379 reads

Curious Case of PLATO: the cold war Internet

by Pen MagnetOctober 5th, 2018

Too Long; Didn't Read

Millennials, myself included, would never get it.

People Mentioned

Companies Mentioned

Coin Mentioned

Curious Case of PLATO: The cold war Internet

Millennials, myself included, would never get it.

Because millennials are accustomed to hear of revolutions born in a garage (Google, Apple, Amazon) or a dorm room (Facebook, Wordpress), hitting eight figure revenues in the cradle, sometimes despite being loss-making.

They miss a point: those innovations are built from solid blocks invented decades ago - inside the lonesome corridors of military labs and Universities.

PLATO, a contrived acronym that stands for Programmed Logic for Automatic Teaching Operations, was one such founding stone.

On the face of it, it was merely a classroom teaching simulator system, powered by military grade vacuum tube mainframes and arrays of dumb terminals made of displays and keyboards, spread through out university campus(es).

Apparently, its being namesake of the Athenian Academy master seems like a cosmic coincidence.

Like fire invented by primitive man, it spawned a chain reaction of innovations — including Plasma Display, Touch Screen, ASCII Art (father of emoticons), games that later became genres, Chat rooms, Email, forums and social networks — all of it during 70s and 80s - before the dawn of personal computer, let alone world wide web.

And just like every other revolution whose time hadn’t arrived, its steam ended up powering other vehicles.

How did it all started?

It was the beginning of 1960s — Baby Boomers had arrived. A decade past World War II, both USA and Russia had finished warming up for the cold war — the war had already begun.

Patent offsprings from war research were taking over the world. Transistor had just delivered its first computer. In the USA, military and its sister organizations were nourishing themselves with fresh talents from educational institutes, at the same time giving back their innovations to the institutes, and the society at large.

During 1960s, funding picked for “Institutional Support Program” — an umbrella under which US Universities got research grants from NSF. The event that triggered these developments was the launch of Sputnik in 1957.

Soon after this, US Airforce had asked for proposals about computer aided training systems. In response, PLATO’s first version got built inside University of Illinois. Two physicists (Chalmers Sherwin & Daniel Alpert), an Electrical Engineer (Donald L. Bitzer) and a mathematician (Peter Braunfeld) toyed with the idea during 1959–1961.

Eventually Bitzer built it, loosely basing it on the Radar concept which had delivered much more for its age during WW II.

Despite lack of a Personal Computing devices or a communication network like Internet, PLATO grew up to the level of being a cult phenomena with limited following.

Or perhaps, precisely because of those shortcomings.

What it was really like to program in 60s?

Before trying to understand how PLATO was born, and commercially died much before its time had truly arrived, one must delve into history of computing past World War II.

Allied prowess in World war II showed that much more could be done with computing power of an enigma machine. It also showed that dots that gleamed on the radar screen could be made to educate students as well.

While these two presented an infinite number of technological growth possibilities, all this puffed up muscle weight was useless without bone stability — distributed computing power, and higher-abstraction programming language.

Contrary to today’s times, Computers weren’t multi-purpose, and good software was always verbose, boastful of its steeper learning curve.

In other words, hardware exactly did one job that it was built for, and software, despite having infinite number of customizations, couldn’t be easily ported to another hardware, and simply perplexed layman.

There was no cloud to store your data. There was no open source to minimize development costs. Alas, no internet to even hire a skillful engineer or programmer, and not everyone could learn programming.

A few engineers did everything they could with circuits and assemblers. Judgements were simple, the devil lied in the details, and failure was their destiny.

Yet, they followed some predefined rules:

Assemble from an already proven hardware and make compatible software, if needed create a language that suits that purpose.
Or go the other way around, design / buy your hardware around an already popular programming language, and be constrained within physical limits you are given.
Do both from scratch, in whatever order it succeeds.

To understand 60s better, we must understand what it lacked compared to 80s.

80s growth was fueled by two factors: Invention of Personal Computers and High Level languages. These two events changed the course of computing history in a disproportionately disruptive manner.

In 80s, PCs entered consumer space to become a household commodity apart from enterprise machinery. With integrated circuits, it turned into a device of assembly, supplier footprint grew quite large. It became trendy to assemble a home computer and program it. 80s wasn’t called home computer boom without a reason.

High level programming languages (like C in 70s) and object oriented programming languages (like C++ in 80s) added another boost.

A low level language is typically made of machine codes / commands directly understood by device. Conversely, High level language simply meant that a machine could be dictated by its software to perform certain kind of task in a more human-understandable manner rather than machine-code-like manner.

Object oriented languages were an extra (set of) abstraction layer(s) of human-centered programming — it allowed code to mimic real world entities (objects).

This meant that one simply needed to know about the automation-objective in order to write code; without needing to understand the hardware specifics. Which simply meant: Anyone could be a programmer with minimal training.

So how was PLATO system programmed?

While COBOL (language of the Mainframes before 70s) could not be categorized as low level language, it wasn’t high-level language either. Its syntax was filled with commands which made the programs too verbose.

The first two versions of PLATO system (PLATO I & II) had been programmed using FORTRAN. While it was widely accepted language for scientific programming, its learning curve was steep for an application developer dealing with educational courseware.

So did they make educational videos in FORTRAN?

All courseware had to be produced in interactive (see figure below for benchmark) manner. No verbose descriptions, no speeches.

Videos were beyond imagination for education. (isn’t it easy nowadays?)

Courtesy: Wikimedia Commons (https://upload.wikimedia.org/wikipedia/commons/0/02/PLATO_chem_exp.jpg)

But the real challenge was — how much could one program?

PLATO wasn’t a tool to operate on certain kind of data. It was the data. Content at scale. It was a paradoxical choice:

If anything could be taught using computers, why not everything? And if everything must be taught with computers, who would program it?

To create computer simulation for each subject, an instructor had to be subject matter + programmer. Having instructors produce an enormous courseware for single discipline was beyond singular human capability.

And without scale, it wouldn’t be cost-effective.

Couldn’t they crowdsource it?

They weren’t the days of user generated content. Why? Precisely because even users needed tools to generate content.

Example?

Instagram is a great app, but it relies on image-capturing capabilities of iOS and Android software. Those took years to get written, tested and modified to present developers with on-the-fly APIs that could capture and store a photo in couple of lines written in Swift / Objective C or Kotlin / Java code.

How did we arrive at this enormously helpful level of abstraction? By trial and error.

Everyone along the way kept adding their own level of abstraction, and programming languages evolved.

Remember the birth of Java? It was written to operate appliances in a platform-independent manner — obliterating the need to recompile for every platform. It became instrumental in shaping the web and enterprise alike.

Developers of PLATO did just the same.

They chose the most suitable candidate for their purpose, and built upon it: TUTOR — an authoring language developed by a student biologist Paul Tenczar.

TUTOR:

TUTOR was a collection of commands an author could use to have desired simulation / illustration on the terminal screen. Much like COBOL, its capabilities far outranked even today’s MOOCs transmitting audio-visual renditions.

Apart from basic text / shape rendering, it could ask questions, judge the answers, and provide comprehensive reports to instructors.

With time, it was adorned with animations, and became love of game programmers.

For the complete list, here is the source.

But how did one play on PLATO?

Or rather, use PLATO?

PLATO II terminal features TALK/DATA button that can dial into a mainframe machine. (Courtesy: http://www.bitsavers.org/pdf/cdc/plato/97405900C_PLATO_Users_Guide_Apr81.pdf)

PLATO ran on terminals — a device which in its most basic form was incapable of doing any of the computing (storing, processing and memorizing of data). It merely took user input from keyboard (and later from mic, touch), sent it for processing to the mainframe computer (via telephone / video signal) that did the real stuff.

Later on, Terminal came to become known as thin client, limiting itself to just user data exchange, as opposed to thick client that was capable of processing the data.

PLATO III student terminal (courtesy: https://archives.library.illinois.edu)

In 1969, TUTOR had matured. Third iteration of PLATO, known as PLATO III, was fully functional, capable of supporting 20 terminals over single mainframe CPU. This was simultaneously used by Springfield High School (remote terminal) and University of Illinois Champaign-Urbana campus.

Due to its astounding results, National Science Foundation came up with assured long-term funding, and CERL (Computer-based Education Research Laboratory) — a new body dedicated to PLATO’s development within campus was founded.

Computers that ran PLATO III system (courtesy: https://archives.library.illinois.edu)

PLATO IV got a stronger head start. The days of uncertainties were over. Donald Bitzer, the father of PLATO, became busy ornamenting PLATO with never-before feature: Plasma screen. Its fast baud rate allowed for better graphic rendering and smooth animation experience, preparing the basis for later game development.

As Bitzer imagined, Plasma screen became the prime attraction for corporate investment. But little did he imagine it could be a game-changer for TV industry. In hindsight, it was a decision driven by necessity being the mother of invention. Every pixel of the Plasma screen contained a display memory bit — hence eliminating high costs involved with memory device and astronomical bandwidths.

They were the times when telephone bills ran in hundreds of dollars per month per classroom; when very few companies manufactured computers; and when memory cost $2 / byte. Imagine paying $2,000,000 for 1 GB RAM, or $300 / month / classroom for telephone bills, and you will get it.

Bracing up, PLATO came up with yet another innovation that we only know from smartphones: Touch sensitive screen. Invented beforehand, PLATO IV was the first fully functional large scale implementation of touch sensitive UI that happened inside college campus.

PLATO IV (Courtesy: https://archives.library.illinois.edu)

But there was yet another area where PLATO IV demonstrated it true technical prowess, to the extent of ultra-disruptiveness: Teaching of Music.

Since PLATO IV terminal was also capable of connecting to an extra audio device that was capable of detecting sound properties, including tone, pitch and rhythm. It could also access music media (eg tapes) on random-access basis. This meant that a teacher (or software) could select and play a clip from storage at random, and ask students to identify it.

At the time when storage media (hard disks and such) were in the cradle, this was a breakthrough. It enabled testing music students on their singing and playing performances. Illinois School of Music successfully ran this Learn-music-at-your-own-pace for quite some years.

The CDC Hardware:

All discussions around PLATO cannot be complete without the business angle — the hardware it ran on: Computers provided by Control Data Corporation (CDC).

PLATO I & II were developed using ILLIAC I, a computer developed by University of Illinois itself. ILLIAC I was running off of vacuum tubes — classified as 1st generation of computers — bigger in size and nasty in handling. Yet with it, in PLATO II, Bitzer was able to show that efficient usage of resources was possible by simultaneously supporting 2 users.

Due to its massive courseware target, the University started looking for external commercial hardware provider. CDC was one of the 9 corporations that manufactured mainframes and super computers — and the chosen one.

Historically, it was an offshoot of US Navy’s WW II code breaking team, later incorporated as ERA (Engineering Research Associates) after a series of acquisitions. CDC was led by William Norris, a visionary businessman who envisioned PLATO’s future outside mainstream classrooms.

CDC 1604 mainframe was a 2nd Generation computer made of transistors (higher in speed, lower in power consumption than their vacuum tube counterparts), and CDC’s first foray into mainframe business where it took up against hardware giant IBM.

Used by Bitzer in PLATO III’s development, CDC 1604 was the brainchild of Seymour Cray — who later earned the title of the father of supercomputer CDC 6600 — the first truly successful supercomputer.

Trivia: Due to its phenomenal success, IBM became preoccupied with designing 6600 killer, but could not go beyond campaign, taking a heavy toll on CDC’s critical orders. CDC sued, and IBM ended up paying damages to CDC worth $600 millions.

More users = breaking even:

Timesharing meant that a powerful mainframe’s computing time could be shared among a number of dumb terminals, without terminals noticing it.

The ability to timeshare the PLATO III system was quite intriguing — because that was where any profitability lay. It captured Bill Norris’ interest in taking PLATO worldwide.

Time sharing wasn’t an invention that was made as part of PLATO experiment, yet it had to be engineered for newer CDC 1604 computers, which despite being faster than ILLIAC 1, weren’t capable of time sharing.

It was vital, because it was unimaginable to secure a machine for every programmer/student/teacher on 24-hour access basis. Bitzer used SIMILLIAC, an ILLIAC simulation program that enabled CDC 1604 to timeshare just like ILLIAC. Additionally, Bitzer’s personal rapport with CDC salesperson Harold Brooks ensured that PLATO lab obtained an additional refurbished CDC 1604 for experimentation.

With all the smart networking engineering within campus, PLATO III could timeshare 20 terminals at a time.

When PLATO IV was almost ready for production with Plasma screen, touch sensitivity, auxiliary inputs and text to speech synthesizer, William Norris envisioned its spread to educate people in remote areas, thus equalizing access to education. He started providing CDC machines to the University campus for development, and along with that, put his own team that could learn PLATO programming and hardware-engineering capabilities from research people.

Due to his marketing efforts, he was able to sell PLATO IV terminals to ESCOM — a power plant in South Africa which used it to train its workers. Later on, Madadeni College near NewCastle adopted PLATO system to teach its students of school curriculum — most of which were from Zulu ancestry.

This was at a time when South African schools had rarely any electricity, yet PLATO terminals were being used to teach students in air-conditioned rooms. Oftentimes, students also connected with outside world using PLATO’s own chat application term-talk.

Contrary to his mainframe business that involved single enterprise order of 6-figure cost, here Norris would sell an entire array of geeky terminals for almost same cost to university campuses — be it in Quebec, Alberta, Delaware.

So what eventually lead to (CDC) PLATO’s disappearance?

PLATO never disappeared from Les Lieux de Mémoire — or this article wouldn’t have taken place. But its commercial fate was sealed.

Norris had acquired rights to market PLATO from the University (1976). But due to high costs of content authors that were hard to find outside University, system was already flying on contingency fuel when PLATO IV was in operation.

Single terminal cost $12000 to CDC.

When PLATO V came out, microprocessor revolution had begun, and PLATO V terminals too contained it — the very basic Intel 8080. It also had 16 KB memory — 8 K each for ROM and RAM. While this put some brain inside the previously dumb terminal and made graphics and animations lot better, it was ultimately all data: Educational content that Terminal couldn’t store. It did little to provide any cost advantages to CDC company for selling training materials.

Since PCs were just around, institutes grew reluctant to invest heavily into distributed systems, especially due to telecom tariffs. CDC’s authoring labor and hardware costs were prohibitively large. To cover it, during the early 1980s, CDC had to charge $50 / hour for content delivery alone, something on top of telephone bill that a client would have to pay.

And yet, William Norris had enormous faith in the system so as to proclaim:

50% of CDCs revenue could come from PLATO by 1985.

In 1980s, he was relentlessly marketing it in print and tele-media, including radio. One day his inner revolutionary would roll out a PLATO-based crop-information system for farmers. The other day he would put PLATO terminals inside his shareholder’s premises to re-affirm the lost faith.

Intrigued by the viability of his ad claims, Minneapolis Tribune held a survey to measure PLATO’s effectiveness. The survey results showed that:

While PLATO system was immensely popular among students of all ages, its overall effectiveness wasn’t above average of that of mainstream classroom teaching — conducted by humans.

It became a $600 million behemoth, and was finding it hard to find supporters within CDC. As Norris stepped down in 1986, CDC began to shrink in its overall business. The final attempt to save CDC’s PLATO was made by introducing Micro-PLATO — an all-in-one TUTOR capable terminal that did not involve any online content, and cost $5/hour to the end customer.

Yet it was costly, because CDC couldn’t find cost-effective tutors that matched the prowess of the University staff. An hour of content creation cost CDC $3,00,000. Final nail in the coffin was put with the PC suppliers breaking up: firstly Texas Instruments (TI-99/4A), and later Atari (Atari 8-bit family).

Here is leaf out of CDC PLATO’s brief involvement with TI:

image credits: http://www.mainbyte.com/ti99/software/plato/plato.html

In 1989, CDC stepped out of hardware business, and PLATO Learning became a public company, though it had completely changed from the original PLATO — now simply reduced to selling course content in CDs.

Interestingly, CDC’s PLATO, now known as Plato Learning, came across yet another acquisition by a private equity trader Thoma Bravo in 2010. After acquiring yet another competitor Archipelago Learning, it is today known as Edmentum.

But PLATO still lives…

If you were born in 80s, you will surely remember yahoo chat rooms during the 90s and early 2000. In my home country India, cybercafes (now almost extinct) complemented it by provided the luxury of Internet access. The mere joy of net surfing was compounded with the sense of community and togetherness.

PLATO created quite similar atmosphere in campuses during 70s, and even high school kids frequented University’s PLATO classrooms. PLATO as an educational software simulator wasn’t quite some success, but PLATO enormously succeeded-

as super-performant (perhaps the only) game platform
that, with multiplayer capability
as the only social network

Empire, a highly successful multi-player shooter game written by John Daleske clocked 3,00,000 playing hours between 1978 and 1985.

In 1974, Jim Bowery developed Spasim, most probable father of 3D first person shooter multiplayer online game based on space flight simulation, which supported 32 players at once.

Spasim, in turn, also became inspiration for probably world’s first Tank Simulator Game, also on PLATO, named Panzer.

And then there were Backgammon, Chess, Life, Galaxy….

Courtesy: http://daleske.com/plato/plato-culture.php

The quote says it more than anything: PLATO wasn’t simply about nerds programming in their cubicles or cushions — it was about real people meeting their counterparts. The above letter was given to John Deleske (creator of Empire game) by his Dean to work and program on PLATO at hours that were intended for system maintenance.

The real reason behind such a thriving community was far from commercial, where PLATO missed to make its mark. In 1976, when Norris purchased rights to commercialize PLATO brand with CDC, he paid the University with a CDC Cyber machine. The commercial fate that CDC met was described here above, and is also the first branch of PLATO bloodline below.

Image Copyrights: Nirav Bhatt

I was curious about NOVANET, and googling it did not help quite: Page 1 results were filled with a medical network, a marketing software company, an online library consortium, a VoIP provider, and such. On Page 2, I saw a lonesome youtube video selling Grade 6–12 learn-at-your-own-pace courseware online, uploaded by Pearson School — with no apparent mention of its legacy. Not quite an upgrade over 70s PLATO. Sources say that despite being highly profitable division of Pearson, they shut it down in 2015.

But how did PLATO become NOVANET?

Soon after CDC purchased rights to sell PLATO software under its own umbrella, University of Illinois had to come up with its own brand. University’s students & staff were the authors. But University lacked funds to maintain the Cyber machine it got in exchange for PLATO rights. At the same time CDC was finding it prohibitively costly to sell the courseware due to lack of its expertise in course-making.

At one point, Bitzer’s idea of taking it up commercially through satellites + telephone kicked in (incorporated as UCI — University Communications, Inc.), and this resulted in commercial conflicts between CDC and the University. Not sure what might have followed, but within some time Bitzer moved on for another position in 1989. UCI and university kept providing the courses authored within campus, and also expanded in other universities, but the momentum got lost.

Past 1994, UCI was renamed as NOVANET, and eventually got acquired by National Computer Systems, and finally Pearson Education in 2000.

The third and non-commercial branch was Cybis, which even to this day indulges the PLATO enthusiasts with terminal emulations of PLATO and much more.

What did they all gain?

They were not the days of software IPs. PLATO creators and PLATO authors did little to derive much financially from PLATO itself.

But the cerebral offshoots of PLATO were never amiss. Soon after TUTOR’s development, David Woolley developed PLATO Notes which eventually became Lotus Notes and ended up with IBM.

Courtesy: http://www.computerhistory.org/revolution/the-web/20/377/2154

The ideas that powered PLATO have been endlessly feeding even todays’ social media — forums (Stackoverflow), message boards (reddit), Chat (Yahoo messenger) and email. In the 80s, it was quite mainstream in tech media to mention PLATO’s educational achievements.

PLATO’s end consumers enjoyed it thoroughly, without being sold advertisements, and without being sold to companies themselves.

But what PLATO’s original developers got was neither profits, nor the joy of accomplishing the online education — PLATO’s primary goal.

Their take? Designing huge number of hardware pieces that talked to each other, sharing nostalgic tidbits on terminal simulations, reminiscing about whom they met online in 70s that altered their lives, write books, speak in interviews, create vintage fan art, or just shout to millennials: