The Address

Before social media told you who you were online, you had to build it yourself.

GeoCities gave you an address. Not a username. Not a handle. An address — in a neighborhood, on a street, with a number, in a town that existed entirely on the internet and felt more real than most of what came after it.

The neighborhoods had names: Hollywood, SiliconValley, Tokyo, Athens, WestHollywood, EnchantedForest. You chose one based on what you were about. geocities.com/SoHo/1234 implied art or fashion. geocities.com/SiliconValley/789 implied someone who wanted you to know they thought about the future. geocities.com/Heartland/ implied someone’s grandmother had learned HTML and was not going to let that go to waste.

The URL was a handshake before the page even loaded.

That’s what made GeoCities different from every hosting service before it and most of what came after. It didn’t give you a blank container. It gave you a place — a shared address space with implied neighbors, implied culture, implied belonging. You weren’t publishing to the void. You were moving into a neighborhood.

At its peak, GeoCities was the third most visited site on the internet. In 1999, Yahoo bought it for $3.57 billion. On October 26, 2009, they turned the lights off and deleted everything.

GeoCities died twice. The second death was the Tuesday in October. The first death was quieter and took ten years.

What Was Actually On It

It is worth being specific, because the specifics are what’s gone.

A typical GeoCities page had a header in the largest font available, usually glowing or drop-shadowed via a technique the page owner had found in a tutorial on another GeoCities page. A biography section. A hand-curated list of favorite links, described in the owner’s own words — not an algorithm’s summary, not a category tag, actual sentences about why this link mattered. A visitor counter from a third-party service that meant nothing statistically and everything psychologically. A guestbook, which was an open invitation to strangers to leave a note, functioning as the comment section of an era that didn’t have comment sections yet.

And almost always: an embedded MIDI file that started playing the moment the page loaded.

Not because anyone decided this was good design. Because the web was new and MIDI files were free and someone had figured out how to do it and the default assumption was that if you could, you should.

The whole thing was a home page, a journal, a mixtape, and a calling card stacked on top of each other. It required no permission from any institution. It cost nothing. It was visible to anyone on earth with an internet connection.

Some of those pages were the first thing a person ever published anywhere.

A teenager’s fan fiction. A retiree’s model train collection. A small city’s page for its annual harvest festival. An eight-year-old’s shrine to their favorite cartoon, with a hand-drawn GIF and a complete episode list written from memory. A widow’s page about her late husband, updated once a year on his birthday.

Yahoo deleted all of it.

The Neighborhood System

The neighborhoods are the detail most people get wrong when they try to explain what GeoCities was.

They treat the neighborhoods as organizational taxonomy — a way of sorting pages by topic. That’s technically accurate and completely misses the point.

The neighborhoods created identity. Choosing Hollywood over SiliconValley over Athens wasn’t picking a category from a dropdown menu. It was a statement about who you were and who your neighbors were. It was the early web’s first functional attempt at building community through address rather than through algorithm.

David Bohnett and John Rezner launched the site in November 1994 under the name Beverly Hills Internet — a name that correctly suggests an enormous amount of confidence for a startup running out of Los Angeles. The neighborhood concept was there from day one. The cities were named after real places, but their meaning was constructed entirely by the people who moved into them.

Hollywood collected entertainment fan pages, actor shrines, and pop culture obsessives. Tokyo gathered anime fans and video game communities years before either had mainstream Western audiences. Athens hosted religion and philosophy. WestHollywood became the early web’s most significant home for LGBTQ personal pages at a time when mainstream platforms didn’t exist and mainstream culture wasn’t particularly welcoming. EnchantedForest was for children’s content and family pages.

The neighborhoods made the web feel inhabited in a way it genuinely wasn’t yet. You could walk through them — click from one address to the next via guestbook signatures and favorite link lists — and encounter a version of the internet that was built entirely by people who had something they wanted to say, and no one to say it to, and a free hosting service that handed them a neighborhood and said: go ahead, build something.

What Yahoo Bought and What It Did With It

By 1999, GeoCities had 38 million users and was the third most visited site on the internet, behind only Yahoo and AOL.

Yahoo paid $3.57 billion in stock for it on January 28, 1999. Yahoo’s stock was trading at $367.75 that day. The same stock was at $14.36 when they announced the shutdown ten years later.

The acquisition looked like validation. The web’s scrappiest mass platform had been recognized as worth billions. The neighborhoods would be preserved. The community would be maintained.

What happened instead was slower and more boring than a betrayal, which is how most betrayals actually work.

The first visible sign was the URLs. Shortly after the acquisition, Yahoo restructured GeoCities accounts under Yahoo member names. geocities.com/Hollywood/1234 became geocities.com/yourname. The neighborhood was still technically there in the directory structure, but it had been drained of meaning. The address no longer told you anything. The handshake was gone.

Upload limits tightened. The interface accumulated advertising layers. New features arrived that served Yahoo’s business model rather than the people building pages. The experience of building a GeoCities page started to feel like negotiating with a property management company rather than painting a room you owned.

The users who had been there from the beginning noticed. The pages they’d built were still technically accessible, but the place they’d built them in had been converted into something else. The joy was procedurally extracted — not in one decision, but in the accumulated weight of a hundred small ones that each made sense as a business choice and collectively dismantled the reason anyone had cared.

This is the first death. It is the slower one, and in many ways the more complete one.

The Announcement

On April 23, 2009, Yahoo posted a message to its help pages.

GeoCities would be shut down later that year. New signups were stopped immediately. Existing users would be encouraged to migrate to Yahoo’s paid hosting service. The announcement contained this line: “Due to the high cost of maintaining these sites, we have made the difficult decision to discontinue GeoCities in the U.S.”

The high cost of maintaining the sites.

The sites that 38 million people had built.

TechCrunch’s headline that day was “Yahoo Quietly Pulls The Plug On GeoCities.” Quietly is doing a lot of work in that sentence. Yahoo didn’t hold a press conference. They posted to a help page. The third most visited site on the internet at the peak of its existence was discontinued via a help page update.

The reaction was immediate and disorganized and desperate in the specific way that only happens when people realize something is ending that they had assumed would simply always be there.

Forum threads appeared with instructions for manually downloading HTML files. Browser extensions circulated for mirroring entire pages. People wrote farewell letters directly onto their pages. Some users discovered their accounts had already been abandoned for years — pages they’d built in 1998 and forgotten about, now about to be deleted rather than just sitting there, which felt different in a way that was hard to articulate.

Yahoo offered an export function. It was brittle, incomplete, and failed consistently on pages that were large, old, or dependent on embedded resources hosted elsewhere.

The shutdown date was October 26, 2009.

What Was Lost and Who Counted It

On October 26, 2009, GeoCities did not explode. It went dark.

URLs that had existed for fifteen years returned 404 errors. Guestbooks stopped accepting signatures. MIDI files stopped playing. The visitor counters stopped counting. The under-construction GIFs, which had been promising an update on some pages since 1997, were finally, permanently, accurate.

The Archive Team — a volunteer group that had formed specifically in response to the shutdown announcement — had spent the preceding months scraping every page they could reach. Their documentation of the effort, preserved on the Internet Archive, contains this line:

“Yahoo! succeeded in destroying the most amount of history in the shortest amount of time, certainly on purpose, in known memory.”

Archive Team assembled nearly 100 volunteers and pulled what they could. Their haul was massive and incomplete. They collected broken pages, missing images, dead embedded files, and pages that rendered as fragments because the external resources they depended on had already vanished. They uploaded the results to the Internet Archive in eight compressed packages totaling hundreds of gigabytes.

The Internet Archive itself had been crawling GeoCities since the mid-1990s, but its captures prioritized HTML structure over media files. A page captured in 1998 might have its text saved but not the embedded MIDI, not the images hosted on another service, not the interactive scripts that required a live server. The Archive saved the bones. The rest was already gone.

What survives is navigable at Oocities and Reocities — mirror sites built from the Archive Team data. They look like a collage of fragments rather than a living neighborhood, because that is what they are. A guestbook that lists names without the messages. An image placeholder where the GIF used to be. A favorite links list where most of the links go nowhere.

Nobody knows the exact number of pages that were deleted. Yahoo never disclosed it. Archive Team’s attempts to get the figure from Yahoo were rebuffed. The best available estimate is in the tens of millions.

The Three Kinds of Loss

Preservation is not one thing, and what GeoCities lost happened in three registers simultaneously.

What was physically lost is the most documented: the HTML files, images, audio files, and scripts that made up the pages themselves. Millions of files that existed nowhere else, deleted in a single operation.

What was structurally lost is harder to measure: the neighborhood system, the guestbook networks, the webs of favorite link lists that connected pages to each other. These were the social infrastructure of GeoCities — the equivalent of knowing who lived next door and having a reason to knock. You can read an orphaned page but you can’t feel where it sat in the neighborhood.

What was contextually lost is the hardest to explain and the most important: what it felt like to navigate the early web when GeoCities was alive. The specific experience of clicking from one stranger’s page to another through their link lists, landing on pages that had been updated six years ago and never touched since, reading guestbook entries from people who are now in their forties and were then in their teens.

That experience was only ever available in the present tense. No archive captures it because no archive can.

The Pattern It Started

GeoCities is sometimes described as a nostalgia story. The MIDI music, the bad fonts, the under-construction GIFs. That reading isn’t wrong. It’s just incomplete in a way that matters.

The actual lesson in GeoCities is a governance lesson, and it’s one the internet has had to learn repeatedly since 2009.

Users built the houses. Yahoo owned the land.

When the landowner decided the maintenance costs outweighed the value of keeping the town standing, the town was demolished. Not because it had failed. Not because people had stopped caring. Because a company ran the numbers and the numbers didn’t work, and nobody with the authority to say otherwise was in the room.

The same pattern played out on Vine in 2017 — a creative community that invented the short-form video format that TikTok later built a company worth hundreds of billions of dollars on, shut down without a meaningful preservation effort. On Google+ in 2019. On Tumblr after Verizon acquired it and implemented a content policy that gutted the community it had spent a decade building. On MySpace, which lost years of uploaded music when a server migration went wrong in 2019 and nobody noticed for months because nobody was checking.

GeoCities was first, and it was large enough that the loss was visible to people who had no other language for what they were watching. It established the template: corporate acquisition, slow erosion of what made the community distinctive, declining traffic used as justification for shutdown, inadequate preservation tools offered too late, deletion executed on a schedule that served the company’s fiscal calendar rather than the community’s needs.

Every platform death since has followed the same general shape.

The Wayback Machine preserves fragments of what GeoCities looked like. You can watch the neighborhood system disappear in the archived snapshots — there one year, restructured into something generic the next. You can find individual pages frozen in whatever state they were in when a crawler last visited. Reading them now is a specific kind of strange: you’re watching a person exist in a permanent present tense, on a page that no longer exists, in a town that was demolished fifteen years ago.

GeoCities didn’t disappear because it stopped being interesting.

It disappeared because the company that owned it decided it wasn’t worth keeping.

That’s the haunting. Not the MIDI music. Not the broken image links.

The fact that someone made that decision, and nobody stopped them, and most of what was lost is gone permanently, and the same thing has happened over and over since, and will happen again.

The URL still resolves. It goes nowhere.

Sources & Further Reading

• Wikipedia — GeoCities: Full documented history from the 1994 Beverly Hills Internet founding through the shutdown, neighborhood system, and Archive Team preservation efforts. https://en.wikipedia.org/wiki/GeoCities

• Archive Team — GeoCities Snapshot (Internet Archive): The Archive Team’s own documentation of the scraping operation, including the “most amount of history in the shortest amount of time” line and the technical scope of the rescue effort. https://archive.org/details/2009-archiveteam-geocities-part1

• TechCrunch — “Yahoo Quietly Pulls The Plug On GeoCities” (April 2009): The original shutdown announcement coverage, including ComScore traffic data showing the 24 percent visitor decline leading up to the closure. https://techcrunch.com/?p=58916

• eWeek — “Yahoo Demolishes GeoCities 10 Years, $3.6 Billion Later” (2009): Covers the acquisition price in context of Yahoo’s stock collapse — $367.75 per share at acquisition, $14.36 at shutdown. https://www.eweek.com/it-management/yahoo-demolishes-geocities-10-years-3.6-billion-later/

• The History of the Web — “An Ode to GeoCities”: Jay Hoffmann’s detailed account of the neighborhood system’s cultural function and the Archive Team’s formation in response to the shutdown. https://thehistoryoftheweb.com/an-ode-to-geocities/

• TechSpot — “What Ever Happened to GeoCities?”: Overview of the Yahoo acquisition’s immediate effect on the neighborhood URL structure and the broader Web 2.0 context that made GeoCities feel obsolete. https://www.techspot.com/article/2401-geocities/

• Oocities: One of the most complete public-facing GeoCities mirrors, rebuilt from Archive Team’s scraped data. The closest thing to a museum of what the platform looked like. https://www.oocities.org

• Wayback Machine — GeoCities Captures: The Internet Archive’s crawl history of GeoCities, showing the neighborhood system changes over time. https://web.archive.org/web/*/geocities.com

• Rhizome — “One Terabyte of Kilobyte Age”: Olia Lialina and Dragan Espenschied’s ongoing project systematically posting screenshots from the GeoCities archive — the best available resource for experiencing what the platform looked like. https://oneterabyteofkilobyteage.tumblr.com