The Friendly Orange Glow

by Brian Dear

  —

  Alpert stayed on as dean of the Graduate College at Illinois. George Pake made a pilgrimage to CERL shortly after he took the reins of Xerox PARC. “I had heard about it, and I wanted to see what I could see. We were already thinking a little bit about distributed computing, and I was frankly not too keen on operating everything as a terminal off a shared mainframe.”

  CERL and PARC went their separate ways. PARC would go the way of distributed computing, CERL would stick to the path of the centralized mainframe. Alan Kay had once famously said, “The best way to predict the future is to invent it.”

  Both labs remained determined to invent very different futures.

  —

  From time to time the labs would check in on each other over the ensuing years.

  By 1972 a Xerox PARC team, led by Alan Kay, had built the Alto computer, a prototype of a “personal” computer with a graphical, mouse-driven, windowing display, running in an object-oriented Smalltalk programming environment. Kay and Pat Suppes’s former graduate student Adele Goldberg continued to keep an eye on PLATO. CERL and PARC began several years of collegial, open channels, with each lab inviting the other to send people out and see and learn and exchange ideas.

  Kay and Goldberg went to CERL and met with a number of staffers there, including Roger Johnson, with whom, over glasses of beer, they had “tremendous battles,” Johnson says, debating the advantages and disadvantages of mouse interfaces versus touch screens, as well as plasma displays versus other, cheaper display solutions. CERL had different, more immediate-use cases for PLATO than PARC had for the Alto. PLATO IV terminals had to be workhorses, withstanding many hours of use per day by many different people, some of whom would not be kind to the terminal. A mouse, stylus, light pen, or other handheld device would mean one more thing that could break or suffer abuse at the hands of users. No, CERL had decided that the simplest user interface for a young child—or any PLATO user, for that matter—was the finger. In 2007 Steve Jobs, long a proponent of using mice, would finally echo the exact same rationale CERL had espoused thirty-five years earlier, when he introduced the revolutionary touch screen–based iPhone: “We’re going to use the best pointing device in the world. We’re going to use the pointing device that we’re all born with. We’re born with ten of them. We’re going to use our fingers.” But in 1972, PARC was not convinced of the benefits of touch screens, and they stuck to using a mouse.

  Adele Goldberg found in the touch screen an effective interface for children, at least when doing “gross manipulations” such as touching a picture of an animal and dropping it in a picture of a bucket. But the 16 x 16 grid of touch-sensitive regions on a PLATO IV terminal was far too low-resolution for any detailed drawing or other interactions. PARC was less complimentary on PLATO’s graphics capabilities, which led to more arguments between PARC people and CERL staffers like Paul Tenczar.

  At one point, Tenczar hopped on his motorcycle and rode out to California to visit PARC. David Frankel, CERL’s precocious, teenage systems programmer, flew out to join him, and for two weeks they hung out at PARC playing with Alto computers attempting unsuccessfully to use Smalltalk to create the kinds of interactions that they found easy to create in TUTOR. “Smalltalk was very, very pure,” says Frankel. “Smalltalk is very, very basic and you have to build up. Now, you can argue that really that’s the core of becoming some sort of object-oriented language where you build a bunch of small objects, and then once you have those you can reuse them and other people can use them and so forth. But it was pretty cumbersome to dive in and get started with that language. Whereas I think a language like TUTOR you could much more quickly in ten commands, boom boom boom, have a little lesson that would invoke spell checking and answer judging and all sorts of other fancy things that were buried inside those command subroutines.”

  The trips continued between the labs, at one point Bruce Sherwood and Rick Blomme visiting PARC. While the visits were cordial, it was impossible to ignore the vast difference between the labs’ philosophies and missions. They were simply set up for different reasons and the pace and output of each lab’s research reflected those differences. PARC found much of what CERL was doing interesting, and some of the software tools that CERL people had created influenced similar tools that would pop up on the Alta, for example PLATO’s character set editor that let you design new fonts. But on their respective system architecture designs—CERL’s being thousands of dumb terminals connected to a supercomputer time-sharing mainframe, and PARC’s being networked desktop computers with all the power at the desktop—neither lab could fathom why the other lab was doing what it was doing.

  “We were explaining to him how it was all gonna go,” says Kay. “What the future was going to be like, and Paul Tenczar looked at me and said, ‘YOU ARE A MADMAN!’ It was great. I mean, he really got upset, incredibly upset.”

  “David Frankel and I went out there and had an enormous amount of difficulty trying to understand the system,” says Tenczar. “There’s very little documentation. It was a very open atmosphere out there, very nice, Alan Kay very graciously opened up the whole place to us. We tried to write some lesson materials as we would write on PLATO and it was almost impossible. You had to start off by defining what numbers were, in Smalltalk. You know, one plus one did not equal two; you had to define what one plus one equaled. And well, that’s great, when you’re a mathematician studying your navel. But it has not much practical importance, when we were interested in doing other things. We were interested in doing biology, foreign languages, stuff like that.”

  Kay and Goldberg also were not impressed that children could not program on PLATO, that everything was programmed for them. This was Seymour Papert’s essential argument: kids should program the computers, not the other way around.

  As for their respective choices of programming languages, the two labs could never see eye-to-eye. Tenczar despised Smalltalk. Kay despised TUTOR. “It was always a terrible language design,” Kay says. “People who did it like Tenczar never understood how to design a language.”

  “There was a fair amount of religious hostility,” says Bruce Sherwood, “between the people working on PLATO and people who were card-carrying professional computer scientists….It was the contrast between the pure aesthete and the technician or engineer getting his hands dirty. We undoubtedly suffered from the fact that we did not have anybody in the group who had anything like a real computer science background. Plenty of very good people on the hardware side who were electrical engineers and computer engineers, and there’s no question there….People who were working on the software, on the system software, were people who came from the sciences and had used computers in the sciences for scientific purposes, so were self-taught in the use of computers. But did not know any computer science theory to speak of.”

  TUTOR did not fit into any model of a proper computer language that a computer science purist would find acceptable. Says Paul Koning, who was a systems programmer at CERL while attending UI, “To borrow a term from linguistics, I’d probably have to call it a ‘language isolate.’ It’s the programming analog of Basque. In other words, there really isn’t anything like it that I can think of. Syntactically it’s faintly like FORTRAN, which makes historic sense. But only faintly.”

  Dave Liddle, who had worked at Owens-Illinois to develop graphics drivers for the plasma panel and then moved out to Xerox PARC, believes that a clash between PARC and CERL was inevitable. “In my view,” he says, “the spirit of TUTOR that I felt was good, was that it was an attempt to match language elements to the pedagogical needs of the people writing the programs. Thus, it was written in a gentle form of educational jargon that wasn’t particularly bad. The problem was…it jumped over all the hard lessons of computer science. So as a result, it actually did contain a number of things that we don’t put into programming languages that are done by people who are computer scientists….The usual problem is that computer scientists per se, sit on their ass, and d
on’t work on the interesting applications, so somebody else who doesn’t know these things comes along and does it. For example, HTTP and HTML make all the same mistakes, because they were done by a desperate physicist who couldn’t wait any longer, and they weren’t done by a computer scientist. So, guess what, was that important, or what? Does everybody like it and use it and has it changed the world? Yes. It still has all these problems in it, because if computer scientists had focused on applications more, somebody would have done this right. And the same thing’s true for TUTOR. So you have to look at TUTOR and say, well, at least these people got off their butt, and built this thing, and it was learnable, and so people learned it. As any kind of a language design purist, you have objections that you can raise. But as a practical matter, it did get done. And it worked. And people who learned it didn’t know they weren’t supposed to dislike it, they just learned it, and it worked fine….Smalltalk didn’t have any of that stuff. I mean, Smalltalk was carefully language-designed…so it avoided all those things, so it’s crystal-clear why you couldn’t have a more dialectical combination of Paul Tenczar and say Alan Kay and Adele Goldberg.”

  Years later, Tenczar would admit that there were aspects that he and David Frankel saw at PARC, like the mice and windowing displays and other desktop graphical interface features—the very things that blew Steve Jobs away when he saw them in 1979—that should have had a lot more impact on them (and in theory on PLATO) but did not. “I think that David and I really missed what was going on at Xerox,” he says. “I think we both missed it. Maybe it was more my fault than his, in that I was the senior person in PLATO.”

  10

  Lessons Learned

  To make way for the transition from PLATO III to PLATO IV, the aging CDC 1604 would be phased out and replaced by a genuine supercomputer, a Control Data 6400. In addition to the new mainframe, there would be many new disk drives—each in that era the size of washing machines—as well as racks of communications and networking equipment, peripheral processing units, and special air-conditioning equipment to keep everything from overheating. Add to all that the arrival of PLATO IV terminals from Magnavox, sporting the brand-new Owens-Illinois–made plasma displays, plus all the wiring and phone connections. PLATO IV was far beyond PLATO III. CERL was for all intents and purposes starting over, finally deploying at the large, meaningful scale it had long dreamed of. A scale orders of magnitude beyond anything Suppes and Atkinson were doing out at Stanford, or MITRE and BYU planned to do with TICCIT, or IBM hoped to do with the 1500 system. PLATO IV’s scale called for 4,096 terminals connected all over the campus, the state, and the country, if not the world. And that was just the CERL system. There were bound to be other PLATO systems in the future.

  Monumental efforts were under way not only on the hardware front, but with system software and utility programs, much of which were going to have to be rewritten from their PLATO III counterparts. Countless new programs would also have to be written, including an editor for authors to program their TUTOR lessons that accommodated the larger 512 x 512 plasma display. Initially, much of the system-level programs that ran PLATO were written in COMPASS, CDC’s assembly language, making it cumbersome and difficult to make changes. CERL decided to rewrite existing system programs in TUTOR itself, by extending the TUTOR language to have privileged “system commands” that could do dangerous things like read and write directly to memory or disk, log users in or out, etc. It would take months, years even, to move everything over to TUTOR, but the effort was worth it. By creating system programs in TUTOR, operators and systems staff could interact far more productively with PLATO by taking advantage of all of the features TUTOR offered for friendlier displays, listing out information, and drawing lines and graphs, for instance.

  All of the functionality of PLATO IV did not appear overnight, however, and the early days tended to be bumpy for users. (CERL would not establish formal “prime-time” hours of service until 1974.) Ruth Chabay, an author of chemistry lessons, tried to run a class of students in room 203b on the second floor of CERL. Room 203b was the “author’s room,” but was also for a while the only room full of the new PLATO IV terminals. It could be booked by a department for its students to take assigned lessons. “Trying to run a class in room 203b,” she says, “was a challenge worthy of an Olympic athlete and a career diplomat. My appearance in the room was invariably greeted with groans from the authors working there, who knew they would be asked to leave. Getting students into the correct lesson meant helping each of them type the file name—there were no student records, routers, and of course no restarts. This meant that when, halfway through the class session, the systems staff decided to reload [reboot the mainframe] to try out a change…I’d have to sprint up the stairs to the computer room, and come panting in the door prepared to offer cookies, pizza, or other bribes to persuade them to keep the system up until the class ended in half an hour. Then I’d dash back down to 203 to throw out the persistent young authors who had snuck back into the classroom. Of course, the system invariably did crash.”

  Chabay had to resort to adding a special code in her chemistry lessons that students could enter using the TERM key, so that should the system crash, when it came back up they could more or less return to where they’d left off. “Between the desperate dashes up the stairs and the negotiations with disgruntled authors, one was glad that classes were usually scheduled for only an hour.”

  —

  Nineteen seventy-two was a turbulent year on the campus of the University of Illinois. Much of that turbulence was due to a report, written by assistant provost Barry Munitz, which laid out sweeping recommendations for restructuring the administration of the university. The heady days of the 1960s were over. Richard Nixon was in the White House. Belts were being tightened everywhere. Munitz urged the university to modernize its organizational chart, optimizing communications and productivity by pruning wasteful and redundant reporting structures. Munitz had been brought in by John E. Corbally Jr., who had taken over as president of the university when David Dodds Henry, long an enthusiastic supporter of Alpert, Bitzer, and PLATO, stepped down the year before. Unfortunately for Dan Alpert, the Munitz report recommended gutting the Graduate College, its functions and areas of responsibility to be divvied up by existing and new administrative posts, one of which would eventually be manned by Munitz himself. In addition, a senior faculty committee had been formed to consider plans for campus reorganization, and unfortunately for Alpert, members of the committee did not see eye-to-eye with him or his way of running the Graduate College. It was as if all the karma from ripping PLATO out of CSL to form CERL to keep PLATO under Alpert’s wing, as well as summarily dismissing Larry Stolurow and his SOCRATES project, were now coming back to haunt him. PLATO IV had just received its millions of dollars from the National Science Foundation, and yet Alpert would not be overseeing Bitzer and CERL any longer.

  “Dan was very supportive of what Don Bitzer was doing,” says George Russell. “I guess not everybody was, but Dan hung in there, he was a good man and he hung in there, and kept it on track. So Dan is owed a lot of credit for fighting a good battle for Don Bitzer.” Despite that long track record, resulting in massive NSF funding for PLATO IV, the university’s politics, long tolerating if not openly supportive of the effort, could no longer be counted on. “Dan Alpert’s more industrial research orientation,” says Jack Desmond, “finally got him into trouble as dean of the Graduate College, because even though he was enamored of the hard science community, he was somewhat critical of other aspects of the university’s complexion. He became kind of hard to deal with, irascible, judgmental, outspoken about the quality of university research, and finally, some senior people asked that he be removed because he was not ‘one of us.’ ”

  All through the spring and into the summer of 1972, The Daily Illini reported new scoops on the Munitz revolution under way on the campus. Alpert, it was reported in May 1972, was out as dean of the Graduate College, and he announc
ed he would resign on September 1, on which date he would start two new posts. George Russell would take over as acting dean of a Graduate College that would soon be dismantled. Alpert was offered a consolation prize of sorts, directorship of the Center for Advanced Study, in no uncertain terms a demotion from which Alpert arguably never recovered. In addition Bitzer offered Alpert, still eager to oversee the success of the PLATO project, the role of associate director of CERL—reporting to Bitzer—wherein he would contribute to long-term strategy and oversee the rollout of PLATO IV in community colleges around the Chicago area, one of the key components of the NSF proposal. Alpert accepted. Bitzer was no longer Alpert’s protégé. He was finally on his own. “He was clearly,” says Desmond, “clearly, captain of his own ship.” Jack Peltason, then university chancellor, asked Russell to become vice chancellor for research, a role that largely overlapped the work Alpert had done at the Graduate College. Bitzer and CERL from then on reported to Russell.

  “It is certainly true,” says Desmond of Alpert, “that he had angered many people, but they gave him a very soft landing…the directorship of the Center for Advanced Study, which was a nice showplace, but anybody who occupied that position had very little power.”

  From Richard C. Anderson’s vantage point, Alpert and Bitzer had run the PLATO project “in an entrepreneurial way….The university had a tough time covering all the IOUs that were out there for the CERL staff, that Dan Alpert had granted rather indiscriminately in the flush days. Though Dan continued to be an honored member of the university community, they got him the hell out of administration….It was fine when things were flush, or if not fine, they say, ‘Well, this is the cost of being the great pioneers in this, we got to do some things that are outside the envelope.’ But when we came on harder times, that was a problem for the university.”