Still Serving: The XDCC Bots That Refused to Die and Are Quietly Winning

Type !list in the right IRC channel and something remarkable happens. A bot — probably named something like [XDCCBot]|Anime-HD or CrAzY-SeRvEr|Pack1 — responds within seconds. It dumps a text wall of numbered file packages into your query window. Pack numbers. File sizes. Transfer speeds. A queue counter. The whole apparatus, completely functional, completely automated, running on software that was already considered legacy infrastructure when The Matrix came out.

This is not a museum exhibit. This is happening right now.

A Brief Primer for the Uninitiated

XDCC — Xabi DCC or eXtended DCC depending on who you ask and how much they care about acronym accuracy — started as a Tcl script for ircII written in the mid-nineties. The concept was almost absurdly simple: you connected a bot to IRC, the bot sat in a channel, and users could request file transfers directly from the bot using DCC (Direct Client-to-Client) protocol. No web interface. No torrent swarm. No peer network. Just a bot, a hard drive, and a direct TCP connection between you and the machine.

For the warez and early anime fansub scene, this was transformative. A well-seeded XDCC bot on a server with a fat pipe could hammer files at you faster than anything else available in the late nineties. The interface was primitive — you literally typed commands into a chat window — but the transfer speeds were real and the reliability was genuine. If the bot was up, the bot served. No tracker going down. No swarm drying up. The bot either had the file or it didn't.

The Anime Underground Kept the Lights On

Warez moved on. The warez scene migrated to private FTP, then Usenet, then BitTorrent, and XDCC became a footnote in most histories of internet file sharing. But the anime fansubbing community — a deeply dedicated cohort of people who were translating and distributing Japanese animation before any of it was legally available in the US — found XDCC to be exactly what they needed and never left.

The logic made sense. Fansubbing groups released episodically. You needed your audience to get the new episode of whatever series was airing in Japan, ideally within hours of the raw video being available. BitTorrent worked, but swarms for older episodes dried up fast. Usenet had retention limits. XDCC bots, maintained by dedicated packagers, would hold every episode of a series indefinitely and serve them on demand. For archival purposes, for catching up, for the person who discovered a series three years after it aired, XDCC was simply better.

So the bots stayed. And they're still there.

What Does the Infrastructure Actually Look Like in 2024

This is where it gets genuinely strange. Some of these bots are running on hardware that has been continuously online for over a decade. We're talking about machines that have survived multiple generations of hard drives through careful maintenance, boxes that have outlasted the ISPs that originally hosted them by migrating through a succession of data center arrangements held together by personal relationships and the occasional PayPal donation.

The software running these bots — primarily iroffer, an open-source XDCC bot program — hasn't had a major release in years. It doesn't need one. It does exactly what it was designed to do. The people maintaining these systems aren't chasing updates. They're running a service that works, and working is sufficient.

Channels on Rizon — currently the most active IRC network for this community — host bots with pack lists numbering in the tens of thousands. Individual bots advertising 50,000+ packs are not unusual. The combined storage represented by the major XDCC networks in the anime space alone is measured in petabytes. This is not a rounding error. This is a significant chunk of digital preservation infrastructure that runs entirely on volunteer labor and nostalgia-adjacent dedication.

The People Still Running This Thing

Spend enough time in these channels and you start to notice the operators. They're not teenagers. The people maintaining the major XDCC bots are, overwhelmingly, adults in their thirties and forties who got into this in the early 2000s and simply never stopped. They have jobs. Some of them have kids. They also have a bot that's been serving files since the Bush administration and they're not about to shut it down because Discord exists.

The motivations are varied but a common thread emerges in the rare moments when these people talk about what they do: preservation. The fansubbing groups that produced these files are mostly defunct. The original distribution channels are gone. The XDCC bots, in many cases, represent the only reliable source for versions of media that simply don't exist anywhere else — fan translations that were never superseded, encodes made with codecs that modern tools handle differently, cultural artifacts from a specific moment in the history of a medium.

These bot operators are, whether they'd use the word or not, archivists.

The Ghost Network Metric

Here's a number worth sitting with: on any given day, Rizon — one IRC network, not even the only one hosting XDCC activity — processes millions of file transfers. The bandwidth involved would be notable for a mid-sized commercial file hosting service. It happens invisibly, in channels that don't appear in any tech press coverage, on a protocol that most internet infrastructure tutorials don't even mention anymore.

The people using these bots aren't using them because they don't know about modern alternatives. They know about modern alternatives. They use XDCC because it works, because the content they want is there and nowhere else, and because there's a certain satisfaction in a system that does exactly one thing and does it without asking for your email address, showing you an ad, or recommending content you didn't request.

Discord is newer. Discord is louder. Discord has GIFs and voice channels and a mobile app and a valuation that has been, at various points, genuinely absurd.

The XDCC bot doesn't care. It's in its channel. It has the files. Type !list and it will tell you what it has.

It's been there longer than Discord has existed. At current trajectory, it may be there after Discord is gone.