Category: Ancient Web

The most ancient times of the internet: 90s, early 2000s.

Posted on by

Ancient Web: A Clementi Piano Page Kept Working After the Site Died

A dead or unreliable webpage can still be alive in the research record. The Muzio Clementi Society’s page on Clementi pianos is a good example.

Try the original Clementi Pianos page

The page itself is difficult to reach reliably now, but references to it are scattered across more than a decade of writing about early pianos.

That makes sense. Muzio Clementi was not only a composer and pianist. He was also deeply involved in music publishing and piano manufacturing in England. His business activity places him directly inside the technological history of the instrument, not merely its repertoire.

The Clementi Society page appears to have served as a compact reference on Clementi-built instruments and piano mechanisms. In 2009, piano technicians were already passing the URL around while trying to identify and understand an old piano. A 2016 article about Clementi listed the page among its references.

More interestingly, a 2025 academic paper on early square pianos still cited the same URL while discussing differences between historical and modern piano mechanisms.

That is a long afterlife for one ordinary HTML page.

Links become archaeological evidence

This is why old-Web research cannot stop at “the page is down.”

A vanished page may leave a citation trail in mailing lists, academic PDFs, forum archives, bibliographies, blogs, and later reference works. Those links tell us what the page contained, who considered it useful, and how long it remained part of a subject’s research infrastructure.

They also expose one of the Web’s preservation weaknesses. Scholars can cite a useful specialist page for years, yet the page may still depend on one aging domain and one volunteer organization continuing to pay a hosting bill.

Clementi himself is an especially appropriate subject for this problem. His work sits at the intersection of composition, performance, publishing, manufacturing, and mechanical improvement. Losing specialist pages about his instruments would erase exactly the sort of detail general biographies tend to compress into a sentence.

For webmasters, the lesson is straightforward: a small reference page can outlive its design, its navigation, and even its server. What matters is whether other people found it specific enough to cite.

CacheRat’s 1,967 Ancient Web Domains research list contains many pages whose value survives partly through these external traces.

The Clementi Pianos page is one of them. The server may be unreliable. The citation network is not.

Try the original Clementi Pianos page

Posted on by

Ancient Web: JapanHemp Preserved a 1990s History of Hemp in Japanese Culture

The old JapanHemp site is unreliable today, but the material behind its /uncleweed/history.htm path did not disappear with the server. It leads back to a substantial late-1990s research project on the history of hemp in Japan by writer Dave Olson.

Try the original JapanHemp history page

Olson’s surviving author archive identifies “Hempen Culture in Japan” as material published at JapanHemp.org, formerly Taima.org. He says the research first appeared in the Journal of the International Hemp Association in 1997 and was later published by Cannabis Culture in 1998.

The JapanHemp version was not a single short essay. Its table of contents spread the subject across pages dealing with hemp’s arrival in Japan, Japanese history, religion, Zen and martial arts, language, regional agriculture, imperial clothing, paper and other arts, food and medicine, law, strains, contemporary business, and a bibliography.

That breadth is the reason the old site keeps showing up in later citations.

Useful source, strong point of view

This is also a good example of why old Web research should be read carefully rather than worshipped merely because it survived.

Olson’s work gathered historical claims, photographs, references, interviews, cultural material, and contemporary hemp advocacy into one project. Some of its arguments are explicitly political. The 1997 article discusses postwar cannabis restrictions, American influence, Japanese hemp activism, and hopes for renewed industrial cultivation. Those are the author’s framing and conclusions, not neutral facts CacheRat needs to adopt wholesale.

Other sections document much older material: textile use, paper making, Shinto associations, agricultural practices, hemp seed in food, regional production, and specialized traditional uses. Olson supplied citations throughout the published article, giving later researchers something to trace rather than simply asking them to trust a webpage.

That traceability matters because JapanHemp itself has become unstable. Olson’s own archive now preserves the article’s publication history and the old JapanHemp section structure, while later books, papers, and reference works continue to cite pages from the domain.

In other words, the research escaped the website.

That is one of the healthiest outcomes for digital preservation. A useful page should be copied, cited, quoted responsibly, mirrored, republished by its author, and referenced elsewhere. The Web is much more durable when knowledge does not depend on one Apache process continuing to answer on the same hostname forever.

The subject here is historical and cultural; this article is not advice about obtaining, growing, or using cannabis. What makes the site worth documenting is the research trail it created and the way that trail remains reconstructable despite the original host’s decline.

CacheRat’s 1,967 Ancient Web Domains research list includes many sources whose value now lies partly in following those trails outward.

JapanHemp is a good specimen: the server got flaky. The citations did their job.

Check the original JapanHemp history URL

Posted on by

Ancient Web: Autism News Network Left a Written Record of the GeoCities Shutdown

Most accounts of the GeoCities shutdown describe millions of pages disappearing in the abstract. Autism News Network preserves something more useful: a webmaster explaining to actual readers, in October 2009, what the shutdown meant for them.

Visit Autism News Network

The site began on GeoCities at geocities.com/fishstep/ and served as a volunteer-run information network for parents of children with autism and PDD. Its surviving homepage says the project originally existed as a physical monthly newsletter that had to be typed, copied, stuffed into envelopes, stamped, and mailed. Eventually the newsletter became a webpage and email network.

Then Yahoo announced that GeoCities would close on October 26, 2009.

The site’s response remains sitting at the top of the homepage.

Readers were told to change their bookmarks. Other webmasters were asked to replace old GeoCities links. The maintainers warned that moving every page would take time and asked visitors to report anything missing. They also explained why they had resisted moving earlier: people had bookmarked the old address, other sites linked directly to individual pages, and relocating a large hand-built site was a substantial job.

That is link rot being documented before the links actually rotted.

The community came before the website

Autism News Network was explicitly noncommercial. The page says it accepted neither advertising revenue nor contributions and depended on volunteer work from parents. An email list connected hundreds of families, while the website collected articles, legislation, school and IEP information, meeting notices, newsletters, contacts, and a parent packet.

The surviving newsletter archive includes material from 2001 through 2003, while other pages discuss Illinois legislation and educational advocacy from later in the decade.

Some medical terminology, treatment discussions, external links, and claims on the site reflect the period in which they were written. They should be read as historical material, not as current medical guidance. The site’s real value in 2026 is as a record of how parents organized, shared information, and navigated education and disability systems on the early social Web.

It also captures the transition from one Internet model to another. The homepage itself reflects on how the project moved from paper newsletter to webpage and notes that a blog or social site might make more sense by 2009.

That sentence is almost a timestamp for the Web.

A volunteer mailing list had become a GeoCities information hub. GeoCities was dying. Social platforms were becoming the obvious next destination. Yet the maintainers moved the material to an independent domain instead, which is why much of it can still be inspected today.

The larger 1,967 Ancient Web Domains research list contains many former GeoCities sites with much less fortunate endings.

Autism News Network survived because somebody did the annoying work of moving the boxes before Yahoo demolished the building.

Visit Autism News Network

Posted on by

Ancient Web: DeadMalls.com Turned Retail Collapse Into a Community Archive

Long before “dead malls” became a YouTube genre, DeadMalls.com was already collecting the evidence.

Visit DeadMalls.com

The site says it has been documenting the lifecycle of American shopping malls since 2000. Its method is wonderfully old-Web: people send in stories, photographs, corrections, and updates, and the site turns those contributions into a growing archive of retail places that changed, emptied out, were redeveloped, or disappeared entirely.

The homepage still describes the project in plain language. Read the dead mall stories. Look at the pictures. Submit a mall. Send corrections. Learn the vocabulary. Browse links. There is even a dictionary for the jargon used by people who study failing retail centers.

That community model matters because malls are difficult things to preserve. A building can survive while the original tenant mix disappears. A shopping center can be renovated until almost none of its earlier architecture remains. A mall can be demolished and replaced by apartments, warehouses, medical offices, or nothing at all. Local newspapers may cover the closing, but they rarely preserve the ordinary interior details people actually remember.

A fan archive can.

DeadMalls.com captures the less formal history: fountains, food courts, anchor stores, odd renovations, empty corridors, local nicknames, stores that hung on too long, and the point where a once-busy mall crossed over into something eerie.

The site itself now belongs to the history it documents. It launched while many traditional malls were still operating and before smartphones made photographing every dying shopping center trivial. Its early pages therefore preserve not only old malls, but an older mode of online local-history collecting.

The project remains active in 2026 and still asks visitors to contribute updates. That continuity is important. A mall listed as dead in 2005 may have been demolished, revived, renamed, or repurposed since then. The archive becomes more useful when the historical layer is preserved rather than overwritten.

For webmasters, DeadMalls.com is a strong example of a site becoming valuable because it gives people a place to deposit memory that otherwise has nowhere obvious to go.

CacheRat’s Ancient Web research list includes many sites built around similarly narrow forms of collective memory.

DeadMalls.com understood early that disappearing places deserve URLs before they disappear completely.

Browse DeadMalls.com

Posted on by

Ancient Web: psDooM Let Sysadmins Shoot Processes

Somebody looked at ps, renice, and kill and decided the obvious missing feature was a shotgun.

Visit the psDooM screenshots page

psDooM is a Unix process monitor built on Doom. Running processes appear inside the game as monsters labeled with a process ID and part of the process name.

The joke goes considerably further than a visual skin.

The program periodically checks the machine’s process table and spawns or removes monsters as processes appear and disappear. Damage to a process-monster corresponds to changing that process’s priority with renice. Killing the monster can kill the associated Unix process.

This is therefore one of those rare pieces of software where “shooting the runaway process” can be technically accurate documentation.

The project grew from the release of Doom’s source code in the late 1990s. The psDooM site traces its lineage through XDoom and a University of New Mexico proof of concept called “Doom as a tool for system administration.”

A GUI nobody asked for

A conventional process monitor turns machine state into rows, numbers, percentages, and sortable columns.

psDooM turns the same abstraction into physical space.

Processes occupy a level. They can be approached. Their labels float in front of them. Administrative actions become game actions. The metaphor is ridiculous, but it is also understandable almost immediately to anyone who has played Doom.

That is what makes the project more interesting than a one-line programming gag.

Interface design is largely the business of choosing metaphors for invisible state. Desktop systems use files and folders. Network tools use graphs. psDooM uses demons.

The screenshots page preserves the proof that this really existed, while the main project documentation explains the process-monitoring behavior, supported Doom versions, custom levels, user filtering, and other details.

It is also a snapshot of a specific open-source moment. Once id Software released Doom’s source, programmers did not merely preserve the game. They treated it as reusable infrastructure and started asking increasingly strange questions about what else its engine could represent.

CacheRat’s 1,967 Ancient Web Domains research list includes artifacts like this because software history gets much more interesting when you keep the experiments that were never supposed to become products.

Modern process managers are cleaner, safer, and more useful.

None of them let you circle-strafe Apache.

See psDooM in action

Posted on by

Ancient Web: Our Digital Heritage Preserved Ratbag Games’ Corporate Site

Sometimes an old company website survives not because the company kept it alive, but because somebody else decided it was worth rescuing.

Visit the preserved Ratbag archive

The surviving landing page is minimal and carries a 2001–2002 Ratbag Pty Ltd copyright. That tiny fragment matters because Ratbag Games itself is long gone.

The Adelaide studio began in 1993 as Emergent Software, later became Ratbag, and built a reputation around racing games. Its titles included Powerslide, Dirt Track Racing, Dirt Track Racing: Sprint Cars, Leadfoot, Dirt Track Racing 2, Saturday Night Speedway, and World of Outlaws: Sprint Cars 2002.

Ratbag’s games used the company’s own Difference Engine and leaned hard into vehicle physics and dirt-track racing at a time when that was still a fairly specialized PC niche.

The company was acquired by Midway in 2005 and shut down not long afterward.

That is where the archive becomes more than a dead corporate page.

Corporate websites are historical documents too

Old company websites are often treated as disposable because they were promotional. But promotional pages preserve details that can disappear elsewhere: screenshots, product descriptions, staff credits, technical claims, logos, downloadable material, release-era language, and the way a company presented itself while it still existed.

Once a studio closes, that material tends to evaporate quickly.

A preserved Ratbag site therefore helps document Australian game development from the late 1990s and early 2000s. It also captures a period when independent and midsize studios commonly maintained their own small corporate sites rather than funneling everything through social platforms and publisher storefronts.

Even the bare copyright line is useful. It anchors the preserved page to the exact period when Ratbag was releasing some of its best-known racing titles.

This is one reason digital preservation cannot focus only on famous homepages and major institutions. The Web’s historical record is full of ordinary commercial sites that nobody expected to become artifacts.

CacheRat’s 1,967 Ancient Web Domains research list contains many pages like this: small corporate fossils whose companies, products, and technologies have moved on.

The Ratbag archive is not glamorous. It does not need to be. It keeps one more piece of the old games industry from becoming a sentence with no supporting evidence.

See the preserved Ratbag site

Posted on by

Ancient Web: Roger Russell Documented McIntosh From the Inside

There are corporate product histories, collector databases, and engineering memoirs. Roger Russell’s McIntosh pages manage to be all three at once.

Visit Roger Russell’s site

Russell’s McIntosh History section covers the company from its early years through decades of amplifiers, preamplifiers, tuners, receivers, loudspeakers, equalizers, car audio, home theater gear, drivers, accessories, and test equipment.

The unusual part is who assembled it.

Russell was not merely looking at McIntosh equipment from the outside. His pages discuss work he personally did, including loudspeaker development and the C26 preamplifier. That puts technical descriptions, company history, product listings, and personal recollection in the same archive.

The site also preserves the culture around the hardware.

One section covers the McIntosh Amplifier Clinics, which began in 1962. Dave O’Brien traveled to dealers and tested customers’ amplifiers in front of them, eventually testing more than a million amplifiers over roughly three decades. It was marketing, engineering demonstration, customer service, and live theater for people who cared very deeply about distortion figures.

The details collectors actually need

Russell’s product-history pages list model names, sales dates, original prices, and technical notes. Other sections collect magazine reviews, patents, advertising material, speaker-wire history, acoustic test equipment, and restoration information.

That is the level of detail that becomes more valuable as hardware ages.

A glossy retrospective can tell you that McIntosh made famous amplifiers. A specialist archive can help identify which amplifier, when it was sold, how it differed from a neighboring model, and what somebody who worked there remembers about why it existed.

The site also has the unmistakable structure of a long-lived personal Web project. Information accreted where it was useful. Pages became deep. Navigation grew around the collection instead of being redesigned every few years to satisfy a branding committee.

That is not pretty in the modern product-design sense.

It is excellent archival behavior.

CacheRat’s 1,967 Ancient Web Domains research list includes sites like this because specialist knowledge often survives on personal pages long after the people, companies, and hardware involved have moved on.

The blue meters get the attention.

The old HTML remembers why they were there.

Explore Roger Russell’s McIntosh history

Posted on by

Ancient Web: theJANG’s Old Texture Directory Points Back to Web Design in the 1990s

An old directory called /textures/ is exactly the kind of fossil you expect to find in a long-lived personal domain. On theJANG.com, it sits inside something more useful than a random bucket of background images: the surviving online history of a web designer who was building interfaces before “UX” became a department.

Visit the old theJANG texture directory

The texture directory itself is not reliably retrievable now, so there is no responsible way to invent a catalog of what used to be inside it. The surrounding domain, however, identifies its owner very clearly.

Bamidele O. Shangobunmi, known online as theJANG, says he entered the online world through bulletin-board systems in 1993 and built his first webpage in 1995. His biography explains that the JANG identity evolved through several aliases between 1993 and 1997 before settling into the name that still fronts the site.

His old resume is a miniature history of commercial Web development.

Shangobunmi describes himself as designing online user interfaces since 1993. During the late 1990s he worked on hand-built HTML, JavaScript, DHTML, cross-browser compatibility, and graphics at a time when supporting Netscape Communicator and Internet Explorer 4 was an actual job requirement rather than a historical punishment.

At Ask Jeeves from 1999 to 2001, he worked first as a senior graphic designer and then as a site engineer. His resume credits him with homepage and search-results redesign work and with developing an ultralight Ask.com homepage of less than 5 kilobytes of HTML, down from nearly 20K, while preserving the visual design.

That sentence may cause physical pain in anyone currently waiting for a 9-megabyte marketing homepage to display three paragraphs and a button.

The personal site became a career archive

The interesting thing about theJANG.com is how much of the surrounding history still survives.

There are work samples from later interface-design jobs, descriptions of browser limitations, and personal material that spans completely different interests. His Diablo II section, for example, preserves content from his active playing years of 2000–2002 and says much of it was resurrected in 2009 for old time’s sake.

That is the context that makes a path like /textures/ interesting. Texture collections were basic infrastructure for early personal-page construction. Small repeating GIFs and JPEGs could turn the browser background into marble, metal, paper, stars, stone, or something that looked like a carpet sample discovered after a chemical spill.

We cannot currently prove which textures were in this particular directory, so CacheRat is not going to pretend otherwise. What survives around it is better documented: a personal domain that traces directly through early Web graphics, hand coding, browser wars, commercial interface work, games, hobbies, and decades of online identity.

CacheRat’s 1,967 Ancient Web Domains research list contains many isolated relics. This one happens to remain attached to the person who made them.

Try the original /textures/ path on theJANG.com

Posted on by

Ancient Web: Vulkaner.no Built a Dreamweaver Nature Encyclopedia With No Off Switch

There is a specific kind of old website that begins with one interest and eventually decides to catalog the planet. Vulkaner.no appears to have reached that stage years ago and simply kept going.

Visit the Fauna and Nature section of Vulkaner.no

The page calls itself Our Beautiful World: Fauna and Nature. What follows is less a tidy navigation menu than a declaration of war on finite free time.

The site advertises more than 500 bird entries, more than 250 animal entries, more than 225 flowers, sea life, travel, Norwegian nature, African animals, Antarctic animals, Asian animals, Latin American wildlife, Kamchatka, gorillas, reindeer, pangolins, whales, dolphins, tree kangaroos, butterflies, cranes, national parks, and assorted environmental subjects.

There is a page for all 15 crane species. There are separate links for ducks in Oceania and America. There are birds from Sangihe Island, Yunnan, Thailand, Menorca, and Lake Hornborga. The Andes project is described as one of the site’s largest undertakings and contains Norwegian text alongside large amounts of photography and links to English-language material.

This is not minimalism.

It is much better than minimalism.

The software fingerprint is still visible

At the bottom, the page proudly states that it was made with Macromedia Dreamweaver.

That little line places the site in a recognizable era of personal web construction. Dreamweaver let people treat a website partly like a visual publishing project while still producing ordinary files that could be uploaded to a server and linked together however the author pleased.

The result here is dense, imperfect, and extremely browsable.

Instead of forcing everything through a modern taxonomy, the page accumulates subjects as the author’s interests expand. “The Great Pacific Garbage Patch” can sit near ducks, flowers, volcanoes, Norwegian scenery, wildlife in Kazakhstan, and a link about a giant tree in Mexico. Navigation grows organically because the site itself grew organically.

That structure would make a UX consultant nervous. It also encourages wandering.

The photography is another important piece. The page states that its pictures belong to Vulkaner.no unless otherwise noted. This is not a scraped encyclopedia assembled from stock imagery. It is an authored collection with the feel of a long-running field notebook.

For researchers, such sites can preserve observations, photographs, terminology, and references that do not always migrate cleanly into newer platforms. For webmasters, they demonstrate the durability of plain pages with ordinary links. No application framework is required to keep hundreds of subject pages discoverable.

The larger 1,967 Ancient Web Domains research list includes many specialist websites like this. They are not elegant because elegance was never the mission. The mission was apparently MORE BIRDS.

Explore Vulkaner.no’s Fauna and Nature section

Posted on by

Ancient Web: Flying Omelette Documented Donkey Kong 64’s Weirdest Glitches

Before game wikis turned every secret into a database row, fans built little shrines around the parts of a game they personally found strange enough to document. Flying Omelette’s Donkey Kong 64 Bananza is exactly that kind of page.

Visit Donkey Kong 64 Bananza

The author is not pretending Donkey Kong 64 is flawless. The introduction flatly says it is not the greatest game in the world. The point of the site is narrower and better: collect the things worth remembering.

Those include the unlockable arcade version of the original Donkey Kong, a character essay about Chunky Kong, cameos and pop-culture references, odd visual details, and a surprisingly rich collection of glitches.

One glitch can make the ending credits roll during the opening sequence. Another lets characters other than Tiny Kong visit the Banana Fairy. A boulder-reflection oddity can make Chunky appear to walk like a zombie. There are methods for swimming under DK Isle, entering a room intended only for a cutscene, and reaching what appears to be a leftover beta-testing area through manipulation of the minigame menu.

This is precisely the sort of knowledge old fan pages were good at preserving. A player noticed something weird, reproduced it, took screenshots, wrote down the steps, and uploaded the result. No engagement strategy was required. The weirdness itself justified the page.

The site also warns readers about spoilers in enormous, unmistakable language. That is another relic of the period: a fan site assuming the visitor might actually be playing the game for the first time rather than arriving from a search result twenty years later.

For preservationists, pages like this matter because they often record behavior that official manuals never mentioned and developers never intended. Glitches, unused rooms, leftover test areas, and undocumented quirks are part of software history too.

For webmasters, the lesson is equally simple. A page does not need to explain an entire franchise to be useful. It can cover one game, one obsession, or one collection of peculiarities and still remain worth finding decades later.

CacheRat’s Ancient Web research list contains plenty of exactly this kind of fan-built micro-archive.

Flying Omelette’s DK64 page survives because somebody bothered to write down the weird stuff.

Explore Donkey Kong 64 Bananza