IRC history is messy in exactly the way real Internet history usually is: university machines, borrowed code, network splits, arguments, improvised fixes, and people naming infrastructure after whatever was happening around them.
Read Daniel Stenberg’s History of IRC
Daniel Stenberg’s page pulls that history into one long chronological account. He says he gathered information from multiple sources to verify stories and dates, and the page remains open to corrections through email, issues, and pull requests.
The story begins in Finland in summer 1988.
Jarkko “WiZ” Oikarinen wrote the first IRC client and server while working at the University of Oulu. The first server was tolsun.oulu.fi. Other universities began running servers, and by November 1988 IRC had spread across the Internet.
The network immediately became a network problem
By the middle of 1989, Stenberg’s history records about 40 servers worldwide. Then came disagreements, experiments, splits, reconnects, and the recurring discovery that letting humans share a network generates administrative problems almost immediately.
The page walks through EFnet, early splinter networks, Undernet, RFC 1459, and DALnet.
Some details are especially good because they explain why features appeared.
Undernet worked on timestamps, routing changes, and channel services partly because netsplits and channel takeovers were a recurring headache. DALnet pushed further into nickname and channel ownership and user protections.
Those features were not theoretical protocol decoration. People were annoying one another at scale, so somebody wrote software.
Stenberg also records IRC’s use during the Gulf War in early 1991, when live reports drew more than 300 concurrent users. That number looks tiny in 2026 and enormous when you remember what a distributed real-time Internet chat system looked like at the time.
For modern network builders, the history is useful because it shows protocol development under social pressure. Identity, moderation, routing, naming, abuse, and governance problems did not suddenly appear with social media.
We have been arguing with strangers through network software for a very long time.
This page is one of 1,967 sites in CacheRat’s broader Ancient Web research corpus.
Explore the 1,967 Ancient Web Domains research list
If you have ever typed /join, watched a netsplit, or wondered why IRC has so many oddly named historical networks, Stenberg’s account is worth the detour.