Thematic history · 14 related entries
Playing with someone far away was possible long before it was practical, and practical long before it was profitable. The history of online gaming is three things converging: network capacity, communities that formed whether or not anyone planned them, and companies learning to sell a service instead of a product.
The first host-to-host message on ARPANET travelled from UCLA to the Stanford Research Institute in October 1969, and nobody involved was thinking about games. Within a few years students were. Maze War, developed by students working at NASA's Ames Research Center in the early 1970s, let players hunt one another through a wireframe maze on linked machines, and games on the PLATO educational system at the University of Illinois did comparable things for a community of users who had access to something almost nobody else in the world did. None of this was authorized in any formal sense, and much of it ran on machines booked and paid for by somebody else.
The decisive early experiment was textual. Roy Trubshaw and Richard Bartle built MUD at the University of Essex in 1978: a shared world described entirely in prose, which many people could inhabit at the same time, and in which what one player did persisted for everyone else. It gave the genre its name. More importantly it surfaced the social questions the field has been working on ever since. Who governs a shared space, what happens when players hurt each other, and how much of the experience is the game rather than the people inside it.
Access was the limit on all of it. These worlds ran on institutional computers reached from terminals, and the audience was whoever happened to be at a university or a laboratory. Through the 1980s consumer online services and bulletin board systems opened narrow doors for people with a modem and a tolerance for the phone bill. Neverwinter Nights on America Online, launched in 1991, put a graphical shared world in front of subscribers charged by the hour, which proved that a persistent world could be a commercial service rather than a campus amenity.
Doom in 1993 did more for networked play than any deliberate effort had. It ran over local networks in offices and computer labs, its deathmatch mode turned four people in adjacent rooms into rivals, and its editable data files meant players changed the game themselves. Quake in 1996 went fully three-dimensional and shipped with internet multiplayer, improved later by network prediction techniques that made play over a slow connection tolerable. Clans formed, players ran their own servers, and a competitive culture organized itself without any company arranging a thing. The infrastructure was improvised and the etiquette was invented as it went, and both carried straight into everything that followed.
Blizzard's StarCraft in 1998 attached matchmaking to a retail game through Battle.net, which removed the need to find a server or know a friend with a fixed address. In South Korea it landed alongside an unusually fast broadband rollout and a dense network of PC bangs, public rooms where people paid by the hour to play together. The result was televised leagues, professional players attached to corporate teams, and the first genuinely large esports ecosystem, years before anything comparable existed elsewhere. Infrastructure and social geography produced that, not the game on its own.
Meridian 59, published by 3DO in 1996, offered a first-person persistent world for a flat monthly fee and is often cited as the first commercial 3D massively multiplayer role-playing game. Ultima Online followed in 1997 with player housing, crafting, a genuine in-game economy and open conflict between players, and passed a hundred thousand subscribers. Its problems became the field's first real research material, because a world in which players could rob and kill one another generated governance questions no game designer had previously been required to answer. Theft, harassment and improvised vigilante justice all appeared quickly, and the operators found themselves running a small society as well as a game.
EverQuest in 1999 moved the form into 3D for a wide audience and organized play around groups and raids, which made the game a scheduling commitment as much as a pastime. Its social pull became a subject of press anxiety. World of Warcraft in 2004 then made the genre mass-market by removing most of the friction: forgiving solo progression, reliable service, and a design that respected the time of people who could only play occasionally. It grew past ten million subscribers and became a cultural reference far outside games. It also demonstrated that reliability and customer service, rather than novelty, were what kept a paying population in place.
The monthly subscription was the important business innovation, because it changed what a game was. A product ships and is finished. A service has running costs, an operations team, a patch schedule and a lifespan determined by whether revenue covers the servers. It also meant a company could keep changing the game after purchase, which is ordinary now and was startling then. And when a subscription world shuts down it is gone in a way a cartridge is not, which several have been, taking their communities with them. That fragility is now one of the central problems in preserving the recent history of the medium.
Sega put a modem in the Dreamcast box in 1999 and ran an online service, which was ahead of the connections most households actually had. Microsoft's Xbox Live, launched in 2002, was built for broadband from the start and bundled a single identity, a friends list, matchmaking and voice chat into a paid subscription. That combination made online play a routine console feature rather than an enthusiast's project, and it established recurring service revenue as central to platform economics. Sony and Nintendo eventually offered comparable services on comparable terms. Voice chat in particular changed the texture of console play, for better and also for considerably worse.
Business models then multiplied. Free-to-play removed the purchase price and moved revenue to items sold inside the game. Fortnite, which added a free battle royale mode in 2017, combined that with cross-platform play and became one of the most profitable games ever released, along the way hosting concerts and events that made it a social space for young players rather than only a shooter. Battle passes, seasonal content and continuous live operation turned games into scheduled products. Whether these arrangements are fair to players, particularly young ones, is genuinely contested. Several countries have examined the question through consumer protection and gambling law, and the answers so far differ from one jurisdiction to the next.
Online capability stopped being a feature and became an assumption. Games ship expecting patches, which changed how finished a release has to be on the day it goes on sale. Storefronts are online, so a library is an account rather than a shelf. Twitch, spun out of Justin.tv in 2011, made watching other people play into a medium of its own, created careers that did not previously exist, and changed how games are discovered. Amazon bought it in 2014 for close to a billion dollars, confirming that game video was a strategic media asset.
The costs are as real as the benefits. A game that requires a server can be switched off, and preserving online-only titles is an unsolved problem rather than a hypothetical one. Harassment and moderation at scale have proved far harder than the optimistic early writing about virtual communities assumed. Always-online requirements have caused public backlashes, most visibly when a console maker announced them in 2013 and then reversed course before launch. And a stable connection remains unevenly distributed, which quietly determines who is able to play what. Rural households, and much of the world outside wealthy urban centers, experience online gaming as something intermittent rather than continuous.
The present arrangement is services most of the way down. Subscription catalogs deliver libraries rather than individual titles. Cloud streaming, after Google's Stadia closed in early 2023, survives mainly as a feature bundled into a subscription rather than a product sold on its own. The largest games are operated continuously, with revenue coming from items and seasons rather than a single purchase, and a small number of them absorb an enormous share of the total time players spend. Competition regulators reviewing the industry's biggest acquisitions treated control of online catalogs as a central question.
What has not changed since Essex in 1978 is the part that mattered most. People go to these places for other people, and the design problem is still how to govern a shared space fairly. The tools are better, the populations are millions rather than dozens, and the money involved is enormous, but the questions about moderation, community and what a player is owed by an operator remain open. They are being argued now by companies and regulators rather than by graduate students, which is a change of venue rather than of subject.
14 entries, oldest first. Expand one for the key facts or open it for the full record.
Students working at a NASA research center turned a wireframe corridor display into a game where players stalked one another across cabled-together machines, an early ancestor of the first-person shooter.
Roy Trubshaw and Richard Bartle built a text world at the University of Essex that many people inhabited at the same time. It named the genre and set the social patterns of every online world since.
A licensed Dungeons & Dragons world ran on a consumer dial-up service, letting subscribers explore and fight together while paying by the hour. It is usually described as the first graphical multiplayer role-playing game sold to the public.
id Software released Quake as shareware on June 22, 1996. Its engine rendered genuinely three-dimensional spaces, and multiplayer over the internet shipped with the game rather than arriving later as an add-on.
Meridian 59 opened commercially in September 1996, offering a first-person persistent world shared by many players for a fixed monthly charge. It is frequently cited as the first commercial 3D massively multiplayer role-playing game.
Origin Systems opened Ultima Online on September 24, 1997. Tens of thousands of subscribers shared a fantasy world with housing, crafting, a player-driven economy and few restrictions on attacking one another.
Blizzard released StarCraft on March 31, 1998, with three factions that played nothing alike and matchmaking built into the game. In South Korea it became the foundation of the first large professional gaming scene.
EverQuest opened on March 16, 1999, presenting a fantasy world in first-person 3D and organizing play around groups. Its subscriber numbers exceeded every internal forecast, and its habits shaped online games for years.
A paid network service opened in North America on November 15, 2002, one year after Microsoft's console. It combined a persistent identity, friends lists, matchmaking and voice chat, all designed for broadband.
Blizzard opened World of Warcraft on November 23, 2004. Forgiving solo progression, a familiar setting and unusually reliable service pulled a genre that had been niche into the cultural mainstream.
A hundred players parachute onto an island, scavenge weapons and fight until one remains, inside a boundary that keeps shrinking. Sold unfinished, it reached tens of millions of players within a year.
Stadia ran games on remote servers and delivered video to browsers, phones and televisions, with no local hardware needed. The technology largely worked; the library, the purchase model and confidence in the company did not.
Microsoft folded game streaming into its Game Pass Ultimate subscription at no extra cost, treating remote play as a convenience rather than a product. Bundling proved far more durable than selling streamed games separately.
Stadia stopped operating a little over three years after launch, with hardware and software purchases refunded. The closure became the standard warning about digital libraries held on a single company's servers.
