Essay · Preservation
A game is more than its files
Archives can keep code, media and documentation. They cannot automatically restore the conditions in which a game was played, understood and changed.
This essay is based on the sources listed below. It does not claim first-hand access to every game, build or institution discussed.
Access is often the first thing missing
Before anyone can argue about an older game, someone has to be able to encounter it. A 2023 US study by the Video Game History Foundation and Software Preservation Network drew a random sample of 1,500 games released before 2010. At the cutoff date, 13 percent were still commercially available; the reported 95 percent confidence interval was plus or minus 2.5 percentage points.
That number belongs to this sample, market and date: 15 April 2023. It does not mean that 87 percent of every game ever made has disappeared. It does not measure private collections, unofficial copies or the quality of later reissues. The narrower finding still matters: the commercial market kept most of the sampled history out of release.
The playable object has several layers
The Library of Congress treats software and video games as compound objects. Its format guidance names human-readable source code, a gold-master build, documentation, metadata and the technical environment. A console game may require its original platform or an emulated version of it.
A build without the right environment can be complete as data and unusable as a game. A recording shows one run, not every possible branch. Source code exposes structures but does not replace operating them through controls. An emulator may support serious research without recreating display latency, controller shape, network services or every undocumented hardware quirk. That difference is not a preservation failure. It is part of the record.
Games also have a social record
Some losses sit outside any executable. Matchmaking populations disappear. Balance patches alter viable strategies. Platform rules change access. Guides, mods and conversations move or are deleted. A restored build can recover mechanics while the culture that once interpreted them survives only in fragments.
Criticism should therefore not treat a contemporary emulation session as direct access to the past. It is a new encounter with a preserved system, shaped by present hardware, present knowledge and the evidence that survived. The useful question is which claims that particular evidence can carry.
Good preservation makes criticism more precise
A manual, a build, a source repository, a capture and a player's recollection answer different questions. Treating them as interchangeable turns availability into authority. Keeping them separate leads to better sentences: name the version and platform, distinguish observation from documentation, mark missing network or peripheral features, and call a reconstruction a reconstruction.
A game is more than its files. That is precisely why the files matter: they can be stabilized, described and compared. The rest needs testimony, context and criticism disciplined enough to say where its evidence ends.
Production note
Source discovery, structure and drafting were assisted by OpenAI Codex. Matthias Ramahi is the responsible editor. The essay was written for the current independent publication and does not reuse a former Nightmare Mode article or contributor identity.