An archive can save the disc and still lose the game.
That sounds backwards.
The disc is the thing sold, shelved, scratched, lent, and eventually placed behind glass. It has a label. It has a territory code. It usually has a date. It looks solid enough to outlast the machine that read it.
But a game isn’t only its media.
It’s the image. The timing. The language. The licensing path. The machine’s assumptions about a display. And sometimes a computer somewhere else that was never in the box.
A book needs a reader.
A game is more demanding about the arrangement.
Preservation answers that demand with admirable stubbornness.
The work dumps the cartridge, disc, or ROM chip byte for byte, then compares the result against a known-good checksum. No-Intro’s catalog makes a bad read distinguishable from a complete one. The file has to exist before worn contacts and a scratched sector become a small permanent lie.
That is preservation as retention.
It needs a stable object.
Without one there’s nothing to compare, study, or run.
An archive can’t preserve an atmosphere of approximate sameness.
It needs a version that can be named.
Yet naming a version is already a choice.
The Sega Genesis Sonic the Hedgehog that reached a PAL television carried the same name as the NTSC release.
It didn’t carry the same tempo.
NTSC transmitted 60 fields per second. PAL transmitted 50.
A naive conversion let anything tied to the display rate run slow. Enemy movement. Physics. Animation. The whole small insistence of the game that it was moving now.
The black borders of PAL letterboxing were visible evidence.
The slower physics were evidence with better manners.
They waited inside play.
This wasn’t a regional content edit.
Nobody removed a scene or renamed a weapon.
It was a technical adaptation, produced by incompatible broadcast standards and an engineering budget. Reprogramming internal timing for 50Hz cost money. Converting the output naively cost the European release a fraction of its speed.
Commerce is excellent at converting an engineering invoice into an authentic regional characteristic.
Later hardware escaped the bargain unevenly.
Some consoles offered a 60Hz mode in PAL territories. Some did not.
A preservation record that stores only the image flattens that history.
One file. One title. An absent condition that once made the thing run differently.
Release dates show the same problem from the other side.
Final Fantasy VII reached Japan in January 1997, North America in September, and Europe in November.
Those gaps weren’t a calendar failing to turn its pages quickly enough.
Post-gold localization means adapting a game after the source-language version is finished. Sim-ship means the regional work proceeds in parallel. The first arrangement structurally produces delay. The second spends more to prevent it.
Keep translating. Keep testing. Keep waiting for one market’s present to finish becoming a past.
Treat the Japanese release as the sole original and the price is clear enough.
It makes every later audience a latecomer to someone else’s work.
Treat every localized build as interchangeable and the price is less flattering.
It erases the labor, the mistakes, the code changes.
And the commercial doubt that made those releases distinct.
Pokémon Red and Blue make the distinction unusually literal.
The Western releases were reconstructed from the newer Japanese Blue codebase rather than translated from the older Red and Green, because that older source was reportedly too deteriorated to work from.
They aren’t the same game in a different language.
They’re an answer to a broken line of production.
Content edits complicate the archive’s appetite for a tidy family tree.
Fallout 3 renamed morphine and removed its injection animation for Australia.
The original submission had met a classification problem.
That edit then appeared in every territory’s build.
Germany separately removed blood and gore.
Japan separately disabled a nuclear-detonation quest option and renamed a weapon.
One territory’s requirement became a global artifact.
Another’s stayed local.
The disc remembers the outcome.
Without its territory and build history, it doesn’t remember who imposed it, why, or how far the change travelled.
This is why “censorship” is often a little too pleased with itself as an explanation.
A classification board, a platform holder’s licensing policy, and a publisher’s own calculation are three different causes. They can produce the same missing animation. They do not leave the same history behind, however efficiently a catalogue compresses them into “edited.”
Preservation as completeness has its own cost.
Every regional build. Every timing variant. Every altered localization.
Keeping them means admitting the work has no single convenient body.
It multiplies storage, description, legal exposure, and the researcher’s afternoon. An archive wants a canonical object partly because cataloguing is hard enough already, without also preserving the argument the publisher had with every territory.
But preservation as singularity has the sharper cost.
It makes a market decision look like an authorial one.
It calls the surviving copy definitive because it survived.
That is the oldest and cheapest kind of authority.
Even an exact dump can’t settle the matter.
A locally complete game can be preserved from its dump.
A server-dependent game needs the server code and data, or a private-server route, to remain itself at all.
When a title is delisted and its servers close, an owner is left holding a client with nothing behind it.
The disc is intact. The disc has not changed.
The disc is also, now, furniture.
The law turns that distinction into another kind of border.
In the United States, a Section 1201 exemption lets qualifying libraries, archives, and museums preserve lawfully acquired games. Playable access stays on the institution’s physical premises. The most recent triennial rulemaking kept that limit for video games.
The saved game may be alive, then.
So long as it stays in the reading room.
Preservation has occasionally discovered that a locked door is also a display case.
No build can be crowned without a loss somewhere.
A single verified build makes scholarship and survival possible.
A shelf of distinct builds keeps technical, economic and regulatory history from becoming an accidental footnote.
The archive gets to be legible or it gets to be true.
Rarely both at once.
One release is the game the way one photograph is a decade.
Which differences must remain visible for a game to have a history?