Chapter 1 · 1 of 6

What you get back

A run hands you three things you can act on — a per-work knowledge base, a runbook, and a suggested machine and toolkit — and, from the same evidence, three audience-specific views: a dependency manifest for the conservator, an asset catalogue for the researcher, and ‘the artist’s hand’ for the art historian. This is a tool for art historians as much as engineers. All of it is a proposal, flagged where it is unsure, never silently applied to the work.

You have been handed an unknown born-digital work — a zip, a disc image (a single file holding the complete contents of a CD or hard disk), a loose folder — and the questions of your first days are always the same: what is this, which file starts it, what software did it need, and what kind of machine would run it again? artwork-autopsy reads the work end to end and answers those questions on paper. It decides nothing irreversible and touches the work at no point. What it produces is a single report that fans out into three things you can pick up and use.

1 A per-work knowledge base

Everything the analysis learned about this work, written up as readable notes plus a structured inventory. It is the dossier you wish came with the files — and the research workbench’s assistant can read it directly, so the help you get there already knows what was found.

2 A runbook

Step-by-step instructions for bringing the work up — the machine, the software inside it, the network it expects — in the right order. Wherever the analysis could not be certain (code it could not read through, a server that no longer answers, a quiet failure it suspects but cannot confirm) the step carries a plain FLAG: check this by hand when you run it.

3 A suggested machine & toolkit

A period base machine to start from — a computer of the right era, re-created in software — and the tools worth pointing at the work: the opening configuration for a session at the bench, so you are not guessing the era from a cold start.

A real analysis report scrolled to its verdict: the work's identity — a Flash web page launched from index.html reaching hundreds of network endpoints — followed by what it does, how it behaves, and what it needs to run, each claim backed by file paths and endpoints. — Figure 1. The report is a plain web page — readable without any special software, now or in twenty years.

Three views on the same evidence

The report, runbook, and suggestions above are what the workbench acts on. But the same evidence also serves people who will never boot a virtual machine (a period computer re-created in software), so the analysis composes three audience-specific views — because this is a tool for art historians as much as engineers:

Dependency manifest

For the conservator. Exactly what must be provided to stand the work back up — the player software it needs, the versions, the peripherals, the dead servers to serve back — distilled into a checklist you can work down.

Asset catalogue

For the researcher. What the work is actually made of — every image, sound, video clip and script the bytes gave up, itemised so the material of the piece is legible without opening it in a period tool.

The artist’s hand

For the art historian. The artist’s own words, harvested from the source and the recovered code — code comments, authorship and copyright lines, contact URLs, the voice inside the work. The pass extracts the words; the historian reads them — mechanical and interpretive in equal measure.

A proposal, never a verdict

Nothing here is presented as fact. Every line is a fallible, labelled suggestion for a person to accept, adjust, or throw out. Two rules sit underneath all of it — the same rules everyone who actually keeps old interactive media alive works by (the Internet Archive, Rhizome, the demoscene, abandonware):

Never destroy the original. The raw bundle is kept untouched; everything produced is a derived working copy, never an edit to the work itself.
Label what we guessed. Every automated finding is marked as a proposal, so you always know what was read from the bytes and what was inferred.

Access-first, and honest about it The stance is re-exhibition, not reverent preservation: get the work running and showable again, with the guesswork visible — rather than freeze a folder of files that no longer runs.