by Brian Dear
Control Data had emerged in the 1960s as the preeminent manufacturer of supercomputers—for a while the 1604 was the world’s fastest. Then, with the help of the company’s genius engineer Seymour Cray, it launched the CYBER 6400, 6500, and 6600 line of supercomputers through the rest of that decade. Bitzer decided one of these would be the perfect machine to power his 4,096-terminal PLATO IV. By 1969, CDC had reached $1 billion in revenue for the first time. Its stock soared. The annual revenue would continue to climb all during the 1970s. The company expanded, adding tens of thousands of employees. NSF funded PLATO IV, CERL bought a CYBER 6500 supercomputer, and the future looked bright. It was time to take PLATO to the world, something CERL wanted to do but was neither chartered nor equipped to do. Norris and CDC were ready and willing.
—
One day in late 1969 or 1970, a group of Ohio State University faculty members flew out to Illinois to visit CERL. In terms of computer-aided instruction research, Ohio State was a big IBM user: IBM 360 mainframe for the hardware and IBM’s Coursewriter tools to write the courseware. Michael Allen headed up R&D for Ohio State’s CAI lab and consulted on the side with IBM to help them develop their educational systems. In addition, he says, “I was teaching faculty members the principles of teaching….The provost set up a program and said it was voluntary but it was like [a] real course, and you could only enroll in it if you were a member of the teaching faculty at the university, and I was their first professor that taught professors, so that was really fun.”
When Allen and the faculty members arrived at CERL, they got the usual dog-and-pony-show demo, including the Fruit Fly and Titrate lessons, as well as having Jack Stifle demo a prototype plasma panel, wires attached all over and looking like a miniature Frankenstein experiment. The engineering was impressive, but perhaps too impressive. “We all looked at what they were doing,” says Allen, “and we thought, it’s too bad that their engineering community isn’t coupled more tightly with an education community.”
Many CAI researchers around the country shared this sentiment. It was acknowledged that PLATO was interesting, but why so centered on engineering and not on education? “There’s some stuff that educators could use,” says Allen, “but…they’re really sidestepping all of what we considered were much more difficult, critical issues which had to deal with how people learn.” Allen would admit that the IBM systems at Ohio were not as sophisticated or as capable as what CERL was building, and the PLATO IV system would make them even less so, but still, he thought that despite IBM’s weaknesses, the Ohio researchers were able to support more effective and varied educational paradigms. “We saw them as a mirror image of us, they were just backwards, from where we were. We kind of rolled our eyes, thinking that gee, the people who were engineering these systems ought to be responding to educational needs, rather than trying to tell educators what they need.”
Then there was the issue of money. Like politicians, academic researchers need funding to survive, and given the costs of computer hardware, software, and staff to run it all, CAI in the late 1960s and early 1970s needed a lot of money. Ohio State had a strong relationship with IBM, but it still had to look for outside funding, and like many labs around the country, they would run increasingly into a brick wall when they approached the federal government for support. “When we would write for grant money to the National Science Foundation, we were always turned down because they said, ‘Well, we’re spending all our money on PLATO and TICCIT.’ And we said, ‘Those things that you’re spending money on are primarily engineering-based projects, and we think that you should be funding some more serious educational research projects.’ To which they’d say, ‘Well, those are serious educational projects.’ And we could never win that argument.”
NSF had simply bet on PLATO and TICCIT largely to the exclusion of all else. The degree to which they were PLATO and TICCIT believers boggled minds at competitive labs like Allen’s. “I remember a conversation I had with someone at NSF,” Allen says. “I was saying something about, well, they haven’t even evolved their system to the point where we could test a particular educational issue. And his response back to me: ‘They have because I have a terminal sitting right here at my desk.’ Oh, no wonder we can’t get you to think about anything else, you’re sitting there using PLATO, and you guys have been wrapped around their little finger.” He was happy NSF was pouring money into CAI, but he, and many other CAI researchers around the country, felt NSF was being nearsighted devoting so much of it to just PLATO and TICCIT. “They were kind of opposite ends and there was no funding of the middle ground,” Allen says.
Then, as PLATO IV emerged, Allen’s objections began to fade. He began to see what the fuss was about, including the impact that the new wave of creative young people had had on PLATO, transforming it in a short span of time from being a development-and-delivery platform for interactive educational lessons to being that plus a platform for rich and varied communication and collaboration. “Education is at its heart, communication,” says Allen, “and the strength of communication that PLATO had was so strong that not only could you develop good courseware, but you could also have this interpersonal communication system and you could do beyond that even and have real-time games. Its ceiling was so high in the level of communication that it provided a good basis for the development of [a] big variety of applications. You would never develop those kinds of things on TICCIT, for example. TICCIT had again these blinders on: ‘We’ve discovered the truth and the end of educational design, and now all we have to do is incarnate it in silicon and we’ve got it.’ It was just incredible to me.” Allen, like many others, including even some PLATO people, remained impressed with the video capability built into TICCIT, but the rest of what PLATO offered towered over the best of TICCIT.
At one of the many computer conferences in the early 1970s, Allen and Bitzer appeared on a panel together. Allen went first, and told the audience, “I think it’s important for the research community out there to appreciate what Don Bitzer has done.” Allen began to see Bitzer’s view as not telling educators what they need, but, rather, building as open and flexible a system as possible, so that educators could do whatever they wanted. “Instead of waiting for the laborious research on education to define exactly what systems requirements there are, and then try to build those, they’ve gone ahead and tried to build almost everything imaginable that educators might want, and then said, ‘Here it is, here’s your laboratory! You can do many things unimaginable before, try them out and see which ones are really important.’ ”
After a while, Allen changed his mind about PLATO altogether. Perhaps having seen with his own eyes what the extreme alternative—TICCIT—was actually turning into, and what limitations instructional designers were going to face trying to create courseware on that platform made him come over to the PLATO camp. “I started to be very, very appreciative of what Don was doing.”
By 1973, Control Data was looking everywhere for an educator they could bring on board to help with a commercial rollout of PLATO—someone from academia with serious CAI credentials. They wanted Michael Allen. Michael Allen wasn’t sure he wanted Control Data.
“They went after me,” Allen says. “They had done a nationwide search for someone to come in and my name kept coming up.”
Allen went up to Minneapolis for interviews. He told them he wasn’t sure he was a good choice because he’d been outspoken against PLATO. To his surprise, that was exactly what CDC wanted to hear. “The more I stated that, that I’d been critical of it, the more people at Control Data said, ‘You’re exactly the person we need. We hear from all of these educators at the University of Illinois who have been involved in the system and you know this is just the ultimate in computer-based instructional systems. We want to make a business of this and we need someone who is willing to ask tough questions and yet has educational credentials so you’re not just slinging mud or responding in an unknowledgeable way. You know what’s going on. We really, really, n
eed you.’ ”
CDC had hired the well-respected executive John Dammeyer away from IBM in the late 1960s and put him in charge of Control Data Europe. But then around 1972 Bill Norris handpicked him to lead the new effort to commercialize PLATO, and brought him to Minneapolis to form a team and get going. Dammeyer was keen on hiring Allen. “I was surprised at how receptive he was to me,” says Allen. “I actually went to Control Data on a sabbatical. I never told anybody at Control Data I was here on a sabbatical. Because I intended to just learn what goes on in corporate life, and see how they were thinking about using computers in training, and then I intended to go back, because I was really happy [at Ohio State]. And Dammeyer basically said, ‘What do you need?’ ”
Allen had been a mere graduate student a year and a half earlier—the ink was not yet dry on his PhD diploma. Now, Dammeyer offered him not just an office with a private bathroom, but an entire empty building. “Now, you got to staff up,” Allen recalls Dammeyer saying. “Who do you want? What do you need for computer equipment and stuff?”
“Basically, everything I had to scratch for and scream for at the university,” Allen says, “Dammeyer just called a truck and delivered it, I mean, everything I wanted, he gave me. And so, by the end of my first year, I was thinking, ‘Whew, this is a pretty cool place,’ so I ended up staying on.”
CDC knew that the PLATO system would need enhancements and new capabilities so it didn’t simply reflect the needs and interests of people at the University of Illinois, but rather related to CDC’s clients in the future. “That was kind of my job,” Allen recalls.
But there was something else: an area in which he felt PLATO was weak was at the same time one he felt was crucial to CAI’s success. “When I came in, I found out there was no activity going on with computer-managed instruction. All around the country we had seen that if you want to individualize instruction, or introduce computer-assisted instruction, you really needed the backbone of a management system in there. That was one of the things we started.”
—
John Dammeyer had already recruited Bob Morris and Bill Cole to help with the overall planning to determine the feasibility and costs of launching a major PLATO initiative, as well as researching the market potential, marketing strategy, product strategy, and overall direction the PLATO project would take to maximize success for the company. Morris and Dammeyer flew out to San Diego for an off-site brainstorming session that lasted several weeks. More trips were made to CERL, and more CERL personnel traveled up to Minneapolis to confer with colleagues at CDC. In April 1974, Dammeyer and Morris presented their comprehensive business plan for PLATO to William Norris and his executive committee. “The program was large in scope,” says Morris, and for the first time in Control Data’s history required the concurrent and integrated development of a complete system (e.g., hardware, software, terminals, communication networks, applications, languages, and educational content), and the business strategies for its entry into the market. Norris gave the project the green light.
Over the following months, Morris and Dammeyer picked a formidable team of bright, experienced people to begin the work of transforming a university-developed computer system into a commercially viable product. By 1974, they had gotten PLATO running on a part-time basis on a borrowed mainframe within the company, and a growing number of staffers had terminals. Bill Cole recalls a fad that emerged within CDC at its Arden Hills, Minnesota, facility, where much of the PLATO project was staffed. On one of his visits to Minnesota, he noticed that CDC executives and senior managers promptly turned into gamers at 5:15 p.m. each night. “Somebody said,” he says, “Okay everybody, Airfight time!” There were “real addicts,” says Cole. “I went and watched what they were doing, and what was happening was that over in the Tower [Control Data’s headquarters building in Bloomington, where, for exercise, William Norris walked fifteen flights up the stairs each day to his office], a whole bunch of people, including quite senior vice presidents, were hopping on Airfight. And they were shooting each other down.”
John Dammeyer Credit 34
One of the people brought into the PLATO project was Jock Hill, a proud, bald-headed Scotsman with a magnificent Scottish brogue and the baritone of Garrison Keillor, the kind of voice you imagine leading men into battle. Born in Dundee, he grew up in Inverness, and after graduating college pursued a career in electronics and applied mathematics, alternating jobs at Rolls-Royce and IBM across the 1950s and 1960s. He wound up in Brussels, Belgium, in the early 1970s working for Systems International, a Rolls-Royce subsidiary, where he crossed paths with John Dammeyer, who had had a stellar career at IBM (including a major feature article on business management in Life magazine in 1967). When William Norris appointed Dammeyer to begin a small task force to evaluate the PLATO opportunity, Dammeyer brought in all the best people he knew: Jock Hill, John Cundiff, Bob Morris (originally a high school dropout who “attended the college of IBM” for a decade, including working for Dammeyer), and an engineering manager named Bob Moe who had joined CDC in 1960 and spent most of the 1960s working for Seymour Cray on the 6600 and 7600 supercomputers at Cray’s fabled Chippewa Falls, Wisconsin, laboratory. A year after Hill joined Systems International, the company went bankrupt, and everyone, including Dammeyer, Cundiff, Moe, and Hill, was out of a job. CDC snapped up Dammeyer, who gave the rest of them an ultimatum, says Hill: “He’d cut us off dead, or we’d join him on his staff, so we joined Control Data together.”
It is a testament to Norris that he chose someone as capable as Dammeyer for the PLATO job. Great people want to work with, and wind up attracting, other great people, and Dammeyer had his share of great people to pick from for the PLATO project: people he knew, people he trusted, people who would get the job done. “Dammeyer was a genius,” says Hill. “Bob wasn’t quite the genius, but Bob Morris was a scientist. And he intended to follow the guidelines and produce the most inspired leadership that I’ve ever been subjected to. Just incredible type of leadership. It was an exciting time to be around there.” It was also fortuitous that CDC’s fledgling PLATO team equally respected their counterparts at CERL: Don Bitzer, Jack Stifle (a hardware engineering genius whom Hill would describe as being able to walk on water), Dominic Skaperdas (another hardware wizard), and the rest of the hardware and systems software teams. Teams of good people all around. To an eternal optimist like Don Bitzer, his belief in a future of one million terminals and 250 PLATO mainframes seemed not only attainable, but already a done deal. What could possibly go wrong?
—
One piece of technology CDC had to evaluate in the business plan for PLATO was the terminal. They had discovered that the most expensive component in the overall PLATO equation—the thing that was keeping the Holy Grail of costs, cost-per-student-contact-hour, high—was the PLATO IV terminal, the Rube Goldberg box made from over three thousand parts. In particular, CDC blamed the flat panel gas plasma display. As friendly and familiar as its Orange Glow might be, it was simply too expensive. Morris wanted to know if a cheaper terminal could be built, so he tasked people to find out.
Around this time, Dammeyer and Morris learned that TICCIT was for sale. “MITRE Corporation came to us,” says Morris, “suggesting that TICCIT was a superior technology to PLATO, and that we ought to consider putting our funding and our energies into TICCIT.” Dozens of companies were interested to one degree or another. Of particular interest to William Norris was TICCIT’s color television graphics and video support—features PLATO continued to sorely lack. However, in the end, Norris decided that to acquire the TICCIT project might come back to haunt them, in that the acquisition might be deemed anticompetitive—simply a means of eliminating a competitor. CDC passed. TICCIT wound up being sold to the government contractor Hazeltine Corporation.
Times had changed since Bitzer, Slottow, and Willson had invented the plasma panel a decade earlier. Moore’s Law had indeed come true, and was on a tear all during the 1970s: the cost of integrated circuits was
plummeting each year, while at the same time the capacity and power of the circuits were doubling. For instance, video RAM. Gone were the days of memory costing $2 per bit. PLATO’s 512 x 512 display, which, if driven from video RAM with a CRT instead of running on a gas plasma panel, no longer would cost hundreds of thousands of dollars—per terminal—for the 262,144 bits of memory. Still, CDC now had access to this amazing plasma technology, and directed some engineers to study it and figure out what kind of business could be made of it. Patents resulted from their research, but in the end CDC decided not to pursue the plasma display market, nor use a plasma display in its own PLATO terminal. That meant they would need to do something else, and a CRT with video RAM seemed the best bet. Could it be done cheaply?
Bob Morris Credit 35
“One of the disappointments to me,” says Morris, “was inside of Control Data my PLATO team was not allowed to develop the new terminal.” CDC had a division, the Terminal Systems Division (TSD), located in Roseville, Minnesota, devoted to building terminals for its customers. As is often the case when separate divisions of a company disagree with each other turf wars ensue. Morris’s team was allowed to do the design, to deliver a set of requirements, but TSD insisted on building the actual product. “So we got the team together,” says Morris, “we got the people from the Terminal Systems Division together. And I presented them with the requirements, and we used Jack Stifle to do functional design.” Morris told TSD he wanted the terminal in nine months from initial sketches to finished prototype. TSD said that was impossible, that it normally took them two and a half years. Morris was adamant: nine months.
“We came to an impasse,” says Morris. “By this time Jock had joined me, and Jock Hill had responsibility for the PLATO network and terminals, and we literally locked these people up for two weeks, every day, and we took, one at a time, every one of their objections about why it couldn’t be done in nine months, and at the end of two weeks they agreed they could do it in nine months.”
