Black Tuesday: The Web Ring War Nobody Outside It Noticed

Before search engines were any good, finding a site about something you cared about meant walking a circle.

A web ring was a set of sites on one topic, each carrying a navigation bar with previous, next and random links. Click next enough times and you came back where you started, having seen every site somebody had judged worth including. A person made that judgment. There was no algorithm, no ranking, and no fee.

I described it at the time as following a pathway through a dense forest, moving from village to village. That is still the closest I can get.

I ran a lot of rings. At one point my site belonged to more than two hundred of them and I actively managed over a dozen. By the time I closed my account it held 2,900 memberships. I wrote about the craft of it at ring-master.net: how to run a ring, what a ringmaster owed the members, why the code fragment on each page mattered more than anyone thought.

What follows is what happened to all of it, and what being inside the fight taught me.

Where web rings came from

On 22 December 1994, Denis Howe at Imperial College in London started something he called EUROPa, the Expanding Unidirectional Ring of Pages. He wanted to see how many people would join.

The system was as simple as it sounds. If you wanted in, you built a page for EUROPa and emailed the person you had taken the page from, asking to be added. That person added you by pointing their next link at your site. No software, no database, no central anything.

In June 1995 a young man named Sage Weil found the ring and thought he could improve it. He wrote a CGI program that automated joining. Jerry Hierro suggested running the program centrally instead of on each site, and that combination became WebRing. The first ring built on it was called ESLoop.

Growth was fast. Over a thousand rings by 1997. More than ten thousand by 1998. By 2000 the number topped eighty thousand, carrying somewhere over a million sites.

What did a ringmaster do?

A ring took real work to run, and most people who joined one never saw any of it.

The first job was writing down your criteria before you accepted anybody. It sounds bureaucratic and it was the difference between a ring and a pile of links. If the criterion was quality Star Trek sites, then a Star Trek site that was three broken images and an away message did not get in, and neither did somebody’s excellent site about lighthouses.

Then you enforced it. You reviewed applications and turned people down. You checked members periodically, because sites vanished without telling anyone and a ring with three dead links in it stopped being worth walking. You wrote the welcome page, which was how a stranger decided whether your ring was worth joining. You answered mail from people whose fragment had stopped working.

I wrote at the time that a badly run ring made the ringmaster’s site look bad, and a well managed one made it look professional. That was the honest self-interest underneath the volunteering. Your ring was a public sample of your judgment.

None of the work paid anything, and nobody expected it to.

The uncomfortable thing I found in my own server logs

In 2001 I published a finding nobody wanted to hear, including me.

I spent hours going through server statistics trying to work out where my traffic came from. Search engines, web rings, ezine ads, link exchanges, FFA pages. The conclusion was inescapable. Excluding a handful of enormous rings, membership in a web ring was a minor source of traffic. Rings were not a traffic engine and anybody selling them as one was selling something.

People ran them anyway because the point was never the numbers. What you built was a small path through the wilds of the internet, guiding visitors through your own pick of sites that reinforced your theme. You were giving sites a way to talk to each other, and the path was the value.

I think about that finding a lot now. People poured years into something with almost no measurable return because they thought it was worth doing, which no business plan would ever approve and which was exactly what made the early web worth being on.

The fragment, and why ringmasters argued about it

The fragment was the small block of code each member pasted into their page, carrying the ring navigation. It sat on somebody else’s website and represented your ring, so it had to be small, plain and unbreakable.

There were two ways to build it, and the argument between them turned out to matter enormously.

A local HTML fragment lived on the member’s own page. They controlled how it looked, it loaded at the speed of their own server, and nobody could change it without their knowledge. The cost was that updating a ring’s navigation meant coordinating with every member.

A remote JavaScript fragment called back to the central hub and drew whatever the hub sent. Easier to install, easier to maintain, precise statistics, and the ringmaster kept total control over appearance.

I wrote an analysis of the tradeoff and came down hard against the remote version, on grounds that read differently now than they did then. A member who pastes remote JavaScript into their page is trusting the central hub completely. The hub can change what appears there at any moment, to anything, without notification and without consent. Advertising. Pornography. Anything at all.

I rated the systems in a table and gave Webring a one out of five overall, with a note in capital letters saying it was not recommended for any purpose due to the security weakness of the remote navbar. That was not a popular position.

Nobody in 2001 called this supply chain risk. It is the same thing every argument about third-party scripts has been about ever since.

Black Tuesday

Weil sold WebRing to Starseed in 1997. GeoCities bought Starseed in late 1998 and changed almost nothing, which was the correct decision. Then Yahoo bought GeoCities in early 1999 for $3.57 billion, and WebRing came along in the box.

For a while nothing happened. New rings were created at a good rate and on the surface everything looked normal.

Then a memo went out to ringmasters announcing great and wonderful changes.

On 5 September 2000 the new system arrived. Ringmasters lost control of their own rings. The customizable list page, which was where a ring showed what it was and who belonged to it, was replaced by a Yahoo hub page carrying links to Yahoo properties. Ads could no longer be suppressed. The navigation people had spent five years refining was replaced with something that worked worse and could not be changed.

We called it Black Tuesday, and the name stuck for years afterward.

What the community sounded like that week

The postings are still worth reading, and they are not the measured commentary of people discussing a product update.

A ring with around eight hundred members announced it was moving to RingSurf. Somebody wrote that the system could not be more broken, insulting and ridiculous if it had been put together by crack-smoking chimps. Somebody asked whether anyone could lend her a handful of Valium. A ringmaster wrote that she had shut down half her rings already and would do the rest when she got home, and that she would not be assimilated.

One posting stayed with me more than the angry ones. A newcomer wrote that she had started with webring.org and everything had been simple and clear for someone who had no idea what she was doing. Then Black Tuesday. She said she now felt like she was no one as far as the new system was concerned.

The change took a system a beginner could understand and replaced it with one that made her feel like nobody.

In October a ringmaster named Jack Smit wrote an open letter to Sage Weil, who had sold the thing three years earlier and had nothing to do with any of it. He wrote that whatever anyone thought about web rings, central to the concept was a sense of community, and that Weil’s creation had produced familiarity, volunteerism and a spirit of openness beyond borders. Then he named the date it ended.

People do not write open letters to a stranger about a piece of CGI software. They write them about something they loved.

The kingdom of Webring

A year on I wrote the whole thing up as a fairy tale, and warned readers it was one of the old ones, not the sanitized kind where nobody gets eaten.

There was a kingdom called Webring, united and content, holding over eighty thousand villages and towns and well over a million souls. Each town was ruled by a ringmaster with unlimited authority over his own population. There were the usual problems of any large community, and in general people were happy.

Then a rumor wafted through the kingdom, horrible to the ears. The empire of Y! was visiting with an eye towards assimilation.

I stand by the framing. It is easy to read that now as a grown man being dramatic about a website, and I would sooner be that than the alternative. Something real was taken apart by people who never knew it was there.

The war nobody outside the community noticed

What followed went on for years, and almost nobody outside the community knew it was happening.

People organized. There were protest clubs and mailing lists and coordinated migrations. The great meeting hall was a group called Webringnews, until its moderators went on a rampage and people scattered to other clubs. RingSurf took a wave of refugees because it worked the way the old system had. Ringlink grew. Rings that had run for half a decade in one place picked up and moved member by member, because a ringmaster emailed everybody and asked them to follow.

Some of it was useful and some was shouting into the darkness. The clubs helped people work out what to do next, which mattered when the answer was not obvious. They also burned an enormous amount of energy hoping Yahoo would change its mind.

Yahoo did change its mind, eventually. It gave up the webring business. By October 2007, WebRing had bought back the stock Yahoo still held and was privately owned for the first time since the nineties. RingSurf came off autopilot and had humans behind it again. Ringlink thrived, and its rings lasted longer because its ringmasters stayed involved.

I wrote a piece then called “We won the war, have we lost the peace?” and described it as a Pyrrhic victory. I was closer to right than I understood.

Baggage

Somewhere in the middle of all this I worked out a word for what had been done to us, and I have not found a better one since.

Baggage is the part of a service that exists for the benefit of the provider and not the user. Advertising you cannot suppress. Hub pages linking to places you would not send anyone. Support desks that answer with sarcasm. Popups and popunders on a service you already pay for.

Once you have the word, you see it everywhere. Search engines put ads on the results page: baggage. FFA lists require you accept spam: baggage. Free utilities install spyware: baggage. The user pays a cost that has nothing to do with the thing they came for, and the cost is usually invisible until somebody names it.

I wrote that the only reliable ways to avoid it were to pay for the service outright, or to find good tools built by people who understood that money is not the basic motivation of life.

There is a reason I bring this up. Twenty-five years later the same trade is being made on a much larger scale, and the baggage now is that being findable requires paying platforms to vouch for you. I have written about that at length, and the argument is one I first worked out over a navigation bar.

Why did web rings disappear if the community won?

The thing that ended them was never Yahoo.

While we fought over who controlled the ring hub pages, search engines got good. By the time we had our victory, nobody needed to walk a circle of twenty sites to find things about a subject. They typed the subject into a box and got an answer. The reason to be in a ring evaporated, for reasons that had nothing to do with the war.

I left too, eventually. The decision took years of going back and forth, and when I finally made it I transferred every ring and closed the account. They gave me no clean way to delete 2,900 memberships, so I left them sitting there instead of clicking through the lot by hand. I offered the members of a few rings a move to privately hosted ones and about ten percent came.

webring.com went dark on 15 August 2020. Almost nobody noticed.

What it taught me

Three things, and I did not learn any of them at the time.

The first is that beating the villain does not save the thing. We won and it died anyway. The corporation was the visible part of a change that was coming regardless, and every hour spent fighting Yahoo was an hour not spent on the question of what people would use instead.

The second is that the loss was real even though the replacement was better. Search was more useful than walking a ring, and I am not nostalgic about that trade. What went with it was a mechanism where a person who knew a subject decided what belonged in it, and better ranking never replaced that.

The third is the one I keep returning to. Rings had a property nobody wrote down as a principle: they would consider a site nobody had heard of. A ringmaster looked at your page and decided whether it was any good. There was no requirement that somebody else had vouched for you first.

That property was not unique to rings. Directories had it. Early search had it. It survived about twenty-five years and then it went, and I have now watched a version of the same thing happen twice.

The road moved

Route 66 is the closest comparison I have found.

Little towns strung along a road, each different, each there because somebody built it and stayed. You drove through, you stopped, you saw what they had. Then the interstate went in a few miles away and the traffic stopped coming.

Nobody set out to kill those towns. The road went somewhere else. The towns are still standing.

Curlie is still there, the human-edited successor to the Open Directory Project, with over three million sites in it and a tagline reading “Humans still do it better.” It works. Nobody scrapes it. That is what a bypassed town looks like.

The road has moved again, and this time it is search itself that got bypassed. I have written about what AI search costs the small web, which is this story running a second time with better technology and higher stakes. If you would rather have the practical version, most of what is being sold as the answer does not work.

I still have a few of those GeoCities pages saved. One was called the Coconuts of Quendor, a Zork site somebody made because they loved Zork. It was never going to rank for anything. Somebody found it anyway, and that used to be normal.

Frequently Asked Questions