The People Who Actually Saved PC Gaming History
Go look up any DOS game you played as a kid. Pick one. Try to buy it legally right now, today, from a store that will give you a working copy on a modern machine.
If you picked something from the top 50 most popular titles of the era, you might get lucky on GOG. Maybe Steam has a version someone bothered to maintain. But if you picked anything outside that narrow window (and there were thousands of games outside that window), you’re going to hit a wall. No storefront to buy from, no downloads that are guaranteed safe, just a game that resides in your memory and possibly decaying on an aging floppy sitting in someone’s attic or garage.
The Video Game History Foundation has an official number as of 2023. They surveyed over 4,000 titles from before 2010 and found that more than 80 percent of them have no commercial availability of any kind. The comparison that stuck with me: the survival rate for American silent films is about 14%. Pre-World War II audio recordings, about 10%. Classic video games sit right there alongside them at 13%. We’re talking about the same category of cultural loss that we’ve already accepted as tragedy in film and music, except most people have no idea it’s happening with games.
And the people actually doing something about it aren’t the ones you’d expect.
The Commercial Platforms Aren’t Fixing This
I want to be fair here because this isn’t all bad news, and one company in particular deserves real credit.
GOG began their official Preservation Program in 2024, by February 20th of this year they were up to 267 preserved games. From a 50 foot view that means they are focused on actively maintaining compatibility, keeping games DRM-free, and someone will actually answer when something breaks. The Wolf Among Us and The Last Express just made the cut. They’ve got a GDC 2026 panel lined up too. Credit where it’s due: nobody else on the commercial side is even attempting this.
Steam, on the other hand, has nothing. No preservation program. No backward compatibility initiative. If the developer who published a game in 2003 doesn’t maintain it themselves, Valve doesn’t step in. Old titles break because of outdated DirectX calls, or gameplay logic that was tied to CPU clock speed, or DRM that conflicts with modern Windows, and nobody at Valve picks up the phone about it. Their position, whether they’d phrase it this way or not, is that this is someone else’s problem.
And even GOG’s preservation isn’t bulletproof. I ran into this firsthand with Wing Commander III. Windows 11 updated to 25H2 and suddenly all joystick controls were dead. If it wasn’t forcing the joystick to the upper left part of the screen no matter how calibration went, the controls were completely gone. The game couldn’t see the hardware correctly due to Microsoft’s reworking of HID device identifiers and GOG’s DOSBox build was passing this hardware to the emulation layer. I spent days trying to figure this out, something the retro community fights actively. I was going through every GOG forum, Win 25H2 rollback threads, I even found an ISO of 24H2 and tried to get it working again. It eventually turned out to be that GOG had a newly updated DOSBox version in the installer. When I jumped over to my LaunchBox and ran the eXo version of WC3 I almost fell out of my chair when it just worked out of the box. Everything worked, the joystick with all button mappings, I was even running a synthesized Roland MT-32 game soundtrack. Less than a minute and I was playing WC3, cursing at Kilrathi. That’s the difference between a commercial platform maintaining 267 titles and a volunteer who’s tuned 7,666 of them to actually work.
So here’s the number that reframes everything. GOG has preserved 267 games through their program. A single anonymous volunteer who goes by the name eXo has preserved 7,666 DOS games. Just the DOS collection. He has three other collections on top of that. We’ll get to those.
That gap (267 versus over 10,000) tells you everything about what commercial viability filtering does to preservation scope. GOG can only justify preserving titles that enough people will buy. eXo doesn’t have that constraint. He’s preserving all of it.
The Underground That Built the Foundation
None of this started recently. The preservation community grew out of the same file-sharing and emulation underground that most people either forgot about or never knew existed in the first place.
If you were in the MAME scene at any point in the 2000s, you knew Pleasuredome. It launched in 2004 as the home of the MameFTPGroup, one of the first communities to use BitTorrent specifically for distributing MAME ROM sets and CHD files. Losslessly compressed, properly verified, maintained with an almost obsessive attention to quality. The staff controlled what got distributed and how, not because they were power tripping, but because they’d seen what happened when preservation archives got commercialized by people selling burned DVDs of other people’s volunteer work. They ran that operation for 17 years before shutting down with a “game over” message on their site. Seventeen years of volunteer infrastructure. Free. No ads. No paywalls. Just people who gave a damn.
Underground Gamer was the other critical hub, a private torrent tracker where eXoDOS itself was born. The first versions of the project were built and shared through UG’s forums. And when Underground Gamer disappeared one day (deleted entirely: the tracker, the forums, all of it, just gone) and it nearly killed the eXoDOS project. eXo walked away for a full year after that. The burnout was real. Losing the community that had been the project’s home from the beginning will do that.
Then there are the groups most people never hear about but whose work underpins everything. No-Intro maintains the strictest verification standard for cartridge-based game dumps: verified, unmodified, one correct copy per title. Redump does the same for disc-based media, preserving complete raw sector data from CDs and DVDs. TOSEC fills in the gaps for systems and formats the other two don’t cover. All three produce .dat files that ROM managers use to verify whether what you’ve got is a clean preservation copy or something corrupted or modified. Without their work, game preservation would just be a pile of unverified files and no way to tell the good from the bad.
MAME has been running since February 5, 1997. I have to stop on that for a second. Twenty-nine years. The project has documented over 32,000 systems and made 10,000+ titles playable. Their mission statement puts it plainly: “MAME prevents this important ‘vintage’ software from being lost and forgotten. This is achieved by documenting the hardware and how it functions.” They’re not just making old arcade games playable; they’re reverse-engineering the actual hardware behavior, chip by chip, so that the knowledge of how these machines worked can never be lost even after every original board has failed.
And Archive.org hosts roughly 15,000 PC games and over 6,900 MS-DOS titles that you can play in your browser right now through their emulation system. They’ve also been hosting eXoDOS v5 as a distributed backup. The Internet Archive is under legal pressure at the moment from publisher lawsuits over their book lending program, which is a whole separate conversation, but their role in game preservation has been enormous.
Every single thing I just described is volunteer work. All of it. Nobody drew a salary. Nobody signed a contract. People just decided this history mattered and showed up.
eXo: It Started With Discworld
In 2007, someone wanted to play Discworld.
That’s really how it started. A guy known to the community as eXo fired up DOSBox one night to get Discworld running. It worked. So he figured he’d get all of his DOS adventure games running while he was at it. And then all of his DOS games. And then, somewhere in there, “all of my DOS games” turned into “all of the DOS games.”
Kotaku ran a feature on him in 2019. The line that stuck with me: eXo estimates he’s put in “thousands of hours’ worth of work, with the months before a new release easily crossing the line from ‘hobby’ to ‘full-time workload.'” But I actually heard the origin story directly from him three years before that piece ran. I reached out to eXo via email in August 2016 and asked about the Win3x and Win9x collections. I was trying to find if they existed and where I could grab them. eXo wrote back right away. At that point the project had received exactly three donations total. His explanation for why he built it was straightforward: “I noticed that the number of folks proficient enough with dos/dosbox to get these games running were tapering off. I figured the best way to encourage preservation of DOS games was to actually make them usable. And that is how eXoDOS got started.” No press release. No mission statement. Just a guy who saw something disappearing and decided to fix it.
The first public release, eXoDOS Vol. I: Adventure, dropped on the Underground Gamer forums in early 2008 with about 650 titles. That was eighteen years ago.
Today the scope of the project is staggering. On the Windows PC side alone, he maintains eXoDOS, a complete DOS archive, eXoWin3x for Windows 3.x, eXoWin9x for Windows 95 and 98, and eXoScummVM, a specialized ScummVM collection with tweaks and improvements on playability. There’s also eXoDreaMMachine, eXoDemoscene, an Interactive Fiction collection, a Retro Legacy Project, and platform-specific archives for Apple IIGS and Linux. What was once a pet project to get every DOS game working for a single enthusiast has turned into a massive digital gaming archive that on the surface might seem like just another collection, but on the inside is a complete historical record of all the original material.
On the PC side alone, the numbers tell their own story. eXoDOS currently stands as the top tier collection at 7,600 DOS games as of version 6.04. Every single one of these games is fully configured and working. The Internet Archive’s entire MS-DOS collection sits at 6,900 titles, and they are mostly playable through your browser. 2,500 of these titles were literally pulled from eXo’s collection in 2019. Push forward in time and you also have eXoWin3x with 1,138 Win 3.x titles, eXoScummVM with 671 titles, and the most recent 2025 release of eXoWin9x (the one I was super excited about) which has around 662 games from the Win 95 and 98 era with all the gpu fixes to play them. We are talking a complete hand configured curated collection consisting of over 10,000 PC titles from practically a single passionate volunteer.
The first time I ran the eXoDOS collection, it was running on MEAGRE before it was converted to LaunchBox. I was blown away by the time and care put into each title, every single manual and extras that you would find inside a big box when you bought it. This was something more than your standard game preservation site. This was a complete nostalgic experience for everyone that can no longer get their hands on the physical copies due to rising prices of collectibles.
That’s just the PC side. Demoscene productions, interactive fiction, Apple IIGS, the DreaMMachine project: those are entirely separate collections. What eXo actually maintains is a lot bigger than those numbers suggest.
The full eXoDOS collection alone weighs in at 637 GB with an optional 114 GB media pack: magazines, books, catalogs, soundtracks, even documentaries from the era. There’s a 5 GB lite version that downloads individual games on demand if you don’t have the storage for the full archive. eXoWin9x was released in November 2025, preserving the Windows 95/98 era including Windows versions of Doom II, MechWarrior 2, Myst, the original Diablo, Quake and more. Volume 2 is planned to cover titles from 1997-2000.
But the raw number isn’t what gets me. What gets me is that all of them work.
What “Preservation Through Playability” Actually Means
eXo’s stated principle (and he’s been consistent about this across every interview and every project FAQ) is “preservation through playability.” The logic goes like this: make the games easy to run, package them so anyone can access them without technical knowledge, and distribute them as widely as possible. The more people who have the collection, the harder it becomes to lose. Put the archive on ten thousand hard drives around the world and you’ve built something that can survive the death of any single server, any single platform, any single organization.
This isn’t piracy rebranded. This is the same principle behind every serious preservation effort that’s ever worked. The Library of Congress doesn’t keep one copy of anything. They keep multiple copies distributed across locations because redundancy IS preservation. eXo is doing the same thing, just without the federal budget.
The challenge he took on is genuinely insane when you think about what PC preservation actually requires compared to console work. Console games targeted standardized hardware: one CPU, one GPU, one sound chip, known specifications. DOS games? Hundreds of sound card configurations. Dozens of display modes. Memory managers fighting with each other. TSR programs eating conventional memory. IRQ conflicts between your Sound Blaster and your mouse driver. Anyone who lived through that era remembers the autoexec.bat and config.sys ritual of trying to get enough free conventional memory to actually launch a game.
Getting 7,666 of those games running correctly on modern hardware through emulation means solving thousands of individual configuration problems, one at a time, each with its own quirks, its own hardware expectations, its own ways of breaking.
You Download It and It Works
Every eXo collection comes fully configured with emulators and per-game settings tuned for the best possible experiences and options. All of this runs through LaunchBox which simplifies an ever growing collection into one organization tool. You get box art, metadata, 30-second preview videos, and a clean browsable library out of the box. If you want the couch experience, BigBox mode turns it into something that looks genuinely polished on a TV.
You download it, extract it, and launch. That’s the whole process.
No DOSBox knowledge required. No config file editing. No hunting through forums for the right settings for a specific game. I’ve messed with DOSBox configurations enough over the years to know how much work goes into getting even one game dialed in properly. eXo did that for seven thousand of them. For free.
Metal Jesus Rocks (nearly a million subscribers, one of the biggest retro gaming channels on YouTube) has covered eXoDOS twice, and both times as a genuine recommendation, not a curiosity piece. His take: “like a digital Noah’s Ark for DOS games, meticulously rescuing and preserving thousands of PC classics from the treacherous seas of obsolescence.” That kind of third-party validation doesn’t come from marketing; it comes from someone being genuinely impressed.
The community behind the project runs 10 to 20 volunteers at any given time. None of them were assigned, none of them get paid. The Discord (running since 2018) is where most of the actual coordination happens. These are people doing the work because they think it matters. That’s the whole pitch. That’s the entire business model.
Why Commercial and Institutional Preservation Always Falls Short
The game industry has had decades to solve this problem and hasn’t. The reasons are structural, not a matter of will or resources.
Commercial platforms operate under licensing agreements. To preserve a game you need rights from whoever holds the copyright, assuming that rights holder still exists as a company, is reachable, and is willing to negotiate a deal that makes economic sense. For the biggest titles it works. For the 87% that have no commercial availability, it never adds up. Nobody is signing a licensing deal for a game that 8,000 people worldwide want to play. The economics don’t justify it, so the game doesn’t get preserved. Commercial preservation has a hard economic ceiling baked into the model.
Public institutions tried to work around that ceiling from a different angle. The Video Game History Foundation spent three years, from 2021 through the 2024 triennial review, petitioning the US Copyright Office for a DMCA exemption that would allow libraries and archives to provide remote digital access to out-of-print games. The argument was reasonable: the game can’t be bought anywhere, a library has a preserved copy, why not let them lend it digitally the way they lend books?
In October 2024, the Copyright Office rejected the petition. They sided with the ESA (the Entertainment Software Association, the game industry’s own trade group), arguing that remote access could harm the market for commercial re-releases. The same re-releases that don’t exist for 87% of the catalog. Libraries can provide on-site access only. You have to physically travel to the archive to play a DOS game through an official channel in 2026. Institutional preservation has a legal ceiling, and the industry’s own lobby group helped build it.
So commercial preservation is constrained by economics. Institutional preservation is constrained by law. Neither path covers the full scope of what needs to exist.
The community doesn’t have either constraint. A volunteer with a hard drive and time doesn’t need a licensing deal. They don’t need a DMCA exemption or a publisher’s sign-off. They preserve what they decide is worth preserving, which turns out to be everything. That’s why the collections we’ve been talking about exist at the scale they do, and why no corporate equivalent comes close. It’s not a question of resources or intention. It’s a question of structural freedom, and the community has it in a way that commercial entities and public institutions simply don’t.
This Isn’t About Missing the Good Old Days
I can already hear the response forming in some people’s heads. “Okay, old guy misses old games.” That’s not what this is.
Think about how many people play video games today versus how many go to the movies or buy albums. Games overtook both of those a long time ago as the thing most people actually spend their free time on. And the stuff being built today (the design language, the storytelling approaches, even the business models) all traces back to decisions made in those early PC games. Id Software figured out the first-person shooter. Sierra figured out narrative adventure gaming. Maxis figured out simulation. Those weren’t just products, they were blueprints. Losing 87% of that record is like torching most of a library and shrugging because the top shelf is still intact.
The people behind eXoDOS, MAME, No-Intro, Pleasuredome, Archive.org’s gaming collections: they are doing cultural heritage work on the same level as film preservation or music archival. Unpaid. Under legal threat. With no institutional support from the industry whose history they’re saving. Frank Cifaldi at the Video Game History Foundation has been fighting the DMCA battle with limited resources for years. The MAME project has been running for nearly three decades on volunteer energy alone.
If you grew up configuring autoexec.bat to play Wing Commander, or spending an entire Saturday with a Sierra adventure game, hoping your parents wouldn’t notice you called a 900 number for hints due to moon logic. Those experiences exist in a playable form today because someone volunteered to make sure they survived. Not because a corporation decided they were worth maintaining. A person decided. And then they spent 18 years doing the work.
Getting Involved
I don’t love wrapping articles up with a numbered checklist because it always feels like a college essay conclusion, but in this case the practical steps actually matter.
Go to retro-exo.com. Download whatever collection catches your eye. The DOS stuff alone will keep you busy for the rest of your life, but the Win9x collection just dropped in November and it’s worth grabbing even if you only play Myst and MechWarrior 2 out of it. Every copy on another hard drive makes the whole archive harder to lose; that’s the point.
The eXo Discord is worth joining. Same with the MAME community spaces. I know “join the Discord” sounds like every other call to action on the internet, but these are genuinely different; people actually help each other and the conversations go somewhere. If you’ve got some money to throw at the problem, the Video Game History Foundation could use it. They launched a digital library in early 2025 with over 30,000 archived files and they’ve been fighting the DMCA fight basically on fumes. The Software Preservation Network does the institutional coordination side and they need support too.
The other thing (and this one costs nothing) is just bringing it up. I’ve dropped the 87% number in conversations with people who’ve been gaming for 30 years and they had no idea. They figured somebody was taking care of it. Some official body, some industry group, some well-funded archive. Nobody is. Or rather, the people who are taking care of it are doing it out of their own pockets and their own time, and the official channels have spent more energy blocking preservation than supporting it.
eXo wanted to play Discworld one night in 2007. Nineteen years later, there are over ten thousand PC games preserved and playable because of that one decision. I don’t know what else to call that except a legacy. And it’s one the rest of the industry should be deeply uncomfortable about not having built themselves.
Sources:
Published on Kyle’s Geek Den | February 2026
