PLATO

The University of Illinois teaching system, conceived in 1960 to fight functional illiteracy, which students quickly turned into the first mass platform for multiplayer games, chats, and forums. PLATO never caught on as a mass learning tool, but the genres of multiplayer games and online socializing grew up right there, twenty years before they became commonplace on the internet.

In brief

PLATO was a teaching system at the University of Illinois, conceived in 1960 by Professor Donald Bitzer to fight functional illiteracy among US high school graduates. It never took off as a mass learning tool: Control Data put about a billion dollars into commercializing it and never earned the money back. Instead, students who got access to a programming language that nobody had planned to make a programming language wrote the first mass multiplayer games on PLATO within a few years (space battles, flight simulators, dungeons), along with the first group chats and forums, twenty years before this became the norm on the internet.

How it started (the founders)

Donald Bitzer taught electrical engineering at the University of Illinois at Urbana-Champaign, and he was struck by a statistic that about half of US high school graduates were functionally illiterate. In 1959–1960, together with the physicist Daniel Alpert, he founded CERL (Computer-based Education Research Laboratory), and as early as 1960 the first version of PLATO was running on the ILLIAC I mainframe. This was not a startup in the usual sense: two university faculty members were building research infrastructure on grants, not a product for the market.

The key decision that shaped the system's fate was made not by the founders but by a staff member. In 1967 Paul Tenczar, a biology graduate student, came up with TUTOR, a language meant to make it easier for instructors to write lessons, nothing more. TUTOR was not conceived as a general-purpose programming language; according to the source, its turning into one was unplanned. It was exactly this accident that opened the door: since you could write anything on PLATO, not just a lesson, students began writing games. By 1971 the system was running on mainframes from Control Data Corporation (CDC), which had been supplying PLATO with hardware and money since 1970, so a large corporate partner was on board before games had even appeared.

Year-by-year timeline

1960: Bitzer and Alpert found CERL; the first version of PLATO runs on the ILLIAC I fact
1967: graduate student Paul Tenczar comes up with TUTOR as a language for lessons; later, without anyone planning it, it becomes a general-purpose programming language fact
1971: PLATO moves to Control Data Corporation mainframes fact
1973 (spring): John Daleske, a student at Iowa State University, writes Empire I as a course project; Silas Warner provides the disk space fact
1973 (fall): Empire II (a tactical rework); in parallel, Doug Brown writes Talkomatic, the first group chat with the screen split into windows fact
1973-08-07: David Woolley launches PLATO Notes, expanding a bug tracker into full-fledged forum discussions fact
1974 (March): Jim Bowery releases Spasim, the first truly 3D multiplayer first-person game fact
1974: Brand Fortner and Kevin Gorey release Airfight; the same year administrators delete the game m119h completely fact
1975: three key RPGs come out in a single year: Rusty Rutherford's pedit5 (hidden from the admins, deleted anyway), dnd by Gary Whisenhunt and Ray Wood (Wood was a PLATO admin himself, so the game ran in the open), and Moria by Kevet Duncombe and Jim Battin (the first multiplayer first-person RPG) fact
1976 (early): the system reaches 950 terminals and more than 3,500 contact hours of teaching material; development of Avatar begins fact
1977: Jim Schwaiger, with John Gaby and Bancherd DeLong, releases Oubliette, a co-op RPG so hard that surviving alone is almost impossible fact
1979: Avatar comes out and ends up the most popular game in PLATO's entire history fact
1984 (Q4): Control Data's commercial PLATO turns an operating profit for the first time, after almost two decades and about $1 billion invested estimate
1989 (September): CDC sells the education business and the "PLATO" trademark to William Roach (The Roach Organization), and the CYBIS technology separately to University Online estimate

Lesser-known but significant facts

A language for lessons accidentally became a language for games. TUTOR was created in 1967 to make it easier for instructors to lay out teaching material; nobody planned for it to turn into a general-purpose programming language. Without that accident, students would simply have had nothing to write Empire and Spasim in.
PLATO Notes grew out of a bug tracker. David Woolley was extending a program for reporting system bugs and along the way decided that half a screen might not be enough for some entries and that some problems needed a back-and-forth dialogue; that is how the ability to post up to 63 responses to a single entry came about.
The fate of a game on PLATO depended not on the rules but on the author's status. Rusty Rutherford hid pedit5 under an inconspicuous file name, and it was found and deleted anyway. Meanwhile dnd by Gary Whisenhunt and Ray Wood ran in the open and was constantly improved based on player feedback, because Wood himself was a PLATO administrator.
A game was erased without a trace. The early game m119h (1974) was deleted by administrators so thoroughly that only mentions of it remain, unlike pedit5, which at least survived in a re-created version, Orthanc.
The same idea was born independently in two places almost at the same time. While students in Illinois were writing Empire and Spasim on the closed PLATO network, 1,000 miles away Steve Colley, Greg Thompson, and Howard Palmer were writing Maze War at NASA Ames in 1973 on cable-linked Imlacs; from 1974 on it was played between universities over ARPANET, until DARPA temporarily banned the game: its packets made up half of a month's traffic between Stanford and MIT.

Legend vs. the record

Legend: the first multiplayer online games appeared on the internet. The record: Empire on PLATO (spring 1973) and Maze War at NASA Ames (1973, over ARPANET from 1974) appeared independently of each other in the same year, twenty years before the internet became a concept for ordinary people. The PLATO network was not a predecessor of the internet at all: it was CDC's closed infrastructure on leased telephone lines, not connected to ARPANET. Verdict: the legend is wrong twice over, on timing and on substance: the events unfolded in parallel on two unconnected networks, neither of which was "the internet."
Legend: games on PLATO lived underground, hounded by administrators. The record: true for some games and false for others, and what decided it was not system policy but the author's status. Rusty Rutherford's pedit5 was hidden under an inconspicuous name, which did not save it from deletion; m119h was erased without any trace at all. Meanwhile dnd by Gary Whisenhunt and Ray Wood ran openly and was improved based on player feedback, because Wood himself was a PLATO administrator. Verdict: the underground was not a universal rule but a function of the author's access to the levers of the system itself.
Legend: the authors of Air Warrior on GEnie came out of the PLATO gaming culture. The record: Kesmai's founders Kelton Flinn and John Taylor met not in Illinois but at the University of Virginia, and wrote their first game, Air (1977), on the UVA mainframe; not a single source, including a specialized survey of the PLATO RPG genre, mentions any connection between them and PLATO. Verdict: there is no direct link between the authors; theirs was a parallel mainframe culture at another university, and the connection to PLATO is one of genre, not of people.
Legend: Airfight on PLATO inspired Microsoft Flight Simulator. That is how retrospectives of the genre often describe this game. The record: the creator of MSFS, Bruce Artwick, flatly denied having played flight simulators on PLATO, and Wikipedia explicitly calls this attribution mistaken. Verdict: Airfight's influence was real (via Spasim → Airace → Sublogic Flight Simulator), but not directly on MSFS; the legend mixes up two different lineages of simulators.

The first growth lever

PLATO's growth as a gaming platform was set off not by CDC's marketing but by a side effect of the TUTOR language: conceived in 1967 only for laying out lessons, it turned out to be a general-purpose programming language, and students gained the ability to write anything in it. On top of that came free access: terminals stood on campuses across the country, and any student could sit down and write a program without permissions or budgets. Then the network effect kicked in: John Daleske's Empire (spring 1973, a course project) sprouted versions II, III, and IV within months, because other students (Silas Warner, later Chuck Miller and Gary Fritz) added features for free, and users spent about 300,000 hours on the game between 1978 and 1985 estimate. In parallel, Notes (from August 1973) and Talkomatic (from fall 1973) took off, growing not out of the teaching plan but out of tools for bugs and for one-on-one chat. By 1976 the system had reached 950 terminals and 3,500+ hours of teaching material, but the most popular program in PLATO's history was not coursework but the game Avatar (from 1979), with 6% of all the system's hours of use in 1978–1985 estimate. The growth lever was not educational value but the fact that PLATO turned out to be the only place where thousands of students could play, chat, and program at the same time, for free and with no marketing looking over their shoulders.

The paired story

PLATO and GEnie are not a single line of succession but two parallel stories of online games before the mass-market internet, which converge only in mechanics, not in people. On PLATO, from 1973 on, students wrote multiplayer games for free on infrastructure built for teaching (Empire, Spasim, dnd, Airfight, Avatar). On GEnie a decade later the same genres became a paid business: hourly billing for Air Warrior and the other games from the Kesmai studio was one of the main growth levers of GEnie itself (see the GEnie dossier, "The first growth lever" section). There is no direct personal link: Kesmai's founders Kelton Flinn and John Taylor came out of the separate mainframe culture of the University of Virginia, where in 1977 they wrote the text game Air. What ties these stories together is not a baton passed between particular people but a plot of the era: before home internet, gamers had only institutional mainframes, either university ones (PLATO) or those of commercial services (GEnie), and both platforms independently discovered that multiplayer games keep people on the line better than anything else.

Parallels today (projects from the catalog)

DropPop (droppop-depop-automation-saas) is a service for resellers that stands entirely on someone else's infrastructure (Depop, Vinted, eBay) and depends completely on how tolerant the platform is of automation, the same risk faced by game makers on PLATO, who depended on whether the CERL administrators would notice them. The difference of era: with DropPop, access to someone else's platform as a business tool costs $19–79/month, while PLATO students got access to someone else's infrastructure for free, as a side effect of a teaching grant. view this project's dossier →
Instagram automation extension (chrome-ext-instagram-automation-flippa-11670694) is a direct parallel to the story of pedit5: the product exists by disguising itself as ordinary user activity so that the platform (Instagram) does not notice and ban the account. It is the same trick of hiding a file name from the administrator, only with social network moderation in place of CERL. The difference of era: in 1975 hiding meant renaming a file on disk; in 2026 it means imitating the behavior of a live human against detection algorithms. view this project's dossier →

What a builder can take from this in 2026

A general-purpose tool outruns a narrow-purpose one, even if the narrow one was the plan and the general one an accident. TUTOR was supposed to simply lay out lessons; the fact that it turned out to be a programming language created PLATO's entire gaming industry, which nobody designed.
Free access without permissions is a growth hack even when it is not meant as one. Students did not ask CERL for permission to write Empire or Talkomatic; the very ability to program on shared infrastructure was invitation enough.
The fate of a product on someone else's platform is not about the rules but about who you are to the platform. dnd lived and grew for years because its author was an administrator himself; pedit5 was hidden and erased anyway. When you build a business on top of someone else's infrastructure, the deciding risk factor is not compliance with the formal rules but your relationship with the people who can hit "delete."
A by-product (the social layer) can outpace the main product in engagement. Courses were not the reason people sat at PLATO terminals through the night; the reason was the game Avatar, which took 6% of all the system's hours over seven years estimate. A product's unplanned social side effect can turn out to be the real product.
Parallel invention is the norm, not a coincidence that has to be explained by a conspiracy. Empire and Maze War appeared independently in the same year on two unconnected networks. If an idea is technologically ripe, it will be invented in several places at once; there is no need to look for a lone genius or a single primary source.

Discrepancies and what we could not verify

The primary source for the 300,000 hours figure for Empire and the 6% of hours figure for Avatar. Both figures come from a secondary retelling of PLATO system logs on Wikipedia; Brian Dear's book The Friendly Orange Glow, which probably gives the original data, is available on archive.org only in borrow-only mode, and a full-text page-by-page search was not carried out estimate.
The exact amount and details of the 1989 sale of PLATO. The buyers' names (William Roach / The Roach Organization, University Online) are taken from an aggregated retelling of corporate sources; the deal itself (amount, exact signing date) has not been confirmed directly by a document of the era estimate.
The exact year Air Warrior launched on GEnie. In the paired dossier history-genie this fact is already marked estimate with a spread of 1986/1987/1988; we found no new data that would settle it, only confirmation of the 1987 version.
The quarter of CDC's first operating profit on PLATO. Sources differ: CSM-1984 implies Q4 1984, while an independent retelling citing Van Meer 2003 gives Q4 1983; neither has been opened page by page directly, so we are keeping 1984 as the more likely year, but estimate.
The amount CDC invested. The figures of $600 million and $900 million–$1 billion are probably different points in time rather than a contradiction, but the year-by-year growth curve has not been reconstructed estimate.

Sources (primary first)

Secondary (context, cross-checking):

Wikipedia: PLATO (computer system))
University of Illinois Alumni Association: In The Time Of PLATO
Massively Overpowered: The Game Archaeologist: Maze War
Christian Science Monitor, "Control Data graduates from Plato into the computer big-time," 1984-04-27 (aggregated retelling, not opened page by page)
Wikipedia (all the articles above): only as a pointer to primary sources