by Brian Dear
Tenczar decided to try something different. His goal was to dramatically simplify what authors had to go through just to put text and graphics on the screen. He created simple commands like “at” and “write,” such that if you typed the following:
TUTOR example code Credit 20
and ran that code, PLATO would display “Hello, world!” ten spaces in from the left of the screen, and ten lines down from the top of the screen. No more CALL commands, no more subroutines. Prior to this point, displaying such a message on the screen might have taken fifty lines of code. Now it took two.
Just as Gilpin’s Summer Institute was about to start, Tenczar came in and declared to Gilpin and colleagues, I know how to do it! Know how to do what? they asked. I know how to get an arrow on the screen! I know how to do the language! I know how PLATO ought to run!
Tenczar called his new programming language TEACHER, but quickly changed that to TUTOR. Before the Summer Institute was even over, TUTOR was far enough along that Gilpin, intrigued, invited some participants to try it out instead of GENERAL, just to see what they’d create and how productive they could be. The results were eye-opening.
To be an author developing a PLATO III lesson in CATO or GENERAL in mid-1967, you had to be your own systems operator. While the project had evolved light-years beyond the ILLIAC days, PLATO III was still cumbersome. It made no sense for these visiting Summer Institute educators to learn how to operate the 1604, so Rick Blomme served as the operator for all of the participants. Meanwhile, Tenczar was working night and day on TUTOR. “Every day he would have a new command or two up,” says Gilpin, “and so we would take the stuff into the sessions, and the people would do stuff, and write their stuff, and in fact…several people who did that were able to come up with a fair volume of material. Even though they didn’t have any computer skills themselves at all.”
Rick Blomme had seen firsthand what could happen to productivity when you gave authors decent software development tools. He saw in Tenczar’s TUTOR another looming productivity boost. He and Tenczar began to collaborate and build out a fuller set of commands for the language.
Tebby Lyman was not happy with this development. TUTOR was to GENERAL what SOCRATES had been to PLATO: a threat to the officially sanctioned direction of the lab. But unfortunately for Lyman, that wasn’t how the lab worked. If there was an officially sanctioned direction at CERL, it was “Whatever works.” Lyman thought TUTOR “was a lot of nonsense,” says Gilpin. “This was an upstart from somebody who’s not one of us, and it’s deflecting effort away from further development of the GENERAL Logic, which is where we ought to be going.”
Tebby had been the principal programmer for Maryann Bitzer’s nursing lessons, all written using the GENERAL Logic. “When Paul Tenczar came on the scene,” says Jim Payne, “here’s this brash grad student…not a lot of people liked him, but he was a real savvy politician. There was a very hot meeting over whether both of these languages were going to coexist….He ruffled an awful lot of feathers in the process.”
Don Bitzer, a hardware guy through and through, stepped in, and, says John Gilpin, “declined to make a decision.” Which in its own right was a momentous decision, just the kind Bitzer was known to make. “He said,” says Gilpin, “ ‘Both kinds of efforts will continue until one of them works and one of them dies.’ ”
Around this time, PLATO III got another boost that would seal the fates of TUTOR and GENERAL: disk drives. With some fresh grant money CERL obtained an NCR 160 computer and two big Control Data disk drives. They needed the 160 computer because the 1604 didn’t know what disk drives were, so the CERL engineers hacked the 1604 and the NCR 160 in such ways as to make all of the hardware happy and functional. The result was a massive boost in productivity across the system.
TUTOR divided chunks of a lesson into what were called “units,” using the -unit- command, which was an approximation of a custom function statement in other programming languages. To form a more complete interaction with a student, for example, to ask the student to name the third president of the United States, but also take into account wrong answers, took only a few lines of code:
TUTOR code with an -arrow- Credit 21
In this simple example of TUTOR code, the student can type “Jefferson,” “T Jefferson,” or “Thomas Jefferson” and any of those will be judged correct. If the student types “Adams,” “J Adams,” or “John Adams,” or any other answer, the answer will be judged wrong. Over time, many additional TUTOR commands and enhancements would be added, making answer judging incredibly powerful, including requiring strict or lax spelling, case sensitivity options, stripping out ignore words, and long lists of matching vocabulary and concepts, not to mention powerful numerical expression analysis. In keeping with PLATO’s philosophy of being flexible and extensible to meet the needs of any instructor, the language would continue to evolve based largely on the feedback of the authors developing educational lessons.
In order for a particular class of students to take their lesson, the instructor had to make sure that the proper disk pack on the new disk drive was installed and running. Given the communications limitations of PLATO III, whereby only twenty terminals could be active on the network at any time, it was nothing new that the inactive sites were idle. But now, there was another reason to be idle, thanks to the arrival of the disk drive. “All of a sudden everything would shut down at CERL, because they had to put on [a different disk pack],” says Marilyn Beckman. “The nurses over at Parkland also had courseware that they were taking. They would flip some switch, there were terminals over there, and everyone was cut off at CERL. It was very funny.”
Despite the ongoing constraints, PLATO III was progressing at two steps forward for every one back. Soon, multiple authors could sit down at student PLATO III terminals and fire up MONSTER and edit their lessons, saved straight to disk instead of the elaborate ordeal of recording edits to paper tape, then magnetic tape. The combination of disk drives and a new authoring language made authors orders of magnitude more productive—and happy.
TUTOR worked. GENERAL died.
The battle between TUTOR and GENERAL left Tebby Lyman a lesser figure in PLATO. “She really faded away from the scene as a result of that,” says Jim Payne. “I was in some meetings where it was pretty uncomfortable with all the shouting….It just became, as much as anything, a personality battle here. Each one of them had a large amount of their life invested in, a large part of their ego invested in their respective languages….I do know that Paul ruffled an awful lot of feathers. Tebby had been there a long time, she was one of the original people who with Don started the whole PLATO concept, and she got knocked off her pedestal pretty big-time when Paul came in.”
—
Louis Volpp’s reign as director of CERL turned out to be brief. “It was my thought that Bitzer would do better if he had someone else administering and doing the business,” says Alpert. “As it turned out Don wanted very much to be the head. We had the kind of relationship where I could say, ‘Look, Bitz, I think this is the way it ought to be done,’ and my concerns had been that there are a lot of nitty-gritty things that I had always done for the project which I was finding it difficult to do—I had a full-time job. And I appointed Lou Volpp as director and Tebby Lyman as assistant director. It seemed like a reasonable thing. Well, Don practically refused to do any work.”
In April 1967 Max Beberman wrote a letter to Dan Alpert expressing concerns about CERL. “I agree with Don that we have been committing ourselves much too rapidly and that the organization of CERL is becoming too formal.” He suggested that Bitzer focus on the development of the upcoming PLATO IV system, Lyman handle production of materials and operations, and Alpert appoint a part-time business manager.
“The only impression I ever had from Dan Alpert toward Don,” says Volpp, “was that he was unusual and that he was great and that he needed to be supported but that you can’t put him in a harness, that he needed to do it his own way…
.Dan seemed to me always to express great admiration and love for Don’s work and wanted him to do well. From the time I got involved with the program, up until that time, Don had been the centerpiece for that effort in the Coordinated Science Lab, and that didn’t change when the new lab was created. Don was really the boss of everything. I didn’t ever have any inclination to try to tell Don what to do….My role was to find out what Don wanted to do and make sure we had it supported properly….Get funding, and get people out of the way, and get people in that he could cooperate with and if somebody out of another school wanted to help, then get that dean’s nose back out of joint so that person could be released to come and help Don, and work with Don.”
But by early July 1967, Volpp was out. His heart was in business and economics—something Bitzer must have noticed. “While I could appreciate a lot of the things that went on in PLATO,” Volpp says, “I was not going to be an innovator in it, that was not my strength at all, and so I did what I could, and then it was time to go and do something different.” By the fall of 1967 Volpp had moved to Duke University as a professor in, and eventually dean of, its new business school.
8
Knocking on the Same Doors
PLATO was not the only computer-based education project started in the 1960s. Other commercial and academic efforts were under way even as PLATO was just getting started. Government funding from the National Defense Education Act legislation of 1958 was still finding its way to schools, universities, and businesses. From one end of the country to the other, and in spots all over the world, educators, researchers, and engineers were tinkering with computers as the next big wave in education after the realization that teaching machines and programmed instruction were promising but could only go so far. The computer, they believed, could take you the rest of the way.
At Stanford University, two professors, Patrick Suppes and Richard Atkinson, began years of productive research in what they preferred to call “computer-assisted instruction” (CAI). Suppes, who shared with Bitzer the ego, the confidence, the ambition, and, it must be said, the arrogance, embarked with Atkinson on a project to study the psychology of learning around 1960. In 1962 they submitted a proposal to the Carnegie Corporation, which was looking for interesting projects. Says Suppes, “As we wrote it, we realized we could give it an educational learning theory ‘twist.’ We came up with the idea of computer-based instruction.” (Two years had passed since the PLATO project had started at Illinois, but Stanford was apparently unaware of it.) Carnegie awarded the Stanford team $1 million—a huge sum, vastly more than PLATO’s budget at the time. In 1963 Suppes and Atkinson launched experiments with elementary school children using teletypes connected to a small computer teaching elementary math and reading.
Suppes and Atkinson eventually teamed up with a group within IBM eager to tap into the education market. Together they developed a simple authoring language called Coursewriter, and eventually commercialized the package as the IBM 1500 system, with a mainframe, simple graphics terminals, random-access audio, and light pens for interacting with the screens. Whereas CSL and then CERL were determined to tackle the display memory problem (including addressing the prohibitive cost of video RAM) with the high-resolution plasma panel, for the 1500 system, IBM punted on the RAM problem by dedicating a disk drive to serve the same function as PLATO III’s expensive storage tubes. Up to thirty-two IBM terminals would each have their “screens” stored as files on the disk drive, available for constant display by the terminals. The displays were fairly low-resolution, at 320 x 192 pixels, which only required about 7 kilobytes of disk space per terminal. IBM never made any money from the 1500 system, and only sold a few dozen, but they were nevertheless used in a wide variety of school settings from coast to coast.
In September 1966, Suppes, perhaps even more adept at promotion than Bitzer, wrote an oft-cited article in Scientific American entitled “The Uses of Computers in Education,” which probably did more to introduce the concept of CAI to laypeople than any other article up until that time. No stranger to audacious goals, Suppes opened his article stating, “One can predict that in a few more years millions of school-children will have access to what Philip of Macedon’s son Alexander enjoyed as a royal prerogative: the personal services of a tutor as well-informed and responsive as Aristotle.” The article focuses on the work being done at Stanford with IBM, and briefly makes reference to the existence of other centers of CAI work around the country, including Illinois, but never mentions PLATO. Suppes and Atkinson also managed to get major articles in Time and Life magazines around the same time.
Suppes’s emphasis was drill and practice. “In the early days,” says Andrew Molnar of the National Science Foundation (NSF), “CAI was not all that popular. You could add two and two but you had to wait about five to ten seconds to get the answer before you could go on to the next frame. Many of the kids became bored. Drill and practice was usually limited to ten minutes, because that was the limit of the student’s span of attention when doing drill and practice. While it improved performance, it could be boring.” Suppes and Atkinson’s computer lessons applied the age-old principles of Self-Pacing and Immediate Feedback, but now, unlike with Skinner’s teaching machines, the computer was able to generate custom questions based on a real-time evaluation of how well a given student was doing as she went through the problem set. “Suppes was able to go into very poorly run educational environments and produce results immediately,” says Molnar. “The state of California carried out a well-controlled study…the arithmetic did work; it improved performance significantly, but the reading was not as successful.”
In 1967, the ever-ambitious Suppes founded a start-up company, Computer Curriculum Corporation, which lost money until the 1970s, when it finally became profitable, and went on to be acquired in the 1990s.
Thus, Stanford, through the work of Suppes and Atkinson, established a firm reputation as the research hub on the West Coast for CAI. In addition, they pumped out a number of graduate students who went on to play important roles in various organizations including Xerox’s Palo Alto Research Center (PARC). “These guys, Atkinson and Suppes,” says Dexter Fletcher, an educational psychology doctoral student of Atkinson’s, “were real money machines. That’s what it takes to become a professor at Stanford. You attract your research funding. And they were just phenomenally successful at that.”
An interesting contrast can be made between the CERL lab and Suppes’s IMSSS (Institute for Mathematical Studies in the Social Sciences). CERL was the exception to the rule; CERL’s openness was legend, thanks largely to the management style of Bitzer. IMSSS exemplified the more normal university lab. Fletcher described the IMSSS culture as follows: “It’s an institute and a laboratory, and you can’t just sort of walk in the door, ‘Hi, here I am,’ to a bunch of busy people and have them turn, drop everything, and [help you land a job there]. So there’s some dues to be paid, and I think a lot of people weren’t willing to pay them.”
—
Buzzwords have always plagued the broad field of educational technology, long before computers arrived on the scene. “Distance learning” was a fancy way of describing “correspondence courses,” and then with the arrival of phonographs and radio and television and tape recordings and film, it meant something else. With Pressey and Skinner came “teaching machines,” and later in the 1950s there was “programmed instruction,” and then with the arrival of PLATO came “automatic teaching” and “computer-based education,” two favorites of Bitzer’s in the early days. It can be taken as a rule that if you are a practitioner in this field, and happen to be ambitious, competitive, and want to establish a name for yourself, you first have to talk about what you’re doing in a way that is slightly different from everyone else who’s doing roughly the same thing. How could Suppes and Atkinson use “computer-based education” as a buzzword if that is what Bitzer and Alpert were using? Instead, the Stanford buzzword was “computer-assisted instruction.” In Suppes’s perf
ect world, mere mention of CAI would be equated with Stanford’s project. Suppes pushed hard to make real that perfect world.
“Nobody liked the term ‘computer-assisted instruction,’ ” says NSF’s Andrew Molnar. “Everybody felt it was a misnomer. But its use in newspapers was so widespread and so many people were familiar with the term ‘CAI’ it became apparent that there was no way of converting them to anything else. Computer-based instruction, computer-managed instruction were used, but you could not change the popular usage of CAI….Probably no other name is more closely associated with CAI than Suppes. For years they were synonymous.”
Bitzer stuck with CBE—computer-based education—as the main buzzword associated with PLATO. He once described its meaning at a conference: “Our definition of ‘computer-based education’ is that if a human and a computer get together, if either one of them happens to learn something, we consider that our field. So it’s very wide indeed.” Not surprisingly, no other major project used the “computer-based education” buzzword—CBE meant PLATO.
In the 1970s, a professional society, ADCIS, an acronym for Association for the Development of Computer-based Instructional Systems, pushed the somehow gentler CBI acronym, for computer-based instruction. Because ADCIS was so heavily weighted down with PLATO-affiliated academics and industry people, CBI generally hinted at PLATO’s CBE.
