Who keeps the lights on: inside the volunteers archiving games their publishers deleted

The Torrent Arcade storefront went dark on a Wednesday morning with six weeks of notice, and Priya Ravel spent five of those weeks buying games she had no intention of playing.

She was not collecting. She was capturing. Ravel coordinates the Ember Index, a volunteer project that documents and, where it can, preserves games that are about to become unavailable. By the time Torrent Arcade closed, the Index had catalogued 613 titles that existed nowhere else — no physical release, no second storefront, no publisher still in business to re-release them. Forty-one of those had never been documented in any form beyond a store page.

When the page went, so did the screenshots, the description, the credits list, and in eleven cases the only public record that the game had ever existed at all. The software is still on somebody’s drive somewhere. That is not the same as the game being preserved, and the distinction is what the Ember Index exists to argue about.

The scale of the problem is difficult to state precisely, which is itself part of the problem. Nobody publishes a list of what has been delisted. Ravel’s working estimate, based on the Index’s own tracking across eight storefronts over six years, is that something between four and seven per cent of commercially released digital games become unpurchasable within a decade of release, and that the rate is accelerating as licensing terms shorten and back catalogues change hands.

How games actually disappear

Almost none of it is malicious. That is what makes it hard to stop.

A game ships with a licensed soundtrack on a ten-year term. A racing game licenses vehicle likenesses. A sports title licenses names. A studio uses middleware whose agreement is renewed on a rolling basis, and when the studio is acquired, somebody reviews the renewal cost against a title selling forty copies a month and reaches the obvious conclusion. None of these decisions is unreasonable in isolation. Their aggregate effect is a slow, invisible deletion.

Consolidation compresses the timeline. When a catalogue moves to a new owner, every one of those agreements gets re-examined at once by people with no institutional attachment to the games. That is the pattern behind the questions now hanging over Longshore Digital’s back catalogue following its acquisition by Aurelian Media Group — not a plan to delete anything, just a portfolio review, arriving at the same place by accident.

Nobody sets out to erase a game. Somebody sets out to close a spreadsheet, and a game is inside it.

Online components make it worse. A single-player game with an authentication check becomes unplayable when the check has nowhere to phone. A game built entirely around its servers becomes, on shutdown, an installer that opens to an error screen. The Index classifies these separately: the first is recoverable with a patch nobody is legally clear to write, the second is closer to a demolished building.

The legal position is worse than most people assume

Ravel is careful in a way that becomes noticeable over a long conversation. The Ember Index publishes metadata, screenshots, credits, manuals and interviews freely. It does not distribute games. Copies it holds are held under specific research exemptions, in a single location, accessible on request to named researchers, and the paperwork behind that arrangement took two years and a great deal of donated legal time.

The gap between what is legal and what is useful is enormous. In most jurisdictions a library may preserve a copy but may not circumvent copy protection to do so, and may not make the result available remotely — which means a preserved game is one you can play by travelling to a building during opening hours, if a working machine happens to be available. For a medium that is fundamentally interactive, that is a strange kind of preservation, and Ravel is blunt about it.

Publishers, for their part, are not uniformly hostile. Three mid-sized companies have quietly handed the Index build archives and source material for delisted titles on the condition that nothing is distributed. Ravel will not name them, because the arrangements are informal and depend on individual people who could be replaced tomorrow. That fragility is the point: the current system for preserving games depends almost entirely on personal goodwill inside companies that reorganise constantly.

What actually helps

Ravel’s asks are unromantic. Notice periods on delistings measured in months rather than weeks, so that documentation can be captured. Publishers releasing offline authentication patches at end of life, which several already do and which costs a few engineer-days. Clear statements when a title is delisted, so archivists know it happened rather than discovering it from a broken link.

She is unsentimental about which games deserve it, too. The Index does not rank. It documents a well-regarded immersive sim and a shovelware puzzle game with the same template, because the judgement about what mattered is not one that can be made now. Cascade, the commercial failure that is now visibly influential a decade on, would have failed any curatorial filter applied in its own release year.

That is the argument, ultimately. Preservation is not about which games were good. It is about keeping the option open to find out later — and about a medium that has now been commercially significant for half a century still having no reliable mechanism for remembering itself.

The Ember Index has eleven regular volunteers and a hosting bill that Ravel covers with a spreadsheet and a donation link. Six hundred and thirteen games from one storefront closure. She is expecting another announcement before the year is out, and she is not wrong very often.