineers has built the tools and the thinking this stands on. It is worth being precise about what they already solved — and about the single thing that is still done by hand.
Running a chosen environment is solved
Emulation-as-a-Service (EaaS / EaaSI; Klaus Rechert and colleagues at Freiburg and Yale, with the Software Preservation Network) lets a conservator run a legacy environment inside a browser and share those environments across institutions. Olive / VMNetX at Carnegie Mellon streams whole encapsulated virtual machines on a click; the Internet Archive and oldweb.today run period systems in the browser. These give you access to an environment — once a human already knows which one a work needs.
Reviving the networked web is well charted
Rhizome — through Webrecorder and Conifer, and oldweb.today (Dragan Espenschied, Ilya Kreymer) — pioneered “symmetric” web archiving and the revival of the dead servers a net-artwork still tries to talk to. That is the closest existing work to the network side of the exhibition controller, and it is mature practice.
The conservation thinking is deep
The Variable Media Network (Guggenheim; Jon Ippolito) reframed media art as behaviour rather than object. Tate’s software-based art programme (Patricia Falcão, Tom Ensom) formalised significant properties, dependency, and the maintain–migrate–emulate decision. Strikingly, that literature already names the exact task this workbench is built around — the “derivation of a recipe for reconstructing a suitable technical environment” for a work’s execution — but describes it as a manual, “laborious and highly specialised” activity, done by hand, one work at a time.
What is not solved — and what vmctl adds
Every tool above still needs a human to already know how to run the work. Emulation-as-a-Service matches a known file format to a pre-built environment from a curated library; its own authors wrote, plainly, that they “don’t yet have Artificial Intelligence algorithms… we are currently manually pre-configuring.” Format-identification tools (DROID, Siegfried, JHOVE) stop at telling you what a file is, never how to run it. Nobody automates the upstream question: hand a system an unknown work and have it propose how to bring it back.
That upstream triage — a conservator’s first and hardest days — is exactly what the analysis module automates, and this workbench is where a human then verifies the proposal and turns it into a conservation recipe. vmctl complements Emulation-as-a-Service and Rhizome; it does not compete with them — it feeds them.