Project Genie: Berkeley’s piece of the computer revolution
Mid-1960s computers were enormous, room-filling machines that only major corporations, government agencies and universities could afford. Prohibitively expensive, they functioned like today’s programmable-memory calculators: no words, no images, no sounds or music, no networking, just number crunching. IBM manufactured most of these mainframes to run the accounting systems and print the paychecks of the world’s corporations.
By today’s standards, these machines weren’t capable of much, but few of us appreciate how difficult they were to operate. Not only did they lack graphical user interfaces but, more fundamentally, computers were far too valuable to allow just anyone to sit at and work on. Computing was done by “batch processing,” which is like corresponding with the machine remotely via postal mail. The programmer submitted a “job” of coded punch cards into a queue, the jobs were fed through the computer by a lab-coated technician and the programmer received the resulting printout the next day.
Even in research circles at the relatively freewheeling labs and universities, computer time was a limited resource, like telescope time for astronomers or beam time for high-energy physicists. This meant that politics and position could trump scientific merit when it came to allotting researchers computer access for their experiments.
But a growing underground of computer scientists envisioned a different type of computing experience, a creative human–machine dialogue in which the computer acted as a true extension of the brain. Innovations were under way at Berkeley, MIT, Dartmouth, Stanford and Manchester University that helped move computing into a golden age of progress and democratization that helped liberate its bottled-up magic, first by re-engineering mainframe computers to make them more accessible, then by inventing the personal computer, a tool that anyone could use. Among these was Project Genie, a research effort at UC Berkeley.
“Projects like Genie have both a technical piece and a people piece,” says David Patterson, Berkeley professor of electrical engineering and computer sciences, whose textbook Computer Organization and Design provides historical perspectives on milestones in the evolution of computing. “If you look over the years at the projects Berkeley is famous for, they share this component of a group of talented grad students and a set of faculty working together to build things—as opposed to making theoretical contributions—to create an exciting new prototype that has an effect on industry.”
The timesharing alternative
In 1963, under the visionary direction of J.C.R. Licklider, the U.S. Department of Defense Advanced Research Projects Administration (ARPA, now known as DARPA) began funding Project Genie at Berkeley’s Department of Electrical Engineering, under principal investigators David Evans and Harry Huskey. The project had a fairly open-ended objective of improving computer technology for the military.
“Back then, the DARPA people were just sprinkling money around to schools where they thought they’d get a return,” says Wayne Lichtenberger, a new visiting assistant professor at the time, to whom Evans passed direction of Project Genie just months after it began. Genie’s primary researcher, graduate student Mel Pirtle, was focusing on building a timesharing system, an interactive alternative to painfully slow batch mode that dramatically reduced the per-user cost of computing. Multiple users could connect to one computer simultaneously, and the operating system would alternate processes so quickly that users felt they had their own machines.
Project Genie continued to attract talent, like grad student Butler Lampson and sophomore Peter Deutsch, who complemented Pirtle’s hardware orientation with system design and programming skills. Later, grad students Ken Thompson and Chuck Thacker came on board. The team built its system on the SDS 930, at $73,000 a relatively low-cost minicomputer manufactured by Scientific Data Systems, whose computers were used by science and engineering labs like Jet Propulsion Laboratory and Brookhaven National Laboratory to guide satellites and run experiments, a far cry from printing paychecks and tracking accounts receivable.
The SDS 930 arrived at Berkeley in September 1964, and the team began by writing a modified version of its assembler and linker, the programs that facilitate all subsequent programming and development. The machine had no operating system of its own.
“We chose the machine partly because it did not carry the immense baggage of an established operating system,” says Lichtenberger. “We had to write the systems software from scratch, which allowed all of it to be done by people who had the same philosophy and general goal.” In the next three years they developed the bulk of the software suite, including the timesharing system, the line-oriented text editor QED, the now-standard fork operation to create new processes, command-line completion, state-restoring crash recovery and many other innovative features.
Genie goes commercial
“Project Genie was the earliest useful realization of timesharing on a minicomputer,” says Robert Spinrad, Xerox vice president for technology strategy at the time, who later became director of Xerox Palo Alto Research Center, better known as Xerox PARC. “Their computer differed from earlier systems in that those were built on large, mainframe computers. This system was attractive to SDS because they made minicomputers and had been thinking of getting into the timesharing business.”
SDS manufactured the SDS 940—basically the SDS 930 with the Genie modifications made more bulletproof and at a much higher price—which became the most successful computer in SDS history, earning $40 million in sales and a devoted following among scientists and researchers worldwide. It ushered in the new business of commercial timesharing and was the initial hardware base for two major timesharing service companies.
“One of the great triumphs of Project Genie is that they used a smallish machine that people could afford,” says Jim Mitchell, now a VP at Sun Microsystems who worked at Xerox PARC in its early years. “There were a number of 940s around, so that created openings for other people to do other projects on them, like the Engelbart NLS, which he developed on a 940.” The NLS was the On-Line System, developed in 1968 by Douglas Engelbart (M.S.’53, Ph.D.’55 EECS), which featured the first graphical mouse- and windows-based interface.
The Project Genie team, brimming with ideas for even greater innovation, decided to leave Berkeley in 1968 to start their own company, Berkeley Computer Corporation (BCC). With Pirtle as president and $4 million in venture funding, some of it from the UC Regents, BCC assembled the best possible technical team, drawing mostly from Berkeley, MIT, Jet Propulsion Lab and SDS. Within two years, the company had created a prototype machine featuring numerous compilers and programming languages, an SDS 940 emulator and even remote-entry batch processing. But it was 1970, the economy was in a recession and, worse yet, BCC was light on sales and marketing talent.
“Mel Pirtle was a genius when it came to computers, but he could not really run a business,” says Lotfi Zadeh, professor emeritus of electrical engineering and computer sciences and department chair from 1963 to 1968. “Tragically, BCC went bankrupt.”
Birth of the personal computer
As BCC was failing, other changes were brewing. Xerox Corporation decided to enter the business computing market, purchased SDS, renaming it Xerox Data Systems, and founded its Palo Alto Research Center (PARC). ARPA’s Bob Taylor, who had overseen Project Genie for the federal agency, was hired to head PARC’s Computer Science Laboratory; he hired 20 employees, 11 of them from BCC, including Lampson, Thacker and Deutsch. Some of PARC’s other researchers who had been fans of the SDS 940 now found themselves working alongside its creators.
“When BCC folded, Xerox gave them a home at PARC, a brilliant move on Xerox’s part and an excellent move for the Berkeley guys,” Spinrad says.
Lampson, Deutsch, Thacker and company had many achievements at Xerox PARC, including the Multiple Access Xerox Computer, according to Lampson “one of the first systems with semiconductor memory.” But the Berkeley guys soon realized they could build a completely different type of machine, a single-user computer with a three-button mouse, keyboard, graphical user interface mode of operation and page-sized display. This was the Alto, a landmark in technological history that many consider the first personal computer, even though it was never sold to the public.
“Xerox was unsuccessful in making a commercial version of the Alto,” Berkeley’s David Patterson says. “But when Steve Jobs saw the Alto, he was inspired to create the Macintosh. Then, when Jobs showed the windows and mouse interface of the Mac to Gates, that inspired Gates to create the Microsoft Windows operating system. There’s a direct line from this extraordinarily talented group of Genie/BCC people to the personal computer.”
Lampson and Thacker championed the new Alto for PARC internal use in 1972. Timesharing had been the first step in democratizing computing, and this was the next: personal computers communicating in entirely new ways, through e-mail and word processing, with graphical user interfaces, networked together via Ethernet and served by laser printers, all of which were developed at PARC during the next decade. The old Project Genie team had uncorked another magic bottle. As a result, interactive technology is now everywhere, and your wish is its command.
Paul Spinrad, a San Francisco–based writer and editor, remembers visiting SDS and Xerox PARC with his dad, Robert, who worked at both.