Speedrunning communities preserve games by doing something publishers rarely bother with: documenting exactly how the code behaves, down to the frame. That work, built over years by unpaid hobbyists probing glitches and memory states, has produced technical records detailed enough that software historians now cite community wikis as primary sources. It matters because publishers are shutting games down faster than most players expect, and a runner's notes may be the only record of how the thing actually worked.
The scale of the shutdown problem is what makes this preservation work more than a hobbyist curiosity. Electronic Arts alone lists a dozen titles retiring in 2026, from Anthem in January to Grid Legends in September, according to EA's own online services shutdown register. Anthem was built online-only, so when the servers went dark on January 12, 2026, every copy of the game stopped working, with no offline mode offered.
Not every closure ends that badly, and the difference usually comes down to whether anyone outside the studio understood the game well enough to keep a version running. We covered a connected angle in Why esports casting language works so well on our brains.
Why do speedrunners understand old games better than the studios that made them?
Casual play only requires knowing how a game is supposed to work. Competitive speedrunning, where fractions of a second decide rankings, requires knowing how a game actually works at the code level, including every bug the original developers never caught. That gap is the entire reason the hobby produces such thorough documentation, according to reporting on speedrunning's reverse-engineering culture.
Runners build tools that resemble professional software analysis more than gaming peripherals. RAM watch utilities let a runner track memory addresses in real time, watching position data, timers, and random-number seeds change frame by frame. Tool-assisted speedrun projects go further, using scripted input sequences to map the absolute limits of what a game's code allows. That work has, on multiple documented occasions, surfaced bugs the original developers never knew existed, decades after release.
The output is a community wiki, not a corporate archive. For a meaningful number of older titles, especially ones from studios that no longer exist in their original form, the most accurate technical record of how the game behaves lives in a fan-maintained page built by people with no formal relationship to the developer at all.
What actually happens to a game after the servers go dark?
The outcome depends entirely on how the game was built, and the record so far is mixed. Volunteers connected to the Stop Killing Games campaign surveyed more than 700 online-only titles and published results in May 2025: over 68% of the games surveyed were already dead or in the process of shutting down, according to the Stop Killing Games volunteer survey.
Among the titles that did survive their official discontinuation, fans preserved 110 of them. Developers themselves salvaged only 16, per that same survey. That split is the clearest evidence yet that community effort, not publisher goodwill, is doing most of the preservation work industry-wide.
Concord is the sharpest example of what that effort looks like in practice. Sony pulled the game offline on September 6, 2024, fourteen days after its August 23 launch, and refunded every buyer, according to BBC reporting on the Stop Killing Games campaign. In November 2025, a small group of developers rebuilt Concord's backend through reverse engineering and ran full matches on custom servers, per the shutdown statistics report covering the incident. Sony responded with copyright claims that took the demonstration videos off YouTube, but the private servers reportedly stayed up.
MultiVersus offers a gentler version of the same story. The game lost online play on May 30, 2025, but kept its local and single-player modes intact, which meant fans preserving it had something to build on rather than nothing at all.
Could publishers just design games to survive their own shutdown?
Yes, and the added cost may be smaller than it sounds, according to Ross Scott, founder of the Stop Killing Games movement, speaking on a recorded discussion recapped by Ars Technica. Scott argued that most online games already run local test environments internally, separate from the live servers, to keep development moving if the main infrastructure goes down. If that test environment already exists, he said, "that's half the battle right there" toward giving players an end-of-life option.
Scott's broader argument isn't that publishers must keep servers running forever. It's that shutdowns should come with planning: an end-of-life patch, an offline mode, or software that lets players keep hosting the game themselves. He compared the current standard practice, no notice period and no fallback, to a warranty that simply doesn't say when it expires.
The European Commission considered exactly this question after the Stop Killing Games petition gathered 1,294,188 verified signatures and became a valid European Citizens' Initiative, per the European Citizens' Initiative register. On June 16, 2026, the Commission said it could not propose a legal requirement to keep games playable, but it did commit to opening talks with the industry on a code of conduct for how games are retired. A separate UK petition on the same issue closed with 189,887 signatures and was debated in the House of Commons on November 3, 2025, though the government said it had no plans to change consumer law.
Why do fans keep fighting for games the industry has already written off?
For a lot of players, the fight isn't abstract. Ubisoft shut down The Crew in 2024, a racing game that had drawn more than 12 million players over its lifetime, and one longtime player told the BBC the shutdown "felt personal" because the game had become something he shared with his kids, per BBC's reporting on the campaign. Whammy4, who founded a fan community dedicated to preserving The Crew, compared a server shutdown to someone breaking into your home and taking something you already paid for.
That emotional pull is part of why older games stay competitive long after their sell-by date: the people keeping a scene alive are often the same people who refuse to let the underlying software disappear. A NYU Stern professor of games business, Joost van Dreunen, put the structural problem simply in the same BBC piece: live-service games function more like ongoing digital communities than one-time purchases, and "every new live-service game invents its own demise" once its audience shrinks and the publisher decides it's time to move on. For related coverage, see Why older games stay competitive long after their sell-by date.
What this means for players and the games they care about
Academia Hagi is an online publication covering competitive gaming, not a broadcaster or an event organizer, and none of this preservation work happens on our platform. What the evidence shows is a pattern: publishers control the shutdown date, but they rarely control what happens to the knowledge afterward. Speedrunning wikis, TAS documentation, and fan-run servers have become the default archive, not because any studio planned it that way, but because no one else was doing the work.
The practical steps available to a community facing an eventual shutdown are limited but real. Documenting mechanics early, before a shutdown notice arrives, gives a scene something to preserve later. Community-run servers, where technically possible and legally uncontested, keep competitive viability alive even after official support ends. And organizing, the way Stop Killing Games has, keeps pressure on regulators even when the near-term answer is disappointing. None of that guarantees a game survives. It does mean the record of how the game worked, thanks largely to speedrunners, usually does.
