PLATO (computer system)

Before the 1944 G.I. Bill that provided free college education to World War II veterans, higher education was limited to a minority of the US population, though only 9% of the population was in the military. The trend towards greater enrollment was notable by the early 1950s, and the problem of providing instruction for the many new students was a serious concern to university administrators. To wit, if computerized automation increased factory production, it could do the same for academic instruction.

The USSR's 1957 launching of the Sputnik I artificial satellite energized the United States' government into spending more on science and engineering education. In 1958, the U.S. Air Force's Office of Scientific Research had a conference about the topic of computer instruction at the University of Pennsylvania; interested parties, notably IBM, presented studies.

Around 1959, Chalmers W. Sherwin, a physicist at the University of Illinois, suggested a computerised learning system to William Everett, the engineering college dean, who, in turn, recommended that Daniel Alpert, another physicist, convene a meeting about the matter with engineers, administrators, mathematicians, and psychologists. After weeks of meetings they were unable to agree on a single design. Before conceding failure, Alpert mentioned the matter to laboratory assistant Donald Bitzer, who had been thinking about the problem, suggesting he could build a demonstration system.

Project PLATO was established soon afterwards, and in 1960, the first system, PLATO I, operated on the local ILLIAC I computer. It included a television set for display and a special keyboard for navigating the system's function menus;[5] PLATO II, in 1961, featured two users at once, one of the first implementations of multi-user time-sharing.[6]

The PLATO system was re-designed, between 1963 and 1969;[7] PLATO III allowed "anyone" to design new lesson modules using their TUTOR programming language, conceived in 1967 by biology graduate student Paul Tenczar. Built on a CDC 1604, given to them by William Norris, PLATO III could simultaneously run up to 20 terminals, and was used by local facilities in Champaign–Urbana that could enter the system with their custom terminals. The only remote PLATO III terminal was located near the state capitol in Springfield, Illinois at Springfield High School. It was connected to the PLATO III system by a video connection and a separate dedicated line for keyboard data.

PLATO I, II, and III were funded by small grants from a combined Army-Navy-Air Force funding pool. By the time PLATO III was in operation, everyone involved was convinced it was worthwhile to scale up the project. Accordingly, in 1967, the National Science Foundation granted the team steady funding, allowing Alpert to set up the Computer-based Education Research Laboratory (CERL) at the University of Illinois Urbana–Champaign campus. The system was capable of supporting 20 time-sharing terminals.

A standard keyboard for a PLATO IV terminal, circa 1976

In 1972, with the introduction of PLATO IV, Bitzer declared general success, claiming that the goal of generalized computer instruction was now available to all. However, the terminals were very expensive (about $12,000). The PLATO IV terminal had several major innovations:

• Plasma Display Screen: Bitzer's orange plasma display, incorporated both memory and bitmapped graphics into one display. The display was a 512×512 bitmap, with both character and vector plotting done by hardwired logic. It included fast vector line drawing capability, and ran at 1260 baud, rendering 60 lines or 180 characters per second. . Users could provide their own characters to support rudimentary bitmap graphics.
• Touch panel: A 16×16 grid infrared touch panel, allowing students to answer questions by touching anywhere on the screen.
• Microfiche images: Compressed air powered a piston-driven microfiche image selector that permitted colored images to be projected on the back of the screen under program control.
• A hard drive for Audio snippets: The random-access audio device used a magnetic disc with a capacity to hold 17 total minutes of pre-recorded audio.[8] It could retrieve for playback any of 4096 audio clips within 0.4 seconds. By 1980, the device was being commercially produced by Education and Information Systems, Incorporated with a capacity of just over 22 minutes.[9]
• A Votrax voice synthesizer
• The Gooch Synthetic Woodwind (named after inventor Sherwin Gooch), a synthesizer that offered four-voice music synthesis to provide sound in PLATO courseware. This was later supplanted on the PLATO V terminal by the Gooch Cybernetic Synthesizer, which had sixteen voices that could be programmed individually, or combined to make more complex sounds.

Bruce Parello, a student at the University of Illinois in 1972, created the first digital emojis on the PLATO IV system.[10]

Early in 1972, researchers from Xerox PARC were given a tour of the PLATO system at the University of Illinois. At this time, they were shown parts of the system, such as the Insert Display/Show Display (ID/SD) application generator for pictures on PLATO (later translated into a graphics-draw program on the Xerox Star workstation); the Charset Editor for "painting" new characters (later translated into a "Doodle" program at PARC); and the Term Talk and Monitor Mode communications programs. Many of the new technologies they saw were adopted and improved upon, when these researchers returned to Palo Alto, California. They subsequently transferred improved versions of this technology to Apple Inc.

As PLATO IV reached production quality, William Norris (CDC) became increasingly interested in it as a potential product. His interest was twofold. From a strict business perspective, he was evolving Control Data into a service-based company instead of a hardware one, and was increasingly convinced that computer-based education would become a major market in the future. At the same time, Norris was troubled by the unrest of the late 1960s, and felt that much of it was due to social inequalities that needed to be addressed. PLATO offered a solution by providing higher education to segments of the population that would otherwise never be able to afford a university education.

Norris provided CERL with machines on which to develop their system in the late 1960s. In 1971, he set up a new division within CDC to develop PLATO "courseware", and eventually many of CDC's own initial training and technical manuals ran on it. In 1974, PLATO was running on in-house machines at CDC headquarters in Minneapolis, and in 1976, they purchased the commercial rights in exchange for a new CDC Cyber machine.

Using the CDC Plato network, c. 1979-1980, with an IST-II terminal

CDC announced the acquisition soon after, claiming that by 1985, 50% of the company's income would be related to PLATO services. Through the 1970s, CDC tirelessly promoted PLATO, both as a commercial tool and one for re-training unemployed workers in new fields. Norris refused to give up on the system, and invested in several non-mainstream courses, including a crop-information system for farmers, and various courses for inner-city youth. CDC even went as far as to place PLATO terminals in some shareholder's houses, to demonstrate the concept of the system.

In the early 1980s, CDC started heavily advertising the service, apparently due to increasing internal dissent over the now $600 million project, taking out print and even radio ads promoting it as a general tool. The Minneapolis Tribune was unconvinced by their ad copy and started an investigation of the claims. In the end, they concluded that while it was not proven to be a better education system, everyone using it nevertheless enjoyed it, at least. An official evaluation by an external testing agency ended with roughly the same conclusions, suggesting that everyone enjoyed using it, but it was essentially equal to an average human teacher in terms of student advancement.

Of course, a computerized system equal to a human should have been a major achievement, the very concept for which the early pioneers in CBT were aiming. A computer could serve all the students in a school for the cost of maintaining it, and wouldn't go on strike. However, CDC charged $50 an hour for access to their data center, in order to recoup some of their development costs, making it considerably more expensive than a human on a per-student basis. PLATO was, therefore, a failure as a profitable commercial enterprise, although it did find some use in large companies and government agencies willing to invest in the technology.

An attempt to mass-market the PLATO system was introduced in 1980 as Micro-PLATO, which ran the basic TUTOR system on a CDC "Viking-721"[11] terminal and various home computers. Versions were built for the TI-99/4A, Atari 8-bit family, Zenith Z-100 and, later, Radio Shack TRS-80 and IBM Personal Computer. Micro-PLATO could be used stand-alone for normal courses, or could connect to a CDC data center for multiuser programs. To make the latter affordable, CDC introduced the Homelink service for $5 an hour.

Norris continued to praise PLATO, announcing that it would be only a few years before it represented a major source of income for CDC as late as 1984. In 1986, Norris stepped down as CEO, and the PLATO service was slowly killed off. He later claimed that Micro-PLATO was one of the reasons PLATO got off-track. They had started on the TI-99/4A, but then Texas Instruments pulled the plug and they moved to other systems like the Atari, who soon did the same. He felt that it was a waste of time anyway, as the system's value was in its online nature, which Micro-PLATO lacked initially.

Bitzer was more forthright about CDC's failure, blaming their corporate culture for the problems. He noted that development of the courseware was averaging $300,000 per delivery hour, many times what the CERL was paying for similar products. This meant that CDC had to charge high prices in order to recoup their costs, prices that made the system unattractive. The reason, he suggested, for these high prices was that CDC had set up a division that had to keep itself profitable via courseware development, forcing them to raise the prices in order to keep their headcount up during slow periods.

A PLATO V terminal in 1981, displaying RankTrek application, one of the first to combine simultaneous local microprocessor-based computing with remote mainframe computing. The monochromatic plasma display's characteristic orange glow is illustrated. Infrared sensors mounted around the display watch for a user's touch screen input.

Intel 8080 microprocessors were introduced in the new PLATO V terminals. They could download small software modules and execute them locally. It was a way to augment the PLATO courseware with rich animation and other sophisticated capabilities.[12]

In 1969, G. David Peters began researching the feasibility of using PLATO to teach trumpet students to play with increased pitch and rhythmic precision.[21] He created an interface for the PLATO III terminal. The hardware consisted of (1) filters that could determine the true pitch of a tone, and (2) a counting device to measure tone duration. The device accepted and judged rapid notes, two notes trilled, and lip slurs. Peters demonstrated that judging instrumental performance for pitch and rhythmic accuracy was feasible in computer-assisted instruction.[22]

By 1970, a random access audio device was available for use with PLATO III.[9]

In 1972, Robert W. Placek conducted a study that used computer-assisted instruction for rhythm perception.[23] Placek used the random access audio device attached to a PLATO III terminal for which he developed music notation fonts and graphics. Students majoring in elementary education were asked to (1) recognize elements of rhythm notation, and (2) listen to rhythm patterns and identify their notations. This was the first known application of the PLATO random-access audio device to computer-based music instruction.

Study participants were interviewed about the experience and found it both valuable and enjoyable. Of particular value was PLATO’s immediate feedback. Though participants noted shortcomings in the quality of the audio, they generally indicated that they were able to learn the basic skills of rhythm notation recognition.[24]

These PLATO IV terminal included many new devices and yielded two notable music projects:

By the mid-1970s, James O. Froseth (University of Michigan) had published training materials that taught instrumental music teachers to visually identify typical problems demonstrated by beginning band students.[25] For each instrument, Froseth developed an ordered checklist of what to look for (i.e., posture, embouchure, hand placement, instrument position, etc.) and a set of 35mm slides of young players demonstrating those problems. In timed class exercises, trainees briefly viewed slides and recorded their diagnoses on the checklists which were reviewed and evaluated later in the training session.

In 1978, William H. Sanders adapted Froseth’s program for delivery using the PLATO IV system. Sanders transferred the slides to microfiche for rear-projection through the PLATO IV terminal’s plasma display. In timed drills, trainees viewed the slides, then filled in the checklists by touching them on the display. The program gave immediate feedback and kept aggregate records. Trainees could vary the timing of the exercises and repeat them whenever they wished.

Sanders and Froseth subsequently conducted a study to compare traditional classroom delivery of the program to delivery using PLATO. The results showed no significant difference between the delivery methods for a) student post-test performance and b) their attitudes toward the training materials. However, students using the computer appreciated the flexibility to set their own practice hours, completed significantly more practice exercises, and did so in significantly less time.[26]

In 1967, Allvin and Kuhn used a four-channel tape recorder interfaced to a computer to present pre-recorded models to judge sight-singing performances.[27]

In 1969, Ned C. Deihl and Rudolph E. Radocy conducted a computer-assisted instruction study in music that included discriminating aural concepts related to phrasing, articulation, and rhythm on the clarinet.[28] They used a four-track tape recorder interfaced to a computer to provide pre-recorded audio passages. Messages were recorded on three tracks and inaudible signals on the fourth track with two hours of play/record time available. This research further demonstrated that computer-controlled audio with four-track tape was possible.[29]

In 1979, Williams used a digitally controlled cassette tape recorder that had been interfaced to a minicomputer (Williams, M.A. "A comparison of three approaches to the teaching of auditory-visual discrimination, sight singing and music dictation to college music students: A traditional approach, a Kodaly approach, and a Kodaly approach augmented by computer-assisted instruction," University of Illinois, unpublished). This device worked, yet was slow with variable access times.

In 1981, Nan T. Watanabe researched the feasibility of computer-assisted music instruction using computer-controlled pre-recorded audio. She surveyed audio hardware that could interface with a computer system.[22]

Random-access audio devices interfaced to PLATO IV terminals were also available. There were issues with sound quality due to dropouts in the audio.[30] Regardless, Watanabe deemed consistent fast access to audio clips critical to the study design and selected this device for the study.

Watanabe’s computer-based drill-and-practice program taught elementary music education students to identify musical instruments by sound. Students listened to randomly selected instrument sounds, identified the instrument they heard, and received immediate feedback. Watanabe found no significant difference in learning between the group who learned through computer-assisted drill programs and the group receiving traditional instruction in instrument identification. The study did, however, demonstrate that use of random-access audio in computer-assisted instruction in music was feasible.[31]

By 1988, with the spread of micro-computers and their peripherals, the University of Illinois School of Music PLATO Project was renamed The Illinois Technology-based Music Project. Researchers subsequently explored the use of emerging, commercially available technologies for music instruction until 1994.

Educators and students used the PLATO System for music instruction at other educational institutions including Indiana University, Florida State University, and the University of Delaware. Many alumni of the University of Illinois School of Music PLATO Project gained early hands-on experience in computing and media technologies and moved into influential positions in both education and the private sector.

The goal of this system was to provide tools for music educators to use in the development of instructional materials, which might possibly include music dictation drills, automatically graded keyboard performances, envelope and timbre ear-training, interactive examples or labs in musical acoustics, and composition and theory exercises with immediate feedback.[32] One ear-training application, Ottaviano, became a required part of certain undergraduate music theory courses at Florida State University in the early 1980s.

Another peripheral was the Votrax speech synthesizer, and a "say" instruction (with "saylang" instruction to choose the language) was added to the Tutor programming language to support text-to-speech synthesis using the Votrax.

During the period when CDC was marketing PLATO, the system began to be used internationally. South Africa was one of the biggest users of PLATO in the early 1980s. Eskom, the South African electrical power company, had a large CDC mainframe at Megawatt Park in the northwest suburbs of Johannesburg. Mainly this computer was used for management and data processing tasks related to power generation and distribution, but it also ran the PLATO software. The largest PLATO installation in South Africa during the early 1980s was at the University of the Western Cape, which served the "native" population, and at one time had hundreds of PLATO IV terminals all connected by leased data lines back to Johannesburg. There were several other installations at educational institutions in South Africa, among them Madadeni College in the Madadeni township just outside Newcastle.

This was perhaps the most unusual PLATO installation anywhere. Madadeni had about 1,000 students, all of them who were original inhabitants i.e. native population and 99.5% of Zulu ancestry. The college was one of 10 teacher preparation institutions in kwaZulu, most of them much smaller. In many ways Madadeni was very primitive. None of the classrooms had electricity and there was only one telephone for the whole college, which one had to crank for several minutes before an operator might come on the line. So an air-conditioned, carpeted room with 16 computer terminals was a stark contrast to the rest of the college. At times the only way a person could communicate with the outside world was through PLATO term-talk.

For many of the Madadeni students, most of whom came from very rural areas, the PLATO terminal was the first time they encountered any kind of electronic technology. Many of the first-year students had never seen a flush toilet before. There initially was skepticism that these technologically illiterate students could effectively use PLATO, but those concerns were not borne out. Within an hour or less most students were using the system proficiently, mostly to learn math and science skills, although a lesson that taught keyboarding skills was one of the most popular. A few students even used on-line resources to learn TUTOR, the PLATO programming language, and a few wrote lessons on the system in the Zulu language.

PLATO was also used fairly extensively in South Africa for industrial training. Eskom successfully used PLM (PLATO learning management) and simulations to train power plant operators, South African Airways (SAA) used PLATO simulations for cabin attendant training, and there were a number of other large companies as well that were exploring the use of PLATO.

The South African subsidiary of CDC invested heavily in the development of an entire secondary school curriculum (SASSC) on PLATO, but unfortunately as the curriculum was nearing the final stages of completion, CDC began to falter in South Africa—partly because of financial problems back home, partly because of growing opposition in the United States to doing business in South Africa, and partly due to the rapidly evolving microcomputer, a paradigm shift that CDC failed to recognize.

In August 2004, a version of PLATO[38] corresponding to the final release from CDC was resurrected online. This version of PLATO runs on a free and open-source software emulation of the original CDC hardware called Desktop Cyber. Within six months, by word of mouth alone, more than 500 former users had signed up to use the system. Many of the students who used PLATO in the 1970s and 1980s felt a special social bond with the community of users who came together using the powerful communications tools (talk programs, records systems and notesfiles) on PLATO.

The PLATO software used on Cyber1 is the final release (99A) of CYBIS, by permission of VCampus. The underlying operating system is NOS 2.8.7, the final release of the NOS operating system, by permission of Syntegra (now British Telecom [BT]), which had acquired the remainder of CDC's mainframe business. Cyber1 runs this software on the Desktop Cyber emulator. Desktop Cyber accurately emulates in software a range of CDC Cyber mainframe models and many peripherals.[39]

Cyber1 offers free access to the system, which contains over 16,000 of the original lessons, in an attempt to preserve the original PLATO communities that grew up at CERL and on CDC systems in the 1980s. The load average of this resurrected system is about 10–15 users, sending personal and notesfile notes, and playing inter-terminal games such as Avatar and Empire (a Star Trek-like game), which had both accumulated more than 1.0 million contact hours on the original PLATO system at UIUC.[38]