Tag: irc

Posted on by

Ancient Web: Floppy Museum Runs a Web Server, IRC and FTP From a Floppy Disk

The best sentence on floppy.museum is also the project specification: “Obsolete hardware running obsolete software documenting obsolete media.”

Visit floppy.museum

The project began after its builder encountered a floppy-disk museum and decided the obvious next step was to make the museum itself run from a floppy.

The result is a bootable 1.44 MB disk containing MS-DOS 6.22 components, Norton DOS, networking tools, mTCP’s HTTPServ, compressed site files and configuration needed to turn vintage x86 hardware into an actual network server.

It got worse from there.

A 286 has services now

The system copies files into a RAM disk, launches DESQView and serves the site over HTTP. It also runs FTP.

Then an IRC server was added.

Visitors can point an IRC client at floppy.museum:6667. The site documents a clustered configuration using vintage systems and even experiments with Telnet-accessible message boards. Modern assumptions about what constitutes “enough computer” are politely ignored.

The software list is a miniature archaeology dig by itself: MS-DOS 6.22, NDOS, QEXT, QRAM, DESQView, packet drivers, UPX, mTCP, ngIRCd and assorted utilities squeezed around brutal memory limits. The hardware target includes 16-bit CPUs, and the documentation talks openly about the compromises required to make multitasking and networking happen on machines that were never intended to host a public Internet service in 2026.

That makes floppy.museum more than a retro display.

It is a live engineering demonstration.

A modern museum page can explain floppy disks. This one demonstrates the mindset of the era by forcing software to live inside the constraints again.

This site was found in CacheRat’s broader Ancient Web research corpus.

Explore the 1,967 Ancient Web Domains research list

It would have been easier to put the museum on a Raspberry Pi. That is precisely why doing it this way matters.

Return to floppy.museum

Posted on by

Ancient Web: Daniel Stenberg’s History of IRC

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.

Read the full History of IRC