The Friendly Orange Glow

by Brian Dear

  Bitzer was furious. In a letter to Bunderson, he said, “In reference to the document written by Ken Stetten and sanctioned and circulated by the MITRE Company, our position has always been to stop its circulation. However, Ken Stetten and MITRE evidently circulated enough of the documents before they changed their policy and decided to stop it, so that most people in the field are well acquainted with its contents. In the hard sciences, such documents become property for scrutiny and I suspect that this document will be treated no differently. We have never had, nor do we have now, plans to compare TICCIT and PLATO, but the MITRE document will force the scientific community to make the comparisons that Ken outlined. Ken Stetten and MITRE produced it, and they will have to live up to its contents.”

  MITRE invited two CERL staffers, Bruce Sherwood and Paul Tucker, to come out to Virginia to take a look at TICCIT firsthand. Before the trip, Sherwood did his own analysis of the TICCIT system architecture, and to his surprise could not figure out how it was possibly going to work. Sherwood believed that MITRE had overlooked some key system design considerations necessary to achieve the equivalent of PLATO’s Fast Round Trip: he believed they were going to run into problems just having a user type on the keyboard and see their text echoed on the screen within a reasonable fraction of a second. Tucker and Sherwood flew out to MITRE and got a demo. Sherwood was presented with an electronics lesson asking him to identify a particular component of an example electric circuit. He knew the correct answer was “resistor,” so he began typing “r” and then “e” and the rest of the word, but no “r” nor “e” nor the rest of the word appeared on the screen. Sherwood recalls the ensuing conversation going like this:

  SHERWOOD: Um, excuse me, why are the characters not echoing on the screen?

  MITRE: Oh, well, we haven’t implemented that part yet.

  SHERWOOD: But how are you going to do that? The way I figure it, there isn’t any way to do that.

  MITRE: Oh? Why do you think there’s a problem?

  [SHERWOOD explains what he believes the problem is.]

  MITRE: Oh, gee, maybe that is a problem. Well, maybe we’ll have to add another whole processor.

  —

  “Double the number of mainframes they’re going to use!” Sherwood would recall in horror years later. “And in front of Paul and me, bemused as we were, these people started brainstorming about how it might be possible to echo keys onto the screen, because they hadn’t really thought through whether there was a difficulty there. They could now see there was. I was shocked out of my mind. Because as I say they had actually almost kind of scurrilously attacked PLATO and Bitzer for not having done a careful analysis that would merit going through the next stage, whereas they had. And they couldn’t even echo keys on the screen. It was terrible. And it was sort of my first, starry-eyed introduction to the world of high-powered, Beltway-defense-contract-vaporware. It was quite an experience.”

  The TICCIT terminal Credit 23

  The surprises kept coming. BYU and MITRE were starting from scratch, and had no authoring language to develop lessons on TICCIT, whereas PLATO had over a decade’s experience going, starting out with machine code, evolving to FORTRAN to CATO to GENERAL and finally to TUTOR. PLATO had painfully paid its dues when it came to authoring tools. The TICCIT people had no such experience to fall back on. For a while they considered using the ALGOL language, but eventually that idea faded away, as it would have been expensive to repurpose a language that had never been designed for CAI. One day, as if to add to CERL’s growing apoplexy, BYU and MITRE reached out to Bitzer and Sherwood to mention that TICCIT still did not have an authoring language…might they use TUTOR? This resulted in a flurry of back-and-forth letters and phone calls, but in the end, perhaps in part to save face, the TICCIT people decided they would implement their own authoring language. TAL, the TICCIT Authoring Language, was the result.

  —

  One academic surprised by TICCIT was Michael Allen, who got to know the system and its creators during its early days. Allen, who directed CAI research and development at Ohio State, found that the big issue for Vic Bunderson back then was learner control. Should the student have a lot of control, or a little? For Bunderson, that was a central issue. Allen argued with him about that, telling him, “I don’t think there can be too much student control. What you can have is too much student control without enough guidance and feedback, but I don’t think you benefit by taking control of the student. If we want to have people who are good learners and going to learn independent of the system and take the good habits forward, then they need to have some discretion and control and then we have to help them practice good habits until they learn them well. To give them no choice doesn’t seem to me to be a very good approach.”

  Allen remembers Bunderson’s presentation at an educational computing conference in New York. “I was just amazed at his feeling,” he says, “that he had an educational paradigm that was so effective that it was time to implement hardware and software so that you even had a keyboard that specifically was limited to that instructional design.”

  “I remember my mouth just dropped open,” Allen recalls when he first saw TICCIT. “I thought, you can’t be serious.” The TICCIT keyboard looked like any other keyboard, except that it had three keys that stood out from the others, boldly declaring their take-it-or-leave-it pedagogical model: RULE, EXAMPLE, and PRACTICE. “I thought it was really premature,” Allen says, “to implement hardware that was really tailored to that specific approach.”

  If nothing else, TICCIT had one big thing going for it. Its creators had bought into the REP model so deeply, so thoroughly, so completely, and the system reflected this deep, thorough, complete buy-in so absolutely, as did the courseware written for the system, as did the data that spewed out from the system when students used it, that the research could not help but reflect that model 100 percent.

  —

  The National Science Foundation decided in the end to fund both PLATO and TICCIT, two diametrically opposed systems with 180-degree different personalities, philosophies, and system architectures. The fact that the projects were so radically different appealed to NSF. They already loved PLATO and had been familiar with Bitzer and Alpert for years. They recognized that TICCIT represented a wholly different approach, both at the scale (supporting a maximum of 128 color TV terminals running on a minicomputer, versus PLATO IV’s much heralded 4,096 terminals running on a supercomputer) and in terms of its instructional design model. “TICCIT was prepared,” says Arthur Melmed, “to demonstrate a certain kind of interaction in a relatively efficient way, and I thought that deserved a crack.”

  In 1972 both systems received roughly $5 million each from Congress.

  The race was on.

  9

  A Fork in the Road

  In July 1968, ARPA invited graduate students working on projects it was funding at universities around the country to attend a conference at UI’s Allerton House. One of the graduate student attendees was a brilliant computer-philosopher genius, viewed by some as a “wild man,” named Alan Kay, then working on his PhD at the University of Utah. He gave a talk on what he called the FLEX computer, which he’d started working on the year before and which formed the basis of his dissertation, “The Reactive Engine,” which he would finish in 1969. The presentation did not go over very well. Audience members, largely other computer science grad students, thought Kay’s ideas were crazy. The FLEX machine was Kay’s idea for a “personal computer,” a notion in 1968 (and years afterward) often met with laughter or worse, an idea still foreign to practically everyone involved in computing in the 1960s, including, importantly, the PLATO team. Kay’s dissertation would introduce FLEX as “a personal, reactive, mini-computer, which communicates in text and pictures by means of a keyboard, line-drawing CRT, and tablet.”

  Kay would later describe FLEX as “a small tabletop computer, and it didn’t have a very good user interface on it, and only after we had done a lot of it did we
realize the user interface was really the critical part. We already had the idea that somehow the novice programmer should be able to program on it and it had an object-oriented language on the first one that I designed but not terribly well-integrated into the user interface.” Kay was influenced by the work of Ivan Sutherland, also at the University of Utah, who was a pioneer in computer graphics, as well as Douglas Engelbart, who had long ago abandoned his plasma storage device research to embark on research into user interfaces and online software for “human augmentation,” resulting in, among other things, the invention of the mouse. Kay also stumbled upon Wallace Feurzieg’s and Seymour Papert’s LOGO work under way in Massachusetts. In Papert he found a kindred soul: Papert was someone who had worked with Jean Piaget and approached education and learning the way Kay did, views that were about as diametrically opposite to Skinner and the behaviorists as you could get.

  While attending the ARPA conference, Kay for the first time saw what Bitzer and company were up to with PLATO and the plasma display project at CERL. As for the PLATO III system and CERL’s plans for PLATO IV, Kay was not that impressed. “My take on that was very different from Bitzer’s,” he says. “From my standpoint as a computer scientist, I always thought that the PLATO architecture was stupid. What you need for anything that’s going to be interactive is guaranteed cycles. The way you get guaranteed cycles is putting them at the user, not trying to time-share them.” Kay and his colleague Adele Goldberg would summarize their distaste for “time-sharing” (the very word was anathema to them) in a famous essay entitled “Personal Dynamic Media,” published in 1977:

  Children really needed as much or more computing power than adults were willing to settle for when using a timesharing system. The best that timesharing has to offer is slow control of crude wire-frame green-tinted graphics and square-wave musical tones. The kids, on the other hand, are used to finger-paints, water colors, color television, real musical instruments, and records. If the “medium is the message,” then the message of low-bandwidth timesharing is “blah.”

  PLATO meant time-sharing, and Kay was religious in his dislike of it. But the prototype of PLATO’s plasma display, with its tiny, flicker-free pixels from which emanated the Orange Glow, was something else entirely. When Kay laid eyes on that prototype he was astounded. “I saw a one-inch-square lump of glass and neon gas in which individual spots would light up on command,” Kay says. “When I saw that flat screen display, I thought, oh boy!” Seeing that panel was an epiphany, something he would later describe as a “big whammy.” Kay had assumed that flat screen displays were still science fiction, devices that might come true at some point far off in the distant future, and therefore his FLEX machine would use, as he had stated in his dissertation, “a keyboard, line-drawing CRT, and tablet.” But all that changed when he laid eyes on the plasma prototype. He realized that he wasn’t going to need a CRT after all. Kay, a believer in Moore’s Law, could now see that the future would include affordable chips and flat displays, components that would make personal computing not only possible but inevitable. He spent the rest of the conference calculating when Moore’s Law would reach the point where all the processing and memory could fit on the back of the FLEX machine’s flat panel display.

  “People had talked about flat screen displays since the 1950s as a concept,” Kay says, “but I never really thought of it as a computer display until I realized that people could actually start building them like that….The thing that hit me on the plasma panel was that people could actually do thin-film deposition now over large areas and get away with it, and so it was only going to be a matter of time before you could actually do some decent display. So yeah, that had a lot to do with it. That immediately gave me a focus for thinking about user interface. And the crux of the thing was that I remembered a saying of [Marshall] McLuhan’s, which was, ‘I don’t know who discovered water, but it wasn’t a fish.’ And I realized that one of the problems, one of the reasons user interfaces were lousy, was we had adults trying to design for adults. And thinking they knew what they were doing but actually taking so many things for granted that it just wouldn’t work out. Somewhere around in there I realized that what Seymour had shown is that the computer is more than just a tool, it could be like a medium and extend into the world. Media are things you want to extend into the world of the child.”

  As 1968 wore on, Kay visited with Papert and the LOGO project in Massachusetts, and then in early December attended a presentation in Palo Alto, California, given by Douglas Engelbart of the Stanford Research Institute on his “NLS” or “oN-Line System.” Known in the history books as “The Mother of All Demos,” the event marked a turning point in thinking about computers, what they could be used for, and how they were best designed. Engelbart walked his rapt audience through a demonstration of a working system that would pave the way for every desktop computer in use today. It was an extraordinary, breathtaking demonstration, and convincingly depicted a future very different from the mainframe-based PLATO IV system, which was still four years away. In the minds of people like Kay and Engelbart in December 1968, PLATO was doomed, unless Bitzer and company embraced distributed computing and the idea of “guaranteed cycles,” which was then and would be for a long time largely anathema to CERL mainly on economic grounds. Soon, Kay’s vision for the FLEX computer, a desktop machine, morphed into a portable tablet, much like present-day iPads. He called this the “Dynabook”—his ultimate dream for a lightweight, portable, multimedia computer that incorporated the flat panel display technology he’d seen at Illinois.

  Two approaches, two different missions, two ways of thinking: two visions for computing collided and bounced away from each other in 1968. It was not the last time they would collide or the last time they would bounce away.

  —

  That same year, high-flying Xerox Corporation, one of the hottest technology companies of the 1960s, was looking to expand beyond photocopying by venturing into the computer business. Documents were no longer sexy. Information was the future. The company recruited Jack Goldman, a veteran head of research at Ford Motor Company, to assume the top research post at Xerox, which was freed up by the departing John Dessauer. Xerox CEO Peter McColough made Goldman an offer he couldn’t refuse. “The Xerox job meant money, stature, power,” wrote Douglas Smith and Robert Alexander in their Xerox PARC history, Fumbling the Future. Goldman was even offered a seat on Xerox’s board. He took the job.

  Xerox already had a research laboratory, located near its Rochester, New York, headquarters. Goldman toured it and met with the lab’s director. He was shocked to find that they did almost nothing with computers and had very little familiarity with digital technology. The lab was focused on the technology behind copying machines, Xerox’s cash-cow business. Then Goldman was hit with another surprise: the SDS acquisition.

  When Xerox explored the idea of getting into the computer business, it realized it couldn’t start from scratch—there was no way it was going to build its own systems. The competition, including IBM, Control Data, Honeywell, Sperry, Digital Equipment Corporation, and a long list of others, was far too far ahead, and Xerox would never catch up. The only way it was going to get into the computer business was to acquire an existing computer business. Control Data and DEC were considered but unobtainable. Then they looked at a hotshot young company called Scientific Data Systems (SDS), makers of a computer called the Sigma, and found it willing to be courted. In 1969, Xerox successfully acquired the company, for nearly $1 billion in stock, a massive sum at the time. Incredibly, Goldman, Xerox’s new chief technologist, had not been consulted, nor was he at all involved in the acquisition. He would have fought against it had he known, believing SDS a lackluster business.

  After seeing how weak Xerox was with digital computing at the copier research lab, it occurred to Goldman that even with SDS now under Xerox’s wing, a new, second research lab, one focused solely on computing, was needed. He wrote up a detailed proposal and presented it to the
board, which reacted negatively to it, but he managed to secure McColough’s go-ahead to proceed anyway. For recruiting ideas, Goldman turned to fellow Xerox board member Bob Sproull, a well-known physicist, former head of ARPA, and at this time provost of the University of Rochester. “I knew him well from the physics community,” says Goldman. “I took him on a plane ride since we were both going to New York and I said, ‘You got any suggestions on who I ought to go for?’ ”

  According to the generally accepted historical record, Goldman would pick a physicist named George Pake to be the director of the cutting-edge research lab. Here’s what Fumbling the Future, which Goldman himself would later recommend as the most accurate history of Xerox’s Palo Alto Research Center (PARC), had to say:

  Jack Goldman recruited a long-standing acquaintance of his named George Pake to set up and manage the proposed Xerox research center. While Goldman could, and would, continue to speak at corporate headquarters on behalf of the effort he’d inspired, he had too many responsibilities as the company’s chief scientist to operate the facility himself. He had to find someone else for that job, and Pake was his first choice.