Reprinted from Personal Computing 2/83, pp. 88-94.

The LISA project design team, in the photo above from left to right: John Couch, Bruce Daniels, Wayne Rosing, and Larry Tessler [sic].

The LISA project – even the mention of its existence sent rumors throughout the computing industry. But where and how did it all begin? Who were the men behind the machine?

Silicon Valley, the geographical cluster of electronics firms 40 miles south of San Francisco, is a fairly small community and, as with most small towns, gossip and rumor are important elements of society. There is never-ending curiosity about precisely what one’s neighbor is doing – what products are under development and what new technologies are on their way to market. But probably the longest-running bit of drama in recent Silicon Valley history has revolved around “LISA,” a secret housed in the scatter of modern, earth-toned buildings that comprise Apple Computer’s headquarters in Cupertino.

LISA, the ultra-friendly, long-awaited addition to the Apple product line (for more details, see Hardware of the Month, page 181), attracted far more attention than did the development of the average new computer. For more than two years, stories about the project have circulated – focused around the notion that this would be the company’s breakthrough product for the 1980s, intended to do for office systems what the original Apple did for the entire field of personal computing.

On that count, only time will tell. But the LISA project did produce a remarkable machine, and the collective nature of its invention is a process worth examining.

John Couch, the head of personal office systems at Apple. His mission as one of the guiding forces of LISA was to develop the prototype computer of the 1980’s.

John Couch, the rangy, energetic head of the personal office systems group at Apple, arrived from Hewlett-Packard in 1978 – at half his previous salary, but with an exciting assignment: to produce a strategy for the company’s future personal-computing software. “To spec out the personal computer of the 1980s,” he says. “To ask, ‘What’s going to be the next Apple II?’”

The first part of the answer was clear: The next great advance would take place in computers for the office, simply because that’s where the resources are to purchase sophisticated technology. Couch recalls sitting around in the office of Steven Jobs, Apple’s co-founder and now chairman, “with our feet up on the desk, thinking about what directions to go in. What we said was that we wanted to replace everything in his office – but we didn’t want to change the way people did things.”

The second part of the answer was Couch’s feeling that the next major step forward would lie in a software revolution, rather than a hardware evolution. Couch became vice president of software and, during a period when all of Apple’s development projects were given women’s names, launched LISA. (Most code names are dropped when projects reach the market, but LISA fit the human dimensions of the machine so well that the name stuck. At the last minute, a suitable acronym, Local Integrated Software Architecture, was reverse-engineered to fit the name.)

Couch recalls that one guiding principle for LISA came from his own father, whom he’d watched using a TRS-80 Microcomputer to run his health club. Every time Couch visited him, his father needed more programming to simplify his work. Couch realized that the machine could be made much easier to use if it integrated all the applications to begin with. At the time he was still working for Hewlett-Packard. “They thought it was a good idea,” he says, “and wanted to put it in a $100,000 machine.”

Thus LISA began with a fairly broad mandate: There was a general sense that the machine should be office-oriented, powerful, and easy to use. But no one was quite sure how to do it.

Bruce Daniels, a software designer, remembers seeing the “mouse” and becoming convinced that he would make it an essential element of the new Apple.

At first, ideas about the user interface – just how people would control the machine – involved then-current concepts like “soft” function keys (which have subsequently appeared in a number of other personal computers), says Bruce Daniels, one of the project’s earliest software designers back when the entire LISA staff – soon to swell to over 200 people – could still fit in a few offices behind a nearby health food restaurant. No one was satisfied, however, and the entire question of how to proceed remained up in the air until December 1979.

That month, a small group of Apple engineers, intrigued by some published papers, traveled a few miles north to the Xerox research center in Palo Alto to see a demonstration of a new language system called Smalltalk. Smalltalk was unusually friendly and made extensive use of a hand-controlled “mouse,” rather than a keyboard, to move the cursor on the screen. “We loved it – what they were trying to put across in ease of use,” says Daniels. “It was almost a snap at that point. We all said, That’s it – that’s what we want to build.’”

Larry Tessler [sic], a project engineer, left Xerox to join Apple six months after he gave a demonstration of Xerox’s Smalltalk language to the LISA team.

A young Xerox researcher named Larry Tessler [sic] gave the Smalltalk demonstration. “I was about the only person there interested in personal computers,” he recalls, “so they said, ‘You can talk to these people from Apple.’” As it turned out, Tessler liked the Apple folks. In fact, six months later he joined the LISA project engineers, who just then were in the throes of designing the user interface.

By then a lot had already happened with the project: The decision to go with the mouse approach had taken three months, and three months after that there was a prototype machine. “(There were) breadboards, wires, all over the floor,” says Daniels. “But it worked, and nobody got electrocuted.” The decision-making process surrounding the hardware – which processor, how many circuit boards – had been relatively brief by industry standards. But the next phase, the design of the user interface – the aspect of the project on which LISA would rise or fall – proved more trying.

A variety of mock-ups were built, and basic software decisions were made. The summer of 1980 saw such concepts as pull-down menus, scroll bars, and what came to be known as “cut-and-paste” procedures – all sophisticated uses of the cursor plus the mouse to replace keyboard input. Each change in the interface was subjected to great scrutiny, says Daniels, “even the question of how many buttons to have on the mouse.”

Make it simple

The mouse question reflects a basic approach of the LISA team. “In any situation when there were arguments in both directions, unless one approach was a big advantage for experienced users, we’d choose the one that made it easier for the beginner,” says Tessler. “In this case, we found that two buttons on the mouse were a slight advantage for experienced users, but not much. For beginners, when there were two buttons, they’d keep glancing down. With one button, they adapted right away. Since we had a very aggressive target – being able to learn the system in less than a half hour – we couldn’t have 20 minutes of getting over your fear of the mouse.”

“That’s pretty much how we spent the first year,” Daniels says, “coming up with ideas, trying them out, finding the problems, and trying something new.”

Designing a new user interface meant sampling a lot of users. Most tests were run on new Apple employees – they would be screened for previous computer experience, and a screened list would be delivered to the LISA group weekly. “Then if I was working on the spreadsheet,” Tessler says, “I could say, ‘I want two managers and three accounting clerks with no computer experience.’ Or else I could request a similar group, except that they’d all used VisiCalc before.” Hundreds of hours of observation ensued, supervised by psychologists, and much of it videotaped.

By the fall of 1980 the user interface was basically agreed on. Only then did the LISA group meet with Apple’s marketing people to decide which application would be developed for first release. Initially, five were chosen – such basic functions as word processing, graphics, financial modeling. Then programming started in earnest. Within three months nearly 20 new programmers were hired, with an average of 10 years experience each. “They came from all over the country, from places with all kinds of styles,” Tessler says. “There was a period of learning to live together, and then we got to work.”

The first rough versions of application packages were ready by the summer of 1981. “They were buggy,” Tessler says, “but at least you could sit down and play with them to get a sense of the user interface.”

Then came perhaps the most difficult task: integrating the application packages so that data could go back and forth, and text could be cut and pasted between operations in much the same fashion that John Couch had envisioned for his father. And all of this would have to run smoothly under one on-screen display called the Desktop Manager.

Something interesting happened in this process. The job proved sufficiently massive and involved so many interdependent tasks that the LISA team began drawing PERT (Performance Evaluation and Review Technique) charts to track the various aspects of the project. The weekly updating of the charts grew tiresome, however, and it wasn’t long before a program was written to do the job on the embryonic LISA.

“Suddenly everyone got excited,” Tessler remembers, “and wanted to make it another official application package. At first, marketing was against it, until they saw it, and then they wanted to use it, too.” Within a week the “LISA project” program was part of the product line, and it was only later that market studies were done to confirm that it would, indeed, be welcome in the world at large.

Not everything occurred so quickly for LISA. In fact, in terms of industry gossip, probably the most talked-about aspect of the project was its scheduling. “Since it’s been leaked from day one,” Daniels says, “the general perception has been that it’s horribly behind schedule.”

Wayne Rosing, director of engineering for LISA, describes the computer’s development as the “largest software project ever undertaken for a microprocessor.”

LISA was indeed long in arriving, and probably no one was more aware of that than Wayne Rosing, a soft-spoken, serious fellow who came from Digital Equipment Corp. to become director of engineering for LISA. “When I arrived in September 1980,” Rosing says, “LISA was supposed to ship in March 1981. In fact, when I was interviewing for the job, I wasn’t sure if I should say what I thought of that. I just indicated that I thought it was a little optimistic.”

LISA, of course, has now appeared almost two years after that original date. Some companies might not have been so tolerant. “But we’ve had great support from top management,” Rosing says. “You can’t just schedule an invention of this complexity. If we’d been forced to sign up to meet a date, we would have had to say, ‘No more invention, no more innovation.’”

But surely there was some pressure to produce a product?

“Sure, there were some very tough, brutal moments,” says Rosing, “but they were between me and top management. Their concern, of course, was, ‘Are we just iterating to make it more and more elegant, and will we ever stop?’ I’d try to explain the situation and then come back to the project and change as little as possible...”

In addition, Daniels points out, many projects are dictated by specific market pressures – having to get a computer out in order to compete with a similar product. “For us, most of the pressure was internal,” he says. “We wanted to get the machine out so people could use and appreciate it, so we’d push ourselves. We’d set ‘best-case’ schedules.”

“Best-case,” Tessler adds, “sometimes meant totally unrealistic.”

The most depressing period of the project occurred during the integration of all the applications under the Desktop Manager. Completion of a crucial mechanism for allowing the applications to share a programming code took far longer than was expected. “That was about a year ago,” Rosing recalls. “We needed a major miracle, and it wasn’t there – the project was slipping faster than the calendar.” Work continued on polishing the user interface, but, says Rosing, “There are times when you go to top management and say, ‘The shipping date has been moved forward again.’ And you never know when the boom will drop.”

The boom didn’t drop, and by the end of July 1982 all the application packages were running simultaneously under the Desktop Manager – the user could easily move from one to the other. It was a milestone, and the generally serious-minded LISA team celebrated with a champagne party. After sufficient champagne had been consumed, a handful of engineers decided to go back to the office and try the cut-and-paste technique that would allow material to be moved between applications. “It wasn’t supposed to work yet,” Tessler says. “We figured it would take another couple of weeks to get it debugged. But they got it working that night. And since people were still hanging around, we had another champagne party.”

After those two milestones, the nature of the software work changed. “Debugging and performance improvement,” says Daniels. “That’s the part of the project that’s not as (much) fun. Getting something up on the screen – that’s fun. But spending hours fixing bugs and making things run a little faster – that’s not so exciting.”

A worst-case analysis

But it was necessary, and it became even more so in light of another event in the life of the Apple company. In the midst of the LISA development project – in June 1980 – Apple introduced the Apple III in a version so thoroughly rife with problems that an avalanche of bad publicity ensued. In turn, extra pressure was placed on the LISA team to produce a solid product. By early 1981, for example, 50 usable LISAs had already been built for the software developers. At that point, however, Rosing insisted that the machines be put through a complete worst-case analysis – “timing, temperature, logic-loading, everything,” he says. “And then we basically started a redesign.” Several more versions were produced in “interim builds,” with the design improved each time for factors such as safety and manufacturability. It was not until February 1982 that a LISA close to its current state was produced.

Similar caution went into software testing. “The review process is set up,” Rosing says, “so that we know there are 1165 bugs in an application as of such and such a date – and that’s 1165 bugs that somebody’s watching, and we have to fix.”

In a sense, the ideas for LISA’s innovative hardware and software developed rapidly in 1980. “The inspiration came quickly,” says Tessler. “The perspiration took a lot longer.”

One exception was the exact format of the Desktop Manager – the program that provides the environment in which the user manipulates the cursor to choose among applications. There had been early problems with its design, and, to compound these problems, in June 1981 Xerox introduced its Star workstation at a computer show in Houston. The machine was also very user-friendly, and it was heavily influenced by Smalltalk. “A lot of people from Apple went there,” says Tessler, “and in a way it was like seeing Smalltalk the year before, in terms of inspiration.” The Xerox machine relied strongly on graphic symbols – icons, in the jargon – to organize its functions.

After mulling it over for nearly nine months, the Apple team decided to go with icons as well, which were integrated into the system only at the last minute. The icons are small drawings of symbolic objects – a file cabinet or a wastebasket, for example – arranged in a column at the right edge of the screen. Characteristically for the LISA project, even the shape of the icons went through various versions; different artists did renderings, and long discussions ensued as to whether the icons should be graphically elegant or simply as understandable as possible. (Also characteristically, the understandable version won out.) “In a way,” says Rosing, “the shape of the icons are like the whole product – you can’t point at something and say, ‘So-and-so did that; it’s all his.’ Every part of the system has a lot of people who contributed to it in various ways.”

The mouse, the icons – clearly, Xerox had a lot to do with LISA. And while the (two) products may ultimately be in the same marketplace, the Apple design team reports friendly relationships with the people at Xerox. “I think that they see the fact that we’re doing something like their product as giving them more credibility,” says Tessler, the former Xerox researcher. “They won’t be the only ones on the market with a bitmapped screen, a mouse, and icons, and so they’ve encouraged us.”

LISA also provided another point of interest within Apple. From the beginning, the LISA project was segregated from the rest of Apple. “The LISA group wasn’t designed to fit in,” says Daniels. “It was placed in another building, far away from the others. It was supposed to be revolutionary, so it wasn’t constrained by having to share older technology.” At first – particularly in 1980, when a whole set of new programmers were hired at the same time – the LISA group had a bit of a set-apart, new-kid-on-the-block, privileged image among the rest of the people at Apple. “There was skepticism,” Tessler says. “What are all these guys doing in that big building?” But this dissolved rapidly, he says, when the rest of Apple began to see the early fruits of the project. And by now LISA has thoroughly become part of the Apple culture. “After all,” Daniels says, “for three years, the sales of Apple IIs and IIIs supported us; now other people are looking forward to us contributing to their research and development.”

What vacation?

On a crisp California afternoon, Rosing, Tessler, and Daniels were sitting in a small conference room in one of the two large building now given over entirely to LISA and the personal office system division. They seemed relaxed, but they were clearly looking forward to less intense times. LISA, in retrospect, was a massive project representing more than 200 total man-years in development – compared to 25 for the Apple II and two for the Apple III. And much of that effort went into painstaking software work. “It’s probably the largest software project ever undertaken for a single microprocessor on behalf of one user,” says Rosing. In scope, it was closer to the kind of effort that produces minicomputers like the VAX – $100,000 machines that serve numerous users at a time.

Bruce Daniels mentions that he has now accumulated his maximum possible vacation time, and Larry Tessler asks, “What’s vacation?” But director Rosing is clearly sensitive about this. “Soul of a New Machine, the prize-winning book about the development of a minicomputer (Avon Books, 1982),” he says, “created a vision that that’s how engineering is done – crazy, intense efforts. But that’s partly because in that case, someone else had introduced a product and Data General was behind and had to get out a product to match. We’ve never had a policy about not taking vacations. People will be working over Thanksgiving, say – the door will be open – but it’s up to them. Sometimes during the project two guys would take a machine home and work day and night for a week, but you can’t keep that up.”

Looking back, it appears that the LISA project produced an unusual kind of group effort – one set up by Rosing to use only as much structure as was necessary. (Neither Tessler nor Daniels can recall what their official titles are supposed to be.) For example, Daniels says, specifications for LISA software started out to be very detailed – 50 or 100 pages. By the end, they were brief and informal; the last set, for a terminal emulator, was two pages. “In the beginning, we just knew it was going to be an exciting office product, with graphics, and easy to use. After a year,” Daniels says, “everyone agreed on what LISA represented – the next level of subtlety.”

“There was no guiding genius,” Rosing says, “only a guiding vision, a shared dream.” A visitor remarked that it sounds similar to a successful motion-picture crew – during the shooting, no one knows in every detail precisely how the film will turn out. But everyone, from costumer to cameraman, knows intuitively what’s right for the picture. “Exactly,” says Tessler. “In fact, on LISA, people had to have a sense of where we were going, because they were pretty much self-directed for a long time. Some people worked for nearly two years before they saw the whole package play together.”

Now, however, the package is put together. In the fall of 1982, sneak previews, mostly for potential corporate clients, had already begun in an elegantly furnished conference room at Apple headquarters, with three LISAs concealed behind an electric screen that moved aside dramatically. Early reaction had been very positive, and the LISA team seemed pleased – although much of their attention was still fastened on last-minute changes and plans for the immediate future. “And of course,” says Daniels, “there are those great ideas coming that nobody knows about now – when someone wakes up in the middle of the night with an inspiration.”

by Michael Rogers, Senior Editor