Early social networks guide: reading old pages

Anyone reading old material from the pre-feed web meets forms that have changed or fallen out of common use, presented as if the reader already understands them. A page ends with a row of links to strangers; a message begins with a quoted block and a name; a guestbook holds greetings from people who never returned.

This early social networks guide explains what each form was for, what it tells you about the people who used it, and what to be careful of when you read it now.

What to take away

The forms

The personal homepage. One person's own page, hand-built, on space they controlled or rented. It was the identity document of the period: what a person chose to say about themselves, in whatever order and style they chose, with no template imposed. Reading one tells you what its author thought worth presenting, and the absence of any standard layout means every choice was a choice.

The trap is assuming it was updated; most were built once and then left, so the date it describes is not the date it was captured.

The webring. A webring linked a set of independent pages in a loop, each carrying a fragment that led to the next. It solved discovery before search did, and it did so by peer judgment: you were in the ring because the ring's keeper accepted you. A ring is therefore a record of who considered whom to belong, and the membership tells you more about a subject's social structure than the pages do. The hosts you are most likely to meet in captures are WebRing.org and RingSurf, with portals such as Bomis listing rings of their own.

The trap is that rings decayed; a captured page may point into a loop half of whose members had already gone.

The guestbook. A page where visitors left a line. It was the first public comment surface, and it held greetings, arguments, spam and occasionally the only surviving trace of a visitor who left nothing else anywhere. Read it as a visitor log with the tone of a postcard. The trap is reading volume as popularity: a guestbook was easy to fill and there was no way to tell a regular from a passer-by.

The mailing list. An electronic mailing list delivered every message to every member. It was the workhorse of serious discussion, and its archives are among the best-preserved records of the period, because each member held a complete copy. The software behind the lists — LISTSERV, Majordomo and Mailman — set the conventions you will meet in an archive: the subscription commands, the message headers, the quoting style and the signatures. Its structure tells you who did the work: the people who answered questions, kept the record, and were quoted.

The trap is that list archives preserve the public list and not the private replies that went alongside it, so an argument that seems to end abruptly was often settled off-list.

The chat room. Live, unarchived, present-tense. Internet Relay Chat carried most of it, on networks such as EFnet, Undernet and DALnet, where a channel was a named room anyone could join unless its operators closed it. Chat produced the communities with the strongest bonds and the weakest records, because nothing said in a room was kept unless a member kept it, and members mostly did not. What survives is logs saved by individuals, which are partial, undated and selected. The trap is treating a log as representative; it is one member's evening, chosen because something happened.

The profile directory. A listing of people, each with a page in a fixed template, searchable by attribute. This was the form that carried a stored identity, and it is the one that most resembles what came later. The trap is reading a directory entry as a stored connection: most directories listed people without recording who knew whom, and that distinction between a stored identity and a stored connection is the one this guide turns on, and the one the early social networks page exists to draw.

The early social networks by name

The forms above are the parts. The services a reader searches for are the assemblies, and each one is best read by asking which parts it combined and which it left out.

SixDegrees.com. A profile, a stored list of friends, and a way to look at other people's lists. Reading a captured profile means reading the friend list as the content: the point of the page was the enumeration, not the self-description. Classmates.com is the nearest sibling here — it tied people to a shared institution and year rather than to a chain of connections.

LiveJournal. Journals with a friends list and threaded comments. Read both halves together: the entry and the comment thread underneath it, which often carried more of the community than the post did. A captured journal without its comment pages is a fragment.

Friendster. Profiles plus explicit friend connections plus testimonials written by other people. Read the testimonials as a performance of connection and the friend list as a claim its owner did not fully control.

MySpace. A profile whose layout the owner controlled, a friends list ordered by the owner, and comments left on the page. The order of the friends list was itself a statement, and it is the detail most often missing from captures.

BlackPlanet and AsianAvenue. Identity-centered services where the profile and the community around it were the point. Read them for what members posted to each other — the boards and messages — rather than for the profile template.

Bolt and Ryze. Two smaller assemblies at opposite ends: one built for a young audience around profiles and private messages, one a business contact service where the list of contacts was the product. Read Bolt for the messaging if it survives, and Ryze for how members described their work.

Orkut and Bebo. Both organized around profiles and communities. Read the community pages, which hold the discussion the profile does not, and note which communities were busiest; engagement there, not the profile, was the draw.

Most of these services have closed or been rebuilt into something else. For the ones that closed, the surviving capture is the record, and it is thinner than the service was.

What the forms tell you together

The right-hand column is where the confident claims come from, and it is empty on purpose.

Reading the material now

The archive kinds described under internet culture archives hold these forms unevenly. List archives survive well. Homepages survive as captures with broken links. Rings survive as fragments. Chat survives barely.

Concrete cases make the forms testable: GeoCities pages could combine a personal homepage, a guestbook, and WebRing navigation; Usenet posts and Apache HTTP Server mailing-list messages show threaded exchange; IRC logs preserve chat only when channel and timestamp context survives.

Use Internet Archive's Wayback Machine to search for dated page captures, Google Groups for archived Usenet posts, and Apache HTTP Server mailing-list archives for list messages. The captures of SixDegrees.com, LiveJournal, Friendster and MySpace, and of the identity-centered services, are the working corpus for the section above; treat each as one dated observation of a page that kept changing. Record the archive's capture timestamp or message-header date separately from any claimed page or community creation date; a saved record's date is not a launch date.

The record you keep should let another reader reach the same capture: the archive, the item, and the date the archive stamped on it.

Worked reading: a captured GeoCities page whose text ends in a row of ring links, with a guestbook holding entries from visitors. The forms tell you the page was built to be found (the ring fragment), built to receive visitors (the guestbook), and built to present an identity (the page itself). What they do not tell you is when the page was made, how many people saw it, or whether the rest of the ring still existed at the same time. The capture's date is the only date you have, and it is the date of the copy, not of the page.

The surviving record is weighted toward the forms that were easiest to keep, which were also the most formal. The material that was warmest and most social is the thinnest.

Apply the norms of the form, not of the present. A one-line guestbook entry was not a low-effort post; it was the whole affordance. A long, quoted list reply was not pompous; it was the medium. And a personal homepage with no comments was not ignored; there was nowhere to comment.

Where the forms combined into something with persistent identity, a stored list of connections and a shared surface, the result is the thing the internet communities page describes as a container. The order in which those parts arrived is the argument that platform timelines shows cannot be settled by dates.

Common questions

Which form is the best evidence of a community?

The list archive, because it records sustained exchange between named people over time. The others record presence without exchange or exchange without record.

Were these forms social networks?

Each held one piece of what the term later meant. Whether the pieces count separately is a definitional choice, and this site does not make it for you.

Why is so much of this material broken when I find it?

Because it depended on links to other people's pages, and those pages went. A form built on peer links dies when the peers do, one link at a time.