Category: Ancient Web

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

Posted on by

Ancient Web: Grafman’s VR World Preserves What VRML Looked Like in 1995

In 1995, putting a three-dimensional world on a webpage was sufficiently strange that building a virtual gallery with doors between rooms counted as a useful technical experiment.

Visit Grafman’s VR World demo guide

Robert M. Free’s surviving Grafman’s VR World pages are a compact record of the period when VRML — the Virtual Reality Modeling Language — looked like one possible future of the Web.

The demo guide is especially good because it does not merely preserve screenshots. It dates the experiments, explains what each one was testing, names the browser technology involved, and occasionally records how long the thing took to build.

The first listed experiment, Grafman’s Gallery, is dated November 22, 1995 and uses VRML 1.0. Free describes it as an exercise in learning the language, built with a text editor and images from earlier projects. The world used textures, inline images, frame-targeted links, and doors connecting rooms.

A few weeks later the experiments were already getting weirder.

The future required a plugin

A December 1995 “Space” demo tested animated textures using one of Netscape Live3D’s early extensions. The page still says plainly that the demo requires the Live3D plugin — a sentence that now dates the experiment almost as effectively as the 1995 timestamp.

Another demo generates a VRML snowman with a CGI/Perl script. Visitors could choose the number of body segments while the script randomized proportions, facial parts, and hundreds of spherical snowflakes. Free notes that the snowflakes could overwhelm the virtual memory of early VRML browsers.

By 1996 the archive includes a randomly generated Valentine, a bunny avatar with algorithmically bent ears, a spinning Earth with a transparent cloud layer, articulated skating avatars, and a surprisingly practical geodesic-dome generator. That dome tool let a builder specify size, symmetry, and frequency, preview the structure, and generate a parts list for hubs, rods, and panels.

The avatar work is equally revealing. Jameraude, a roller-skating figure, could be generated at different levels of detail. Free records a high-resolution version at 78K and a low version at 14K — or about 2K compressed. File size was not an abstract optimization problem when the audience was arriving by modem.

By February 1997, Jameraude had been ported to VRML 2.0 with animation and joint ranges, using SGI’s Cosmo Player and, in another experiment, an external Java applet through the External Authoring Interface.

None of this looks like today’s commercial “metaverse” pitch. That is what makes it useful. It preserves the actual engineering questions people were asking when browser-based 3D was young: How do we animate a texture? How large can an avatar be? Can a CGI script generate a world? Can Java control it? Will the browser survive the snowflakes?

CacheRat’s 1,967 Ancient Web Domains research list is full of pages where obsolete technology remains understandable because somebody left the experiments and notes online. Grafman’s VR World is a particularly good specimen because the failed future is still documented in working HTML.

The plugins are history. The questions are recognizable.

Browse Robert M. Free’s VRML demo archive

Posted on by

Ancient Web: Andrew Turnbull Cataloged How American Traffic Signs Changed

A stop sign looks inevitable until somebody lines up the versions from 1927, 1930, 1935, 1948, and 1954 and reminds you that even the most ordinary visual language had to evolve.

Visit Andrew Turnbull’s Field Guide to American Traffic Signs

Andrew Turnbull built exactly that kind of comparison into his Field Guide to American Traffic Signs. The project focuses not merely on what American signs mean, but on how their designs changed over time.

That distinction turns a familiar subject into a surprisingly deep historical reference.

The guide is divided into regulatory, warning, and guide signs, with dates attached to designs based primarily on the Manual on Uniform Traffic Control Devices and its predecessor documents. Turnbull is careful about what those dates mean: they generally represent the earliest reference he found in the manuals, not necessarily the first time a state put a particular sign beside a road.

That bit of caution matters. Road standards are not software releases. Old designs linger, local experiments precede national standards, and replacement happens one physical sign at a time.

The road is a giant interface nobody thinks about

The regulatory-sign section makes the site’s value obvious. Stop signs are shown across multiple revisions beginning in 1927. Yield appears with 1954, 1961, and 1971 versions. Speed-limit signs, turn prohibitions, one-way signs, parking restrictions, railroad-crossing signs, and dozens of now-obsolete designs get the same treatment.

One note even points out that yellow stop signs, though rare, could still be found decades after newer standards replaced them; Turnbull cites a rusty example that survived in Two Rivers, Wisconsin until 2010.

The warning-sign pages do the same thing for curves, winding roads, junctions, railroad hazards, animals, grades, schools, construction, and the expanding vocabulary required as American roads changed. A traffic circle or roundabout warning, for example, appears much later than the oldest curve and crossroad symbols.

The site’s graphics were assembled with wonderfully ordinary tools: PC Paintbrush, IrfanView, and TrueType versions of Mike Adams’ Roadgeek Fonts. No interactive timeline was necessary. Turnbull simply did the tedious part correctly and put the comparisons on pages that still work.

For a designer, road historian, sign collector, transportation researcher, or somebody trying to identify an old photograph, that specificity is the whole point. A sign shape or symbol can help date an image. An obsolete designation can explain why a surviving roadside artifact looks “wrong.” A sequence of designs shows how public visual language becomes standardized.

This is also exactly the sort of narrow, durable research hiding inside CacheRat’s 1,967 Ancient Web Domains research list. Nobody needed to turn traffic-sign chronology into a content vertical. One person just decided the history deserved to be mapped.

The guide’s last recorded update is November 28, 2010. The signs have kept changing. The old comparisons have become more useful because of it.

Browse the Field Guide to American Traffic Signs

Posted on by

Ancient Web: RailroadMichigan Built a Noncommercial Photo History of Michigan Railroads

A railroad disappears twice: once from the landscape, and later from ordinary memory.

Visit RailroadMichigan

RailroadMichigan is the kind of personal archive that slows down the second disappearance.

The site describes itself as a non-commercial hobby photo collection covering railroads in Michigan and the Great Lakes region. Its purpose is straightforward: document both current and historical railroad transportation through photographs, slide collections, contributed images, equipment rosters, and reference material.

The index is huge.

It ranges across major carriers, short lines, industrial railroads, interurbans, scenic lines, narrow gauge, maintenance-of-way equipment, locomotives, logos, local systems, and companies that vanished decades ago.

Hobby archives often preserve the details institutions skip

Railroad history is unusually vulnerable to fragmentation.

A company gets merged. A spur is abandoned. A locomotive is sold three times. A station is demolished. A slide collection sits in a box until nobody remembers where the photograph was taken.

A hobbyist site can connect those pieces while the information is still recoverable.

RailroadMichigan does that with the stubborn specificity railfans are famous for. One roster can track locomotive model, road number, serial number, acquisition date, previous ownership, disposition, and notes about what happened afterward.

That level of detail is not decorative trivia. It is provenance.

The site’s design also reflects an older publishing ethic: no ads, no feed optimization, no attempt to turn the archive into a lifestyle brand. The webmaster explicitly presents it as a hobby collection and asks visitors to contribute corrections and additional information.

That makes the site collaborative without requiring a platform.

CacheRat’s 1,967 Ancient Web Domains research list contains many examples of this kind of volunteer documentation. Their common trait is not visual style. It is obsessive scope.

Someone cared enough to photograph, label, identify, and keep track of things that were moving out of sight.

Twenty years later, that obsession becomes public history.

Browse the Michigan railroad photo archive

Posted on by

Ancient Web: 9p.io Kept Plan 9 From Bell Labs Downloadable

Some old websites preserve pages. 9p.io preserves an operating system culture.

Visit Plan 9 at 9p.io

Plan 9 began at Bell Labs in the late 1980s, developed by people including Ken Thompson, Rob Pike, Dave Presotto, and Phil Winterbottom. It grew out of the same research environment that produced Unix, but tried to rethink distributed computing, namespaces, networking, and the idea that resources should be presented consistently as files.

The surviving 9p.io site keeps that world directly browsable.

It includes documentation, papers, manual pages, historical downloads, source browsing, mailing-list and community references, and links to related systems such as Inferno and Plan 9 from User Space.

That is important because the original plan9.bell-labs.com server is gone.

The server became part of the preservation story

The Plan 9 wiki now explicitly notes that the old Bell Labs server was shut down and identifies 9p.io as its continuation.

That makes the site more than a mirror.

It is a surviving bridge between an institutional research system and the people who continued using, studying, and adapting it after the institution changed around it.

Plan 9’s technical influence also makes that survival valuable. Its ideas include per-process namespaces, the 9P protocol, early complete UTF-8 support, and a radically network-oriented view of the operating system. Some of those ideas later appeared, in altered forms, elsewhere in Unix-like systems.

But reading about Plan 9 in a retrospective is different from browsing the actual project environment.

The old-style site still exposes manuals, release notes, source trees, downloads, historical context, and the little connective documents that show how the system was actually distributed and discussed.

That is the sort of continuity CacheRat’s 1,967 Ancient Web Domains research list is meant to surface. Computing history is not only preserved in museums and books. Sometimes it remains online because somebody kept the old tree mounted and reachable.

Plan 9 is still there.

Browse the Plan 9 distribution site

Posted on by

Ancient Web: The Typing Injury FAQ Documented RSI Before Ergonomics Became a Product Category

There was a time when learning how not to injure yourself at a computer meant finding somebody’s FAQ.

Visit the Typing Injury FAQ

TIFAQ is exactly what the name promises: a broad educational reference about repetitive strain injuries and ways to reduce the risk and symptoms associated with computer work.

Its navigation reads like a map of the ergonomic-computing problem before “ergonomics” became a retail category. There are sections on general information, furniture, alternative keyboards, speech recognition, pointing devices, accessories, software, services, organizations, and material aimed specifically at children.

The homepage records its last update as July 27, 2009.

That date now makes the site doubly useful.

A FAQ can preserve an entire technical moment

The practical problem of computer-related strain has not disappeared, but the hardware ecosystem around it has changed dramatically.

Alternative keyboards that once seemed exotic are now mainstream products. Speech recognition moved from specialized accessibility software into operating systems, phones, and AI-driven interfaces. Trackballs, split keyboards, adjustable desks, and other accommodations became ordinary commercial categories.

TIFAQ captures the period before that normalization.

It also reflects one of the best habits of the early technical Web: people gathered answers into durable documents because the same questions kept being asked in mailing lists, forums, and newsgroups.

A good FAQ did not need a publishing schedule. It needed maintenance.

That distinction matters.

Modern search often returns hundreds of lightly rewritten pages around a single problem. TIFAQ instead tries to build a reference structure around the subject itself.

That is why old FAQs remain so useful as historical artifacts. Their link lists show which organizations mattered. Their product pages show what solutions existed. Their terminology shows how a technical or medical problem was being understood at the time.

CacheRat’s 1,967 Ancient Web Domains research list contains many sites that function this way: not as media, but as maintained knowledge objects.

TIFAQ is one of them.

Browse the Typing Injury FAQ

Posted on by

Ancient Web: The 62nd Regiment Turned Reenactment Into a Research Archive

Historical reenactment can be costume theater, or it can become a research discipline.

Visit the 62nd Regiment of Foot

The old site for the recreated 62nd Regiment of Foot clearly aimed at the second.

The group focused on the British regiment’s service during the American War for Independence, especially the Saratoga Campaign of 1777. Its website went far beyond event photos and recruiting information. It collected officer biographies, biographies of enlisted men, women and children associated with the regiment, uniform documentation, courts-martial, primary sources, campaign organization, and observations from period military reviews.

That combination makes the site more interesting than a normal hobby homepage.

It is a research archive built by people who intended to physically use the research.

Material culture forces details to matter

If you are merely reading about an 18th-century regiment, a button pattern can be trivia.

If you are reconstructing a uniform, it becomes evidence.

The same goes for food, rank distinctions, baggage, camp followers, weapons, drill, and the composition of a company. Living-history groups often develop unusually specific collections because they have to turn documentary evidence into clothing, equipment, movement, and public interpretation.

The 62nd Regiment site exposes that process.

It also preserves a particular phase of independent historical publishing. Before Facebook groups and large reenactment platforms, a unit could maintain its own reference library and present its standards directly to the public.

The site’s last recorded update is from March 2020, which now gives it another archival layer: it documents not only the 1770s, but the Web culture of reenactment organizations in the early 21st century.

The surviving pages even retain a cancelled 2020 event schedule, frozen by the pandemic.

CacheRat’s 1,967 Ancient Web Domains research list repeatedly turns up sites like this—small groups quietly doing serious documentation without behaving like publishers.

Their work is scattered, but often unusually deep.

Browse the 62nd Regiment research site

Posted on by

Ancient Web: Brookline Connection Built a Neighborhood History Archive in 1998

A neighborhood can disappear without moving an inch.

Visit Brookline Connection

Buildings stay put. Street names survive. But the people who remember why a corner mattered, what used to stand there, which trolley ran past it, or where the old school picnic happened eventually disappear.

Brookline Connection was built against that kind of loss.

The site, designed and maintained by Clint Burton, says it has been online since 1998. It documents Brookline, a Pittsburgh neighborhood in the South Hills, through old photographs, maps, local institutions, transit history, municipal documents, recreation, churches, schools, war memorials, businesses, street scenes, and neighborhood trivia.

The result is not a polished civic portal.

It is better than that.

It is a local memory machine.

The Web was perfect for hyperlocal history

Brookline Connection can move from a 1910 school picnic to trolley service, from real-estate advertisements to municipal council minutes, from mine history to tornadoes, and from Brookline Boulevard storefronts to old aerial photographs without pretending those subjects belong in separate publishing silos.

That flexibility is one of the old Web’s great strengths.

A person who cared enough about one place could build the reference work that no commercial publisher would ever justify financially.

The site traces Brookline’s development from farmland and early settlement through the growth of Pittsburgh, the arrival of streetcars, the sale of ten farms for development in 1905, annexation into Pittsburgh in 1908, and the neighborhood’s 20th-century expansion.

It also preserves material whose value increases with age. A photograph of an ordinary intersection may seem unimportant when taken. Fifty years later, it can answer questions no written record can.

Brookline Connection understands that instinctively.

This is the kind of site that makes CacheRat’s 1,967 Ancient Web Domains research list useful as more than a collection of strange URLs. The old Web contains thousands of privately maintained local archives that collectively preserve history at a resolution institutions often miss.

No algorithm would have decided that one Pittsburgh neighborhood deserved this much attention.

A resident did.

Explore Brookline’s history archive

Posted on by

Ancient Web: Freefall’s Mirror Shows How Webcomics Survived by Being Mirrored

Before giant platforms absorbed most independent publishing, popular websites often survived by having copies elsewhere.

Visit the Freefall mirror

The Glass Wings copy of Freefall is a clean example of that older Internet habit.

Mark Stanley’s science-fiction webcomic began online in 1998 and follows Sam Starfall, robot Helix, and engineer Florence Ambrose aboard the perpetually troubled Savage Chicken. Over the years the strip developed a reputation for mixing comedy with engineering, robotics, artificial intelligence, ethics, and surprisingly careful science.

It also became enormous.

A long-running webcomic with thousands of strips is not just a set of images. It is a publishing system, an archive, a navigation problem, and a continuity problem. If the primary host goes down, years of reading can disappear behind one broken domain.

Mirrors helped solve that.

Redundancy used to be social infrastructure

Old-web mirrors were often maintained by people who simply wanted something useful to remain available. They were not always official archives in the institutional sense. They were practical redundancy.

The Belfry WebComics Index listed the Glass Wings site specifically as a mirror of Freefall. Fans also built search engines and translation sites around the comic because its archive became too large to navigate casually.

That ecosystem is historically interesting.

Today, creators are often told to trust a platform’s storage, discovery system, and account infrastructure. In the older Web, the instinct was more distributed: keep local files, maintain alternate hosts, build indexes, and link directly.

That architecture was messier, but it could be resilient.

The Glass Wings mirror also preserves something about webcomic culture itself. A strip could run for decades without being absorbed into a publishing conglomerate. Its audience could build tools around it. Readers could bookmark exact strips. Other independent websites could link directly into the archive.

CacheRat’s 1,967 Ancient Web Domains research list contains many sites whose survival depends on exactly this kind of redundancy.

Sometimes preservation was not a foundation or a library.

Sometimes it was another webmaster saying: I have disk space. Put a copy here.

Browse the Freefall mirror

Posted on by

Ancient Web: The Somnolescent Archives Treats File Hoards Like Museum Pieces

Some archives preserve famous things. Others preserve the stuff nobody else realized was worth saving.

Visit the Somnolescent file hoard

The Somnolescent Archives belongs to the second category.

The archive describes itself as a browseable museum of Somnolescent history, but its scope goes beyond the art collective that created it. It contains restored websites, old web experiments, abandoned pages, media hoards, obsolete software-era material, and directories of files that would otherwise be extremely easy to lose.

One of those directories is simply filed under ids-clipart.

That name is almost a perfect old-Web artifact in itself. It assumes a visitor is willing to click into a directory, look around, and figure out what the material is. There is no recommendation engine, no infinite scroll, and no attempt to inflate a handful of files into a content brand.

Preservation can be gloriously specific

The larger Somnolescent archive includes old group sites, material recovered from dead CMS installations, Apple iLife sound effects, game wallpapers, MIDI files, discontinued personal pages, and projects that never even became public websites.

That matters because the everyday Web disappears differently from institutional history.

A famous software release may be mirrored everywhere. A folder of promotional artwork, a dead personal homepage, an old iWeb gallery, or a pack of assets attached to a forgotten project can vanish without anyone noticing.

Somnolescent’s approach is unusually honest about that problem. The archive does not pretend every item is culturally monumental. It preserves things because they belonged to a particular computing environment, creative scene, or moment in Internet life.

That is exactly how museums work at their best. A chipped coffee cup can sometimes explain daily life better than a marble statue.

The old Web needs the same treatment.

CacheRat’s 1,967 Ancient Web Domains research list contains plenty of grand technical resources and historically important projects, but the small miscellaneous directories are just as revealing. They show what people downloaded, traded, built with, collected, and considered worth keeping.

Somnolescent has turned that instinct into an archive philosophy.

Browse the surviving file hoard

Posted on by

Ancient Web: PenCapChew.com Mapped Nirvana Fandom Before Social Media

Before fandom became a stream of reposts, a serious fan could build a place.

Visit PenCapChew.com

PenCapChew.com is a surviving Nirvana fan site from that model of the Web. Old music directories describe it as carrying Nirvana news, photographs, books, and a map of Aberdeen with Nirvana-related landmarks.

That last part is the clue to why sites like this mattered.

A fan page was not limited to repeating a biography. It could become a hand-built research environment around the band: images, places, references, discography material, books, local geography, and whatever else the webmaster thought another fan would want to explore.

Fandom had addresses

The pre-social Web produced thousands of music sites like this, but they did not all feel interchangeable.

Each webmaster selected what counted as important. One site emphasized tablature. Another collected scanned magazine pages. Another specialized in rare audio. PenCapChew’s combination of news, images, books, and an Aberdeen landmark map turned Nirvana history into something geographically browsable.

That matters for a band whose mythology is strongly attached to place.

Aberdeen is not just a line in a biography. The town, its streets, venues, houses, bridges, and regional culture became part of the way fans understood Nirvana and Kurt Cobain. A fan-maintained map connected online obsession back to physical geography.

The design and surviving references also belong to a period when fan sites linked heavily to one another. Music directories grouped dozens of independent Nirvana pages by specialty, creating a loose ecosystem rather than one official destination.

Most of that ecosystem has eroded.

CacheRat’s 1,967 Ancient Web Domains research list preserves traces of that decentralized fan culture because it shows how people organized information before platforms organized people.

PenCapChew.com is one small surviving room from the enormous fan-built Web around Nirvana.

Browse the surviving Nirvana fan site