Which early social network truly came first, and why does no one agree

Ask which early social network came first and you get at least six confident answers. Each one follows from a different definition of the term.

SixDegrees.com, launched in 1997 by Andrew Weinreich, is the usual pick. It combined a profile, a stored friend list, and a way to click through your friends' lists. The site closed in 2000.

Classmates.com, started by Randy Conrads in 1995, collected school affiliations two years earlier. Friendster arrived in March 2002, MySpace and LinkedIn in 2003, Orkut in January 2004, Facebook in February 2004.

The lists disagree for a reason more interesting than any one list: social networking had no clean moment when social networking began. Its pieces arrived separately and were combined in different orders in different places.

This page names the pieces, the systems that carried each one, and why the record of that period is much thinner than the confident summaries suggest.

What to take away

The three parts

A persistent identity. Something that carries from one session to the next, so that what you said last week is attached to what you say today. Without it there is conversation but no reputation, and no way to accumulate anything.

The WELL, founded near Sausalito in 1985 by Stewart Brand and Larry Brilliant, gave members handles that lasted years. CompuServe's CB Simulator did the same from 1980.

An explicit list of connections. A record, held by the system, of who you are linked to. Not a group you belong to and not a room you are in, but a personal list that belongs to you and is different from everyone else's.

ICQ, built by Mirabilis in Israel, shipped a contact list in November 1996 and sold to AOL in 1998 for $407 million. AOL Instant Messenger shipped the buddy list in May 1997.

SixDegrees.com kept that list public in 1997, so anyone could see who you knew.

A shared surface built from those connections. A view assembled for you out of what your connections did. This is the part that arrived last and changed the most, because once a system is assembling a view, somebody has to decide what goes in it.

LiveJournal, started in 1999 by Brad Fitzpatrick, gathered your friends' posts onto one Friends page. Friendster built a similar view in 2002. Facebook's News Feed, added in September 2006, ranked that view instead of listing it in order.

Systems existed with the first and not the second. Systems existed with the first two and no assembled view, where you went and looked at each person in turn.

Usenet, live from 1980, ran on identity and conversation alone for years, with no stored list of who you knew. The combination is what people mean by a social network, and pulling the three apart explains most arguments about what counts.

Before the list, there were places

The earlier shape was spatial rather than personal. You went somewhere, and whoever was there was who you talked to.

Dial-up bulletin board systems were just that: a machine you called, with message areas, a local population, and rules set by its operator. Ward Christensen and Randy Suess put the first one, CBBS, online in Chicago in February 1978. Reach bounded that population, which often meant a town.

Distributed discussion systems such as Usenet worked the same way on a different scale, organizing talk by subject across many machines rather than by person. Tom Truscott and Jim Ellis built it at Duke University and the University of North Carolina in 1979, and it went live the next year.

In both, you did not have a follower list. You had places you visited and a name people recognized when you turned up. Your reach was the room. If you wanted more reach, you went to more rooms.

The shift from that model to a personal connection list is the actual structural change, and it moved the unit of identity from the place to the person. Everything that followed, including the argument about ranking that dominates the present, sits downstream of that one move.

Why nobody agrees on what was first

Pick a definition and you pick a winner. That is the whole dispute. The table below names the systems that win under each reading.

danah boyd and Nicole Ellison used the middle rows in a 2007 paper for the Journal of Computer-Mediated Communication. Their definition: a profile, a list of connections, and the ability to view and move through that list. On that test, SixDegrees.com in 1997 comes first.

Most news accounts follow them. Histories that count early directories put Classmates.com in 1995. Accounts that require a working assembled feed start with LiveJournal in 1999 or Friendster in 2002.

Friendster, built by Jonathan Abrams, drew millions of users before technical problems slowed it down, won US patent 7,069,308 in 2006, and shut down in 2015. MySpace launched in August 2003 and sold to News Corp in 2005 for $580 million.

Every popular origin story is really an argument for one of these rows. When two accounts contradict each other, they are usually not disputing the facts. They are using different rows and not saying so.

The honest version of the question is therefore not which came first, but which capability you care about and when it appeared in the form you mean. That question has answers. The other one does not.

The record is thin, and the gaps are not random

Anything written about this period is limited by what survived, and survival was selective in ways that bias the story.

Systems that were private by design left almost nothing. A closed board, a local dial-up service or a members-only space produced conversation that no outside index could reach and no crawler could copy. It is not that the material was lost later; it was never publicly recorded in the first place.

Text survived far better than everything else. Plain messages are small, easy to copy, and easy to migrate between formats. Images, sound, and anything that depended on a specific piece of software to display are much more likely to be gone or unopenable.

Popular things were copied more, so they survived more, and the surviving record therefore over-represents whatever was already widely seen. A history assembled from what remains will conclude that the period was dominated by the things that happened to be copied, which is circular.

Commercial systems that closed took their archives with them. When the operator goes, the record usually goes, and what remains is whatever individual users had already saved for their own reasons.

Cyworld, launched in South Korea in 1999, and Mixi, launched in Japan in 2004, sit in records that English-language timelines rarely read. Xiaonei, later renamed Renren, opened to Chinese university students in 2005.

What this means for anything you read

Treat confident chronology with suspicion, especially where it is precise. A specific claim about who did what first is generally a compressed version of a longer argument about definitions, repeated until the argument dropped off.

Some practical habits:

Where this site draws the line

Nothing here states an audience figure, an ownership chain or a full feature list as a bare fact. Those are the parts of this subject that are most often repeated and least often checked, and a reference page that asserts them adds another mirror to the pile.

Founding dates appear only where a definition and a source name them, and each one carries the reading it depends on. The rest is structure: which capability existed when, and where the record runs out. That produces shorter answers. It also produces answers that stay true.

Common questions

Why not just give the standard timeline?

The standard timeline runs CBBS in 1978, Usenet in 1980, and Classmates.com in 1995. ICQ follows in November 1996, SixDegrees.com in 1997, and LiveJournal in 1999. Then Friendster in March 2002, MySpace and LinkedIn in 2003, Orkut and Facebook in 2004.

Each step is a definition wearing a date. Different sources give different sequences, and each is internally consistent with the definition it never states. Reproducing one of them alone would hide the actual disagreement.

Was there really social networking before the web?

There was persistent identity, sustained conversation between strangers, community norms, moderation disputes and reputation. The WELL had all of it from 1985.

CBBS had message areas and local rules from 1978, and Usenet spread similar habits across machines from 1980. What most of those systems lacked was a personal list of connections held by the machine. Whether that counts is the definitional question above.

Why is so little of the early material available?

A mixture of privacy, format decay and commercial closure. Much of it was never public, some of it depends on software nobody runs, and a great deal of what was public went when the operators did. SixDegrees.com closed in 2000. Orkut, Google's service, shut down in September 2014.

If the record is that poor, can anything be said with confidence?

Yes, about structure. The sequence in which capabilities became possible is visible from the systems themselves and from what people complained about at the time. Dates that sources agree on include CBBS in 1978, Usenet in 1980, ICQ in November 1996 and SixDegrees.com in 1997. It is the priority claims that do not survive close reading.