Chapter 2 - Background
2.26 Personal Distributed Computing -- Xerox PARC -- 1980
In 1979, few saw how microprocessors would penetrate corporate information systems. An exception were the computer scientists at Xerox Corporation’s Palo Alto Research Center (PARC) who had developed the Alto computer system. For Xerox, Alto proved to be of disappointingly little benefit – unlike the SAGE Project to IBM – but to the future world of computing it came to define distributed computing. In less than a decade, this new vision – personal distributed computing – made real in the Alto, propelled the innovation and economics of personal computers and their use. It was not a vision readily or willingly grasped by central MIS departments of large corporations, and by the time it was, control of how organizations informated252 no longer was theirs to divine.
Vision plays an essential role in how computer technologies become instantiated. The need for fast calculators drove the first vision of computers. Then the vision became one of computers able to perform real-time computations. Driving this vision were the needs of the SAGE Project, funded by the Air Force. To those innovating SAGE, real-time wetted their appetite to interact directly with the computer, to be on-line, not to batch-process, the existing mode of interaction.
But in the mid to late 1950’s, computers were big and expensive. Cost prohibited their use. How to make interactive computing more accessible, available, and much more affordable became the design objectives. The vision became one computer supporting many simultaneous users – or timesharing. It was reduced to practice at MIT as Project MAC, again funded by the government, this time as ARPA. Timesharing, while designed with a big computer in mind, proved uneconomical for mainframes due to communications costs. Nevertheless, timesharing found a welcome home in minicomputers.
Xerox PARC came next. The individuals and the culture they created gave rise to the radical idea of giving every person a high-performance computer, architected as easy-to-use, and connected in a seamless manner to every other computer. How preposterous – in 1972, computers and software cost in the hundreds of thousands of dollars, computer-to-computer communications barely existed, and computers were anything but easy-to-use. The vision of personal distributed computing changed the assumption from fast computers and slow users to precisely the opposite, and in so doing, unleashed a torrent of innovation to forever change the nature of computing.
In early 1969, Peter McColough, President of Xerox, convinced that Xerox needed to acquire an existing computer company, made an appointment to see Max Palevsky, founder and Chairman of Scientific Data Systems (SDS). McColough’s convictions were neither new nor surprising. As early as 1965, McColough and Xerox Chairman Joe Wilson had approached several computer companies, including SDS and Digital Equipment Corporation, to sell them on the advantages of becoming part of Xerox – but to no avail. Their motivation was rooted in the fear of coming competition from IBM in copiers and of the looming importance of digital technologies in the future, technologies then foreign to them. To be a contender in the just emerging and potentially gigantic market referred to as the “office of the future,” Xerox, they believed, had to sell computers. As McColough prepared to call on Palevsky, he had good reason to believe this might be Xerox’s best, and maybe last, chance to acquire a computer company – and his own chance to earn a reputation as a leader as bold as Wilson.253
McColough wasted no time. He told Palevsky that Xerox would pay $900 million in Xerox stock for SDS – nine times SDS’s 1968 revenues of $100 million. To Palevsky, who was seeking to scale back his commitment to SDS, McColough’s words must have seemedlike a dream come true. The deal was done in two weeks.254 Xerox had now entered the computer business to the surprise of top technical personnel within Xerox and to Wall Street analysts– who could hardly believe the purchase price. McColough immediately challenged SDS to aggressively target expansion into commercial markets,255 neither fearing IBM nor deterred by the mismatch of SDS’s existing products with the needs of the market nor with the struggle of IBM’s current competitors.
On hearing of the acquisition, Jack Goldman, chief technologist for Xerox, recommended Xerox create a research center for information technologies. With a history of successfully leveraging research innovations – namely xerography – McColough quickly agreed and directed Goldman to find the person to head the effort. By early 1970, Goldman recruited an old friend and physicist, George Pake, with the charge to build a research center equal to AT&T’s Bell Labs. SDS executives objected, arguing the money would be better used to build products they could sell, but they were rebuffed.
In a March 1970 speech before the New York Society of Security Analysts, McColough, now CEO, waxed loftily : “The basic purpose of Xerox Corporation is to find the best means to bring greater order and discipline to information. Thus our fundamental thrust, our common denominator, has evolved toward establishing leadership in what we call ‘the architecture of information.’“256
Opening a facility in Palo Alto, California, in June, Pake began the critical process of recruiting staff. To educate himself on computer science and to find the needed people, Pake arranged a tour of many of the best university programs. He quickly learned that the Advanced Research Projects Agency (ARPA) of the Defense Department had the reputation of funding the most advanced computer research and supporting many of the best researchers. Pake arranged a visit for Bob Taylor, who was now at the University of Utah but had directed the IPTO office of ARPA from 1965 to 1969.
Pake viewed Taylor as strictly a consultant. He did not qualify as a candidate to head the Computer Science Laboratory because he had never conducted research himself – and to Pake, if one had not done research, one could not manage a research organization. On meeting, Pake explained the mission of CSL: supporting the long-term needs of SDS. Taylor expressed critical reservations. Based on personal experience with SDS, Taylor knew SDS did not have a vision of computing emphasizing timesharing and interactivity. When asked what PARC should do instead, Taylor stressed innovation to support of Xerox’s transition from a “copier company to being in office automation systems.”257 Pake realized Taylor held a bold vision of computing – one more resonant with McColough’s architecture of information mission – and he knew who to recruit to bring it into being. Pake offered Taylor a job, and although not the one Taylor wanted, Taylor quickly carved out the role he sought – leading PARC’s computer research in creation of computing after timesharing.
Taylor then recruited some of the absolutely best computer scientists. Their names are legendary: Alan Kay, Butler Lampson, Chuck Thacker, and others. They became recognized as the best computer research laboratory anywhere. (Ironically, some of the researchers before joining Xerox were funded by ARPA to write a timesharing system for a SDS computer that became the SDS 940. Even so, no productive exchange ever developed between CSL and SDS.) With staffing well underway, a set of research objective(s) were needed.
Taylor argued – before relenting – that the model, or metaphor, of timesharing was passÃ©. He saw a future when everyone would have their own high-performance computer, rather than sharing one lumbering behemoth. While others agreed, exactly how to go about implementing such a vision remained very much in question. Knowing they would need tools, including a computing environment, they resolved to focus on creating the infrastructure that would be needed while refining their thoughts on the future system. A timesharing system would be written, and not on a SDS computer, but on a copy of a DEC computer which had a timesharing operating system underwritten by ARPA. By mid 1972, MAXC and other tools were ready, and it was time to take on the larger challenge – one to become known as personal distributed computing – an idea J.C.R. Licklider originated now pursued by his friend and colleague Taylor. Taylor had now surrounded himself with others who believed as he did, especially Kay and his reactive engine – in honor of Babbage and his analytical engine – and best known as Dynabook.258
To premise an architecture of office information systems on every person having their own computer was a radical idea. First of all, what would anyone do with their own computer? At the time, computer output was a string of text or numbers, an answer if you will. But the PARC scientists believed the expression coined by Engelbart at Stanford Research Institute in the mid-1960’s, augmenting man’s intellect,259 better captured the power and usefulness of computers than did the prevailing notion of essentially a very fast calculator. To augment meant to present and accept data in the form best suited to humans, not computers. Humans can process graphical information unbelievably fast, yet computers did not support graphical interfaces. The same is true of communications – people can communicate millions, even billions, of bits per second visually, yet at the time computer communications delivered maximally ten thousand bits per second. If the computer was designed to support people, the computational requirements skyrocketed, which begged the second problem. How could everyone afford their own computer?
The costs of a single computer in 1972, a minicomputer not mainframe, could total in the hundreds of thousands of dollars – clearly not a viable strategy on which to build an economic future. However, the scientists at PARC believed, with good reason, that the costs would soon drop to affordable levels. Lampson in a guest editorial in a 1972 issue of Software - Practice and Experience predicted that within five years, it would be possible to build a computer: “comparable to a 360/65 in computing power for a manufacturing cost of perhaps $500.”260 So the problem of costs seemed more short term, yet real enough to preclude action unless a solution could be found. In response, Thacker conceived multi-tasking261 – the computer would work on many tasks all at the same time. There were various forms of multi-tasking to support timesharing.262
The Alto computer, or workstation, would be powerful enough to satisfy the computational needs of a single user. It would have a high-resolution bit-mapped display, with the resolution of a full page of multi-font text – what-you-see-is-what-you-get (WYSIWYG) quality – and a highly interactive user interface with cursor control provided by a mouse – invented by Doug Engelbart and his group at SRI. To constitute an office information system, a large number of Altos would be inter-connected with a local area network, Ethernet. Specialized computers for services that could be shared either logically or for reasons of cost would also reside on the network. (Little will now be said of Ethernet, or local area networks, for they represent the second paradigm of computer communications and will be reconstructed in detail in future chapters. Hence, PARC will be visited more fully later.)
The Alto display system was first demonstrated at PARC in November 1972. By April 1973, an Alto worked – a year later there were forty of them. During this period and subsequently, it would take on its finished look when enhanced by Ethernet, the first laser printer and application software, in particular word processing (Bravo and Gypsy) and electronic mail (Grapevine).
Meanwhile, SDS struggled. First there was the mismatch of SDS’s products, both hardware and software, and the needs of the market. Then SDS was slammed by the slow down in government spending and the computer recession of 1970-71. In response, in 1972, Xerox reorganized SDS, eliminating it as a separate business unit with its own general manager, and merged it into the copier operations. It continued to lose money, forcing yet another change in strategy in 1973 – no longer would SDS-Xerox compete head-on-head with IBM. Like GE and RCA, SDS-Xerox found it could not match the first-mover advantages and market power of IBM.
The June 1975 issue of Business Week described IBM and Xerox as two of the big winners in office automation. It mattered little. By the Xerox’s Board of Director’s meeting the following month, the painful decision could no longer be avoided. No hope existed either of turning their computer business around or of selling it to anyone else. The only real option was to shut SDS down, take the loss – an estimated $1.3 billion263 – and get on with life. McColough rationalized the action at the time by saying that Xerox did not have to sell stand-alone computers, it only had to be able to handle “information in digital form.”264 Later he admitted: “In retrospect, the Scientific Data acquisition was a mistake.”265
Some of the SDS employees were assigned the task of building hundreds of Alto II – a refined design – shipped primarily within Xerox. Also in 1975, a Systems Development Division (SDD) was created to productize PARC technologies. In 1976, John Ellenby, the leader of the Alto II project, proposed building an Alto III, but an alternative design, one advanced by the Office Systems Division (OSD) in Dallas, won out. Those at PARC and SDD felt shunted aside. In 1977, McColough asked Ellenby to prepare a massive demonstration of PARC technology for a “Futures Day” World Conference for Xerox executives and management. It came off without hitch and impressed all. Even so, nothing happened. SDD, under the leadership of David Liddle, persisted in trying to gain support for their product ideas within OSD, but without any success. In 1978, Xerox reorganized OSD, in essence closing it down and combining it with the copier operations. That didn’t work. In 1979, OSD was re-established under the leadership of Don Massaro, who joined Xerox by way of the acquisition of the disk drive company he helped found, Shugart Associates. All the while the Alto technology remained suspended in organization mid-air.
In November 1979, Ellenby and Liddle traveled to Dallas to present the product plans of SDD to Massaro. Massaro saw immediately the possibilities and encouraged Liddle. Subject to organizational approvals, OSD would market and sell the Star System SDD was developing. Star was intended to be a next generation Alto, both lower in cost and higher in performance. They got the go-ahead and in February 1980 began preparing for shipment in the spring of 1981. The Alto technology finally appeared on track to be commercialized. In April 1981, Xerox OSD began shipping the Star.
Only the Star proved a dead-end. It was too expensive, not functional enough - it didn’t have a spreadsheet – and totally proprietary. Xerox would not compete in the market for desktop computers either. Yet, mastering digital technologies to the extent they did, enabled Xerox to remain competitive in copiers and printing systems.
Any lead time Xerox had disappeared in the nine years that elapsed from the first working Alto to shipment of the Star. In August, IBM shipped the IBM Personal Computer. At first it hardly resembled the Star, but in time, thanks to the visionary developments at PARC, the way had been cleared and personal computers brought an end to corporate information systems centered around the mainframe computer.
Informated as in S. Zuboff’s sense.
See Douglas K. Smith and Robert C. Alexander, “Fumbling the Future: How Xerox Invented, Then Ignored, The First Personal Computer,” William Morrow and Company, Inc., 1988, for a complete discussion of these issues.
Ibid., p. 30
Ibid., p. 124
Douglas K. Smith and Robert C. Alexander, “Fumbling the Future: How Xerox Invented, Then Ignored, The First Personal Computer,” William Morrow and Company, Inc., 1988, pp.. 48-50
Alan Kay and Adele Goldberg, “Personal Dynamic Media,” Computer March 1977 and Alan C. Kay, “Microelectronics and the Personal Computer,” Scientific American Sept 1977, pp. 213-244
Douglas C. Engelbart, “The Augmented Knowledge Workshop,” ACM, Jan 1986
Charles P. Thacker, “Personal Distributed Computing: The Alto and Ethernet Hardware,” ACM Conf. 1986, p. 89
A very limited form of multi-tasking was used in both the TX-1/TX-2 designs of Wesley A. Clark and was calledmulti-sequence counters.Wesley A. Clark, “The LINC Was Early and Small,” ACM Conference, Jan 1986, p.135
It also built on ideas first used in the TX-2 computer, the one designed at MIT Lincoln Labs under the direction of Ken Olsen.
Smith, Ibid., p. 128
Gary Jacobson and John Hillkirk, “Xerox: American Samuri,” Macmillan Publishing Company, 1986, p.217