The .RAR Monks: Inside the Secret Vaults Where 2001 Never Ended

There are people right now, in 2024, maintaining private IRC bots loaded with scene releases from 2003. They're not pirates. They're not nostalgists exactly. They're something harder to categorize: the unpaid, unacknowledged archivists of a digital civilization that the legitimate world refuses to mourn.

We reached out to several of them. Most didn't respond. One replied in IRC, naturally, through a bouncer, from a nick we promised not to print. What followed was a three-week conversation conducted entirely in a channel with one other user — a bot named something like [VAULT]-EU-1 — that turned out to be one of the more illuminating interviews we've conducted about the state of digital preservation.

What's Actually In There

The popular imagination of a warez hoarder is a guy with a basement full of burned DVDs and a persecution complex. The reality, at least among the serious collectors, is considerably more organized and considerably more depressing in its implications for legitimate archival institutions.

The collections we were shown — or rather, described, since nobody was handing over directory listings to a website — span multiple terabytes of material organized in strict Scene naming conventions. Quake.III.Arena-RAZOR1911. Photoshop.6.0-DEViANCE. Software titles that haven't been commercially available for fifteen years. Games that never got GOG releases because the original publishers dissolved in the dot-com crash and nobody knows who owns the IP. Demos. Trainers. Cracks. .NFO files by the tens of thousands, which are — and we've covered this before — genuinely the most detailed contemporaneous record of what software existed, when it was cracked, and who did it.

But it's not just software. There are music collections preserved in the original MP3 encoding of the era — 128kbps rips from CDs that were themselves ripped in 1999, complete with the ID3 tags that somebody hand-typed in Winamp while waiting for a 56k download to finish. There are early web pages, mirrored in the mid-90s by people running HTTrack on a whim, that the Wayback Machine never got to. There are IRC logs from channels that DalNet nuked when it went through its periodic server crashes.

The Preservation Paradox

Here's where it gets philosophically uncomfortable, and we're going to sit with that discomfort for a minute.

The Internet Archive is a 501(c)(3) nonprofit that has been fighting in federal court for the right to lend digital books the same way a library lends physical ones. It is, by any reasonable measure, one of the most important cultural institutions operating in the United States. It is also not preserving most of what these collectors have.

The reason is obvious: legal exposure. You cannot accept a 40-terabyte donation of cracked commercial software from 2001, catalog it, and serve it to researchers without triggering every intellectual property tripwire simultaneously. So the Archive focuses on what it can legally touch, which means entire categories of early digital culture — the Scene, the warez ecosystem, the software that defined what computers actually felt like to use in the 1990s — exist in exactly one place: the hard drives of people who won't tell you their real names.

Our contact, who we'll call Splicer because he said we could call him Splicer, has been maintaining his collection since approximately 2000. He estimates he's lost maybe 15% of it over the years to drive failures, and he's lost sleep over every single percentage point. "The stuff that's gone is gone," he told us. "There's no other copy. That's not an exaggeration. I've checked."

The XDCC Pipeline That Time Forgot

XDCC — Xabi's DCC Script, later backronymed to various things depending on who you asked — was the distribution mechanism of choice for serious Scene material on IRC through the early 2000s. You'd find a bot in a channel, type !list, get back a numbered menu of available files, type !get 47, and the bot would initiate a DCC transfer directly to your client. Bypassed the server entirely. Fast, relatively anonymous, elegant in its simplicity.

Some of these bots are still running. Not many, but some. They sit in channels on obscure networks — not the big three, not Libera or OFTC — on servers maintained by people who've been running IRC infrastructure since before broadband was common. The bots still respond to !list. The queues are still full. The files are still there.

Splicer runs two of them. He gets maybe a dozen requests a month. Sometimes it's researchers. Sometimes it's people who are trying to recover software they remember from childhood. Sometimes it's developers trying to understand how a particular piece of software worked before the source code was lost. "I don't ask," he said. "I just serve."

Intentional Neglect as Conservation Strategy

The strangest thing about talking to serious digital hoarders is that many of them have concluded, through hard experience, that the best preservation strategy is deliberate obscurity. Don't publicize the collection. Don't make it too easy to find. Don't attract attention from rights holders or, worse, from the kind of people who would "help" by uploading everything to a public torrent site where it would immediately get a DMCA takedown.

This is, if you squint at it, basically how medieval monasteries worked. The monks weren't trying to make ancient texts widely available. They were trying to keep them from being destroyed. Distribution was a secondary concern. Survival was the primary one.

Splicer doesn't love the comparison, but he doesn't reject it either. "I think about it like," he said, and then paused for a long time in that IRC way where you can't tell if someone is thinking or has just walked away from their keyboard, "I think about it like I'm holding something in trust. Not for anyone specific. Just for later."

Later. It's a strange word to use about files that already exist. But in the context of digital preservation — where "later" often means "after the last person who remembers this is dead" — it makes a terrible kind of sense.

The .RAR archives sit on spinning rust in a datacenter somewhere. The XDCC bot idles. The queue fills up with files that predate the smartphones of the people who might someday want them. Nobody is coming to catalog this. Nobody is writing the grant proposal.

But the bot is still serving. !list still works. The files are still there.

For now.