If you were online in 1998 and wanted to download a movie, a game, or an album, you didn't go to a website. You went to IRC. Specifically, you went to channels with names like #mp3, #warez, #divx, or #0day on networks like EFnet, DALnet, or Undernet. You typed !list and waited for the bots to respond with their catalogues. Then you picked what you wanted, queued up, and waited your turn.
It was slow, unreliable, and occasionally got you K-lined from the network. It was also the most efficient file distribution system the internet had at the time.
Before Napster: IRC Was Already There
The popular narrative of internet file sharing usually starts with Napster in 1999. But IRC-based file sharing predates Napster by several years. By 1996, there were already well-established channels on EFnet dedicated to distributing pirated software, music, and movies. The tools were primitive — mostly manual DCC SEND commands — but the demand was enormous.
What made IRC attractive for file sharing was its real-time nature and its existing community infrastructure. You didn't just download files — you hung out in channels, talked to people, got recommendations, and built relationships. The social layer made IRC's file sharing stickier than any anonymous download site could be.
The other advantage was decentralization. There was no central server hosting the files. Each bot or fserve was run by an individual user on their own hardware. Taking down one bot didn't affect any of the others. This made IRC-based distribution remarkably resilient against takedown attempts — a lesson that BitTorrent would later learn and improve upon.
XDCC: The Protocol That Made It Work
XDCC — eXtended DCC — wasn't really a protocol in the formal sense. It was a convention for how bots offered and served files over IRC's DCC (Direct Client-to-Client) mechanism. An XDCC bot would sit in a channel and respond to specific commands:
/msg botname xdcc list — Request the file list
/msg botname xdcc send #3 — Request file number 3
/msg botname xdcc cancel — Cancel your current transfer
/msg botname xdcc remove #3 — Remove yourself from queue for file #3When you requested a file, the bot would either start a DCC SEND immediately (if it had free slots) or add you to a queue. Popular bots might have queues dozens deep, and you'd wait hours for your turn. The bots would announce their statistics in the channel periodically — how many files they had, total size, number of slots, current queue length.
The DCC transfer itself was a direct TCP connection between your client and the bot, bypassing the IRC server entirely. This was important for two reasons: it didn't burden the IRC server with file transfer traffic, and the transfer speeds were limited only by the network path between you and the bot, not by the IRC server's capacity.
In practice, transfers were slow by modern standards. A 700MB DivX movie might take hours on a typical late-1990s connection. But compared to the alternative — which was often nothing — it was revolutionary.
FServes: Your Own File Server on IRC
FServes (File Servers) were a different approach, primarily used in mIRC on Windows. An FServe was a script that turned your mIRC client into an interactive file server. Other users would trigger the fserve (usually by typing a specific command in the channel), and a DCC Chat session would open. Inside that session, you could navigate a directory structure, almost like a rudimentary FTP client:
Welcome to Marco's FServe!
Files: 2,847 | Size: 42.3 GB | Sends: 3/5 | Queues: 8/15
Type: dir, cd [dir], get [file]
/> dir
[Music] [Movies] [Games] [Apps]
/> cd Music
/Music> dir
[Rock] [Electronic] [Hip-Hop] [Classical]
/Music> cd Rock
/Music/Rock> dir
Pink_Floyd-Dark_Side_of_the_Moon.zip [112MB]
Led_Zeppelin-IV.zip [98MB]
/Music/Rock> get Pink_Floyd-Dark_Side_of_the_Moon.zip
Added to queue. Position: 3. Estimated wait: 45 minutes.FServes were popular because they were easy to set up — just a mIRC script and a shared directory. No special software needed. The downside was that they were less efficient than XDCC bots and relied on the user's mIRC client being online and running. They were also harder to manage at scale.
There was a whole etiquette around fserves. Channels often had ratio requirements — you had to share a certain amount of files yourself before you were allowed to download. This encouraged reciprocity and kept the ecosystem functioning. Users who only downloaded without sharing (called "leechers") were often banned from channels.
The Channels: #mp3, #warez, and the Ratio Game
The warez channels on EFnet in the late 1990s were massive — some had hundreds or even thousands of users. The bigger channels had strict rules:
- Ratio requirements: You had to maintain a certain upload-to-download ratio, usually 1:3 or better. Share three files for every one you download.
- No begging: Asking for specific files in the channel (instead of using proper XDCC or fserve commands) would get you kicked or banned.
- Quality standards: Files had to meet minimum quality requirements. A poorly ripped MP3 or a cam-recorded movie would get you banned and your bot removed.
- Speed requirements: Bot operators were expected to maintain minimum upload speeds. If your connection was too slow, your bot would be removed from the channel.
The channel operators were often remarkably well-organized. They maintained websites with rules and guidelines, ran automated systems to verify bot quality, and had hierarchical administration structures with different levels of access. It was, in many ways, a precursor to the organized administration you'd later see on torrent sites like What.CD.
The Scene: Organized, Hierarchical, Ruthless
Above the public warez channels was "the Scene" — the organized software piracy groups that produced the actual releases. Groups like Razor1911, Fairlight, SKIDROW, and RELOADED (for games), or aPC, LOL, and FLT (for applications) had their own private IRC servers and channels where releases were coordinated.
The Scene operated on a strict hierarchy. "Topsites" — private FTP servers with massive bandwidth — received releases first. From there, files trickled down through "courier" groups to mid-level distribution sites, and eventually to the public IRC channels where regular users could access them.
IRC was the communication backbone of this entire ecosystem. Release announcements were made on IRC. Coordination between groups happened on IRC. The NFO files that accompanied releases often included IRC channel information where you could find the group. Without IRC, the Scene as it existed in the late 1990s and early 2000s simply couldn't have functioned.
I want to be clear: I'm describing this as history, not endorsing it. Software piracy is illegal, and the Scene's operations caused real financial harm to developers and artists. But from a technical and organizational perspective, what they built was remarkable — a global distribution network with quality control, standards, and governance, running entirely on volunteer effort and IRC infrastructure.
The Decline: BitTorrent Changes Everything
BitTorrent, released by Bram Cohen in 2001, didn't kill IRC-based file sharing overnight. But it fundamentally changed the economics of distribution. With BitTorrent, every downloader was simultaneously an uploader — the more popular a file, the faster it downloaded. This inverted the IRC model, where popular files meant longer queues and slower access.
By 2005, BitTorrent had become the dominant method of large-file distribution, both legal and illegal. IRC-based file sharing retreated to niche communities: anime fansubbing groups, obscure music collectors, and Scene couriers who continued to use IRC for coordination even as the actual file transfers moved to other protocols.
Some XDCC channels still exist today, particularly on networks like Rizon (popular in the anime community). But they're a shadow of the massive distribution networks of the late 1990s.
Legacy
IRC's role in the history of file sharing is often overlooked. Napster gets the credit for starting the peer-to-peer revolution, but IRC-based sharing predates it by years. BitTorrent gets the credit for decentralized distribution, but IRC's bot-based model was already decentralized in the sense that no single server hosted all the files.
More importantly, the community aspects of IRC-based file sharing — the ratio systems, quality standards, and organized administration — directly influenced the private tracker communities that emerged around BitTorrent. What.CD's interview process and quality requirements? That's straight out of the IRC warez channel playbook.
The tools change. The principles — community moderation, quality control, reciprocity — persist.
Frequently Asked Questions
Is XDCC still used?
Yes, though on a much smaller scale than in the late 1990s. XDCC bots remain active on some IRC networks, particularly Rizon, which has a significant anime fansubbing community. For general file sharing, BitTorrent and direct download services have largely replaced XDCC.
What was the difference between XDCC and fserves?
XDCC bots operated automatically — you sent a command, and the bot queued and sent the file without further interaction. FServes opened an interactive DCC Chat session where you could browse directories and select files manually. XDCC was more efficient for large-scale distribution; fserves were easier to set up for individual users.
Was IRC file sharing illegal?
Distributing copyrighted material without authorization is illegal in most jurisdictions, regardless of the technology used. IRC itself is a neutral communication protocol — using it for file sharing is not inherently illegal. The legality depended entirely on what was being shared and whether the copyright holder had authorized the distribution.
Did IRC networks try to stop file sharing?
Policies varied by network. Some networks (particularly DALnet) actively tried to combat piracy channels. Others (particularly EFnet) took a hands-off approach, viewing channel content as the users' responsibility. Most networks eventually implemented measures against mass-file-sharing bots due to bandwidth and abuse concerns rather than copyright enforcement.
FAQ
Was IRC used for piracy?
Yes, extensively. In the late 1990s and 2000s, IRC was a primary distribution platform for the warez scene, using DCC transfers, XDCC bots, and organized distribution channels.
What are XDCC bots?
XDCC bots are IRC bots that serve files via DCC (Direct Client-to-Client) transfers. Users request files by sending commands in the channel, and the bot initiates a direct transfer.
Is the warez scene still active on IRC?
The scale has decreased dramatically. Modern piracy has moved to other platforms, but some groups still use IRC for coordination and pre-release communication.