Chapter 2 - Background
2.23 Timesharing -- Project MAC -- 1962-1968
Timesharing as an idea first surfaced in the late 1950’s. Frustrated with batch-processing, scientists and computer programmers sought ways to interact directly with the computer. In 1959, Christopher Strachey, a British mathematician, gave the first public paper on time-sharing at a UNESCO congress; and, working independently, Professor John McCarthy distributed an internal memo about time-sharing at MIT. Under the leadership of F.J. Corbato, a professor at MIT, timesharing was first demonstrated at the MIT Computational Center in November 1961.133 There timesharing might have lingered were it not for the visionary leadership of Dr. J.C.R. Licklider and his license to invest government funds. In October 1962, Dr. Licklider became the first director of the newly created Information Processing Techniques Office (IPTO) of the Advanced Research Projects Agency (ARPA). His charge – to invest in advanced information technologies. Based on his experiences at MIT Lincoln Labs and Bolt, Beranek & Newman (BBN), and his vision of man-machine interactions,134 too briefly summarized as interactivity, he prioritized funding to timesharing projects. And if projects didn’t exist, he created them. For example, at MIT he helped create Project MAC, for machine-aided cognition or multiple-access computer, under the leadership of Professor Robert M. Fano. He then provided $3 million a year in funding – it would become the most influential effort in timesharing. In 1967, IPTO funding to over a dozen timesharing projects, at both universities and research organizations, exceeded an estimated $12 million.135
Timesharing required creating new software and hardware from that used in batch-processing. The most challenging innovation was designing and perfecting an operating system that could support many simultaneous users – all believing they had exclusive control of the computer.136 What made this sleight of hand possible was the speed of computers compared to that of humans. By the computer switching back and forth between programs fast enough, the illusion of both real-time and on-line was created, something operating systems designed to process programs in batch fashion would never be able to do.
The first computer company to emphasize timesharing was General Electric (GE). In May 1964, a GE computer was used in a timesharing demonstration at Dartmouth College.137 That summer, GE announced its 600 series computers would all support timesharing; using software developed at Dartmouth. And that fall, MIT surprised everyone when it announced it would buy a GE computer for its timesharing operations. IBM, which had abandoned internal efforts to develop a timesharing system and did not support timesharing in its initial releases of System/360, scrambled.138
Two years later, indicative of the challenging to create a timesharing system, IBM introduced its first timesharing computer, the System 360/67. Only the 360/67 lacked compatibility with the other 360 models, a key selling point of buying into the 360 series.139 By then, even IBM admitted to the importance of timesharing, believing it would represent 30% of the computer market. The competition waxed even more ebulliently. GE projected that by 1970, 75% of all computers would support timesharing. Business Week projected a $2.5 billion timesharing market in five years;– up from only $20 million in 1968.140
Timesharing contributed to the popularity of a vision of computing known as the Computer Utility. “Within the next decade electronic data centers are expected to sell computational power to the general public in a way somewhat analogous to today’s distribution of electricity,”141 . wrote Dr. ManleyR. Irwin in 1966. Employed at the Federal Communication Commission (FCC), Dr. Irwin should have known – being resident expert in computers and participant in the on-going Computer Inquiry. (See next Chapter for details.) The Computer Utility seemed a perfect metaphor in a time when computers were expensive, difficult to program, and required trained personnel who were in very short supply. If computers were going to be used by most, if not all, companies, then sharing made sense, especially given the success of the utility concept in both electricity and telephony.142 143
The explosion in growth of computer service bureaus seemed to validate the Computer Utility concept. Service bureaus, as either independent organizations or operations of computer manufacturers, sold computer time and services to other companies.144 In 1966, an estimated 800 service bureaus generated $650 million in revenues – growing at 40% per year.145 In a 1969 Datamation, a leading computer publication, Ralph Zani and William Zani write: “by bringing together the combined advantages of large-scale computing and communications equipment, the services [service bureaus] will surpass in capability what can be achieved with most in-house computer systems standing alone.”146
Service bureaus were not new. The most famous was IBM’s Service Bureau Corporation (SBC), which by terms of IBM’s 1956 consent decree with the Justice Department had to be managed independently of IBM. During this first phase of development, First Generation computer manufacturers used captive service bureaus as sales aids; part of their marketing strategies; a way to demonstrate products;147 a means to pre-install customer software, even to make it “pay-as-you-go;” and/or as a solution to overflow work from customers.
In the early 1960’s, independent service bureaus emerged selling “generalized” systems – a single program usable by many companies.148 But the service bureaus could not provide real-time access to clients until the development of timesharing. Timesharing resulted in a boom for service bureaus and growing demand for modems and multiplexers,149 and the birth of the Data Communications market-structure.150
IBM had no intention of being a by-stander in the service bureau business. While participating indirectly through SBC, still a part of IBM even though it could not publicize the fact, that was not enough. By 1966, IBM also ran a nationwide system of time-sharing computer centers. Dr. Irwin writes: “These new developments in technology and services raise the question, once again, of the status of IBM’s consent decree. Does time-sharing merely permit IBM to sell computer time over telephone lines, or is IBM processing customer data for a fee? What is legitimate activity for IBM as a manufacturer and IBM as a service bureau? The answers to these questions are not clear, but as if to hedge its short term anti-trust bet, both the Service Bureau Corporation and IBM, the parent corporation, have recently introduced nationwide systems of time-shared computer centers. In the long run, however, IBM many find it necessary to convince the Justice Department that new technology has invalidated major premises of its 1956 judgment.”151
IBM, unfazed by hovering Government threat, proceeded to take a lock on the market, selling both the dominant design, the System/360, and becoming the dominant firm, in an oligopoly of one, holding a 70 to 80 per cent market share in computers – mainframe computers. A 1968 Business Week article assessed the growth of IBM: “Since it entered the computer business 15 years ago, IBM’s volume has increased 17 times (to $5.3 billion last year [1967]) and its net income has gone up 20 times (to $651.5-million). Last year, IBM zoomed past Texaco and U.S. Steel to become the nation’s eighth largest industrial company when it added $1.1-billion in revenues. That is like creating another Coca-Cola or another Celanese in just one year.
In Wall Street’s assessment, IBM is now the most valuable corporation around. Early this week, IBM’s common shares were worth $41.5-billion. The common shares of AT&T, with assets eight times larger, were worth $26.3-billion. The stock market appraises IBM stock as worth at least as much as the combined shares of 21 of the 30 companies that go to make up the Dow-Jones industrial average.”152
IBM benefited from the public stock markets insatiable appetite for technology stocks during 1967 and 1968, but so did hundreds, if not thousands, of other companies. The valuation public investors were willing to place on a technology company by buying its stock went through the roof. One relatively new start-up, University Computing Co. of Dallas, Texas – a firm that didn’t make it – had its stock price bid up over three years from $1.50 a share (adjusted for splits) to $155.153 Venture capital became more widely available. New start-ups, hardly moved into their first space, were candidates to go public. Technology was the rage – where technology meant computer-related.154 One group of computer start-ups to be funded were minicomputer companies. But to mainframe watchers, it was as if an unrelated event…as if occurring on Mars.
And then came January 1969. On the 7th, the banks raised the prime interest rate to a then record 7%.155 Then Congress increased the capital gains tax. For anyone who did not appreciate how over priced technology stocks had become, they soon would – for the luster of technology proved more the reflection of investors’ greed than companies’ prospects. As rudely awaken as were investors were the management of the technology companies who no longer had money being thrown at them, but, once again, had to worry about cash flow and making sales. Management of private companies who had raised less money than needed to reach profitability or positive cash flow, believing that the money would be there when needed, only it would be a lot less expensive – investors would place a much higher valuation on their companies which translated into less dilution and lower cost of capital – were shocked. Many, probably most, either had to find survival in being acquired or simply closed down. The good times of 1967 and 1968 turned into the midnight years of 1969 and 1970.156 (These issues will be more fully explored in future chapters.)
In January 1969, Watson Jr. announced record earnings of $871 million – a 34% increase – for IBM. Revenues had increased by $1.5 billion to $6.8 billion. The very next day, on their last day in public office, Attorney General Ramsey Clark and his antitrust chief, Edwin M. Zimmerman, filed a major antitrust suit against IBM. The Justice department was not first to act, for in December 1968, Control Data Corporation filed suit charging IBM with violation of the Sherman Antitrust Act. Then earlier in January, Data Processing Financial & General sued IBM for violation of “antitrust laws, the Consent Decree of 1956, and state unfair competition laws.”157
The Justice department’s suit had been developing for several years and focused on the sale and leasing of general-purpose digital computers. The complaint, in spelling out the extent of IBM’s dominance of the computer market, seemed compelling. Every year from 1961 to 1967, IBM represented between 69% and 80% of the sales of computers. In 1967, while computers and related products only accounted for 43% of IBM’s revenue, or $2.3 billion, IBM constituted 74% of the total computer market. And the nearest rival held but a 5% market share, or revenues of $165 million.158
According to Business Week of January 25: “Essentially, the Sherman Act complaint accuses IBM of obtaining a “monopoly” by practicing manufacturing and marketing policies that deny competitors adequate opportunity in the computer market. The Justice Depart., objecting to four specific kinds of activities, claims that IBM has:
- Inhibited the growth or entry of competition by maintaining a single price policy for equipment, programming, and “related support” (system engineering services and the practice of supplying extra personnel to some customers.)
- Precluded effective competition from other manufacturers for various accounts, by using the accumulated power of its “software and related support.
- Introduced low-profit lines of equipment and announced new models when it “knew” that it was unlikely to be able to deliver on time.
- Influenced future purchases by dominating the educational market through discriminatory price allowances to colleges and universities.”159
Once the Justice department’s lawsuit had been filed, Applied Data Research Inc, and Programmatics Inc., both software companies, filed as well.160 The focus of their complaints also targeted IBM’s practice of not separately pricing hardware and software, nor of charging for software programming services. Since customers were forced to pay for IBM’s software, whether used or not, it made it extremely difficult to convince customers to then pay for third-party software. Unless customers had the choice of not paying for IBM’s software, they argued the software market would be stunted in growth.
While denying any connection between their changes in policies and the antitrust suits, IBM on June 23 announced that as of January 1, 1970, they would unbundle hardware and software pricing, charge for programming services, service their equipment no matter how the customer acquired it,161 and charge for education courses.162 There still remained the question of IBM’s market dominance, but they had eliminated many of their objectionable practices. Clearly the burst of lawsuits had, if not changed IBM managements’ attitudes, forced to make them public. All competitors now had reason to rejoice as the rules of competition had now changed in their favor. Only, what would not become obvious for another few years, IBM had already achieved unassailable dominance of mainframe computers.
In 1969, seven companies contested IBM in mainframe computers: G.E., RCA, Burroughs, Control Data, Honeywell, National Cash Register, and Sperry Rand’s Univac division.163 Known as the “Seven Dwarfs,” none believed they could unseat IBM, nor was that necessary to their survival and success. What each had to do was carve out enough market share, generally thought of as ten percent, to become number two to IBM. And since none of them held more than a five percent market share, the challenge was to double their respective size, not by clobbering one another, but by taking share from IBM. But how? Two stories will be told – GE and RCA, two industrial giants with the resources and commitment to succeed.
By the middle of 1969, GE’s strategy had two components: dominate timesharing and build a new line of computers compatible with the System/360. Timesharing was the easy part. Management believed that they controlled 40% of the timesharing market – through both its timesharing services as well as leases, or infrequent sale, of mainframe computers – and, if timesharing would come to represent as much as three-fourth’s of the market,164 as they believed, all they had to do was maintain their current share and they would be successful.165 Two assumptions drove their confidence in timesharing: “by the late Seventies from one-third to one-half of all computer power in the U.S. will be used for corporate strategic planning”166 and “within a couple of decades it will be as rare to find an in-house computer as it is to find a company-owned electric generating plant.”167
The second leg of GE’s strategy was formulated after March 1969. Newly promoted management knew GE had to respond to the System/360 and developed a plan to do so – it was code named Project Shangri-La.168 “Project Shangri-La very likely produced the most comprehensive and boldest master plan for an assault on IBM ever prepared,” wrote W. David Gardner in a 1970 Datamation article. GE was the only one of the Seven Dwarfs with the resources and size to take on IBM. Estimates for implementing the master plan represented a significant investment, even for GE. The low figures ran $450 to $500 million over six or seven years, covering machine design and development, as well as marketing development, but not production costs.169 In September 1969, GE executives, emboldened with their new strategy, reaffirmed publicly their commitment to the computer business.
RCA also premised their strategy on taking business from IBM in quest for 10% market share and the assumed number two position in the industry. L. Edwin Donegan, Jr., an exIBM’er, who became vice president and general manager of RCA’s Computer Systems Division on January 1, 1970, only one year after joining the company, was quoted as saying: “We have two options if we want to get 10% of the market. We can get a little out of the hide of each of the other Dwarfs, or we can get it out of IBM’s hide. It’s easier for us to get after IBM.”170 Only RCA differed from GE in having essentially only one strategy – out-IBM IBM.
Once general manager, Donegan prioritized the need for a new line of computers within the year. On learning R & D would be of no help, the advanced computers they were developing would not be ready for at least two years, he decided to take their current Spectra series of computers, change the covers, add more memory, and introduce them as a new line.171 Scheduled introduction date – September 1970.
Then on May 20, in “possibly the most startling event ever to occur in the computer industry,”172 or so claimed Datamation in July, Honeywell announced it was acquiring all of GE’s computer operations except the U.S. and Canadian timesharing division.173 (GE management refused to sell their last toehold in computers even though offered as much as $200 million for an operation that had never been profitable.174 ) In exiting the mainframe computer business, GE had conceded IBM held too big a lead, too much technology and too many resources. The investment required was simply too large and too risky to chance. Market consolidation, a sure sign the dominant design and oligopolistic firms have been decided, had begun.
A month later, on June 30, 1970, IBM upped the ante of competing, announcing the System/370 line of computers. The System/370 did not introduce a new generation of computers as had the 360, leaving many disappointed.175 Instead it represented incremental improvement along the technology trajectory established by the 360. A 1970 article in Datamation entitled “IBM’s System/370 Surfaces, But Is That All There Is?” points out that”perhaps that should not have been surprising, for it is unlikely that any company – least likely IBM – will ever shake the foundations of its user audience as profoundly as IBM did with the introduction of the incompatible 360 series.”176 177 The article goes on to say RCA, Honeywell, and NCR had terminals and systems ready to respond to IBM’s announcement.178 What the article fails to see is that the scale and scope of response is insignificant given the advantages IBM had been accumulating ever since announcement of the 360 in 1964.
In September 1970, RCA announced their new Spectra family of computers179 – believing they had priced their line very attractively to IBM’s announced and to be announced products. Only RCA misfired. The machines they announced to take control of IBM’s market were obsoleted in less than a week when IBM announced the System/370-145 which had more memory and was lower priced than RCA had anticipated.180 The System/370-145 was the first commercial computer to ship with not only semiconductor memory, but with virtual memory, which greatly expanded the effective memory the computer had.181
Undaunted, in March 1971, at the annual meeting, Robert Sarnoff, RCA’s Chairman, told shareholders : “Our highest priorities today are the establishment of a profitable computer business and capture of the domestic industry’s No. 2 position. RCA has made a greater investment in this effort than in any prior venture in its history, and we are convinced that the returns will be substantial.”182 RCA’s computer revenues had grown from $14 million in 1960 to $257 million in 1970; although sales growth in 1970 was less than 10% with a reported loss of $10 million.
Abruptly, on November 19, 1971, RCA, without any counsel from those in its computer division, threw in the towel. Univac would buy its customer base of roughly 1,000 machines and 500 customers for $70 million and the potential of another $30-60 million over the next five years, depending on future revenues.183 It made little difference to the competitive dynamics of the market-structure for IBM was solidly in control.
In the early 1970’s, the mainframe computer market-structure was seen as in turmoil, but in fact had become an oligopoly of one – IBM. The two potential alternatives, timesharing or a better mainframe, proved false hope. Timesharing could never overcome the overhead of communication costs – once terminals were over one hundred miles from the computer, telephone line charges made timesharing unprofitable.184 And as for a better mainframe, by the time competitors realized they had to do one better than the System/360, roughly 1969, IBM had an eight year advantage and an insurmountable lead.
While corporate computing had come to mean an IBM mainframe computer and a MIS department, a computer no longer meant a mainframe. Due in part to the stock market frenzy of 1967-1968, and the purchase of a start-up computer company, Scientific Data Systems, by Xerox Corporation for $900 million, or nine times revenues, ninety-one new companies introduced minicomputers between 1968 and 1972. Their story is next.
- [133]:
Fortune Aug 67, p. 91
- [134]:
His paper written in 1960 was extremely influential. J. Licklider, “Man-computer symbiosis,” IRETrans. Human Factors in Electronics,HFE-1:4-11, March 1960
- [135]:
Fortune Aug 67, p. 91
- [136]:
Fortune Aug 67, p. 88
- [137]:
Fortune Aug 67, p. 91
- [138]:
Fortune Oct 1966, p. 206
- [139]:
“Two years ago, for example, IBM rushed out the Model 360-67 computer, a machine designed for the broadest time-sharing applications, when General Electric appeared to have jumped ahead in the field. IBM’s machine was a disappointment, and its time-sharing service bureau operation lagged GE’s effort. “ Business Week, June 15, 1968, p. 88
- [140]:
Manley R. Irwin, “The Computer Utility,” Datamation Nov 1966, p. 22
- [141]:
Manley R. Irwin, “The Computer Utility,” Datamation Nov 1966, p. 22
- [142]:
“Economies of scale are becoming more and more pronounced. As these facts become clearer to computer designers, operators, and users, the time-shared public utility will assume its role as an important factor in fulfilling the computing requirements of the future.” Lee L. Selwyn, “The Information Utility,” Industrial Management Review, Spring 1966, p.17
- [143]:
The search for information content, “data banks” or data bases, brought the computer market-structure in contact with publishing. “The rash of mergers between the electronics industry and the publishing industry may temper one’s optimism with respect to the future. IBM’s acquisition of Scientific Research Associates, Raytheon’s purchase of D.C. Heath, Xerox and Wesleyan Press, RCA and Random House, to mention a few, suggest that the scurry to establish data bases is gaining momentum.”Irwin, 1966 p. 27
- [144]:
John L. Roy, “The Changing Role of the Service Bureau,” Datamation Mar 1970, p. 52
- [145]:
Gilbert Burck, “The Computer Industry’s Great Expectations,” Fortune Aug 1968, p. 142
- [146]:
Ralph L. Zani and William M. Zani, “Towards the Computer Utility: Evolution or Revolution,” Datamation Oct 1969, p. 126
- [147]:
“The scarcity of good programmers alone provides a significant impetus towards the computer utility.” Datamation, Oct 69, p. 126
- [148]:
Ibid., p.127
- [149]:
In 1966, “the Bell System predicts that eventually half of the information transmitted over its network will be data. “ – Manley Irwin
- [150]:
Ibid., p. 127
- [151]:
Irwin, p. 24
- [152]:
“Where IBM looks for new growth,” Business Week June 15, 1968, p. 88
- [153]:
Gilbert Burck, “The Computer Industry’s Great Expectations,” Fortune Aug 1968, p. 93
- [154]:
Leasing companies were a principal beneficiary of the explosive demand for new “high technology” computer company equities in 1967; the price per share of the public stock of “the six largest computer leasing companies showed a staggering 412% gain” in 1967.”Where investors get a run for the money,” Business Week June 1, 1968, p. 103
- [155]:
Business Week Jan 69 p. 38
- [156]:
Aubrey Dahl, “1969: An Overview of the News,” Datamation, Jan 1970, p. 91-93
- [157]:
Data Processing Financial & General Data Corp., asked more than $1-billion in damages. “Data Processing claims that IBM ‘s business practices, particularly its sale and lease prices on new computers, injure computer equipment leasing companies. The complainant is a an eight-year-old operation that had 1968 revenues of $16.7-million. More than 99% of the equipment it leases was made by IBM.” Business Week, Jan 11, 1969, p. 38
- [158]:
Business Week Jan 25, 1969, p. 37
- [159]:
Business Week Jan 25, 1969, p. 37
- [160]:
Datamation, Jan 1970, p. 89
- [161]:
“Where investors get a run for the money,” Business Week June 1, 1968, p. 104
- [162]:
Ibid., p. 91
- [163]:
RCA, UNIVAC, and Buroughs introduced faster machines than IBM. The Chip, p. 126
- [164]:
“It is hardly a coincidence that General Electric, the company that controls 40 percent of the time-sharing market, has also the longest and strongest commitment to management science.” Fortune, Oct 1969, p. 171
- [165]:
Fortune Oct 1969, p. 171
- [166]:
“It is hardly a coincidence that General Electric, the company that controls 40 percent of the time-sharing market, has also the longest and strongest commitment to management science.” Fortune, Oct 1969, p. 171
- [167]:
Tom Alexander, “Computers Can’t Solve Everything,” Fortune Oct 1969, p. 171
- [168]:
Datamation Nov 15, 1970, p. 24
- [169]:
W. David Gardner, “Anatomy of a Merger,” Datamation Nov 15, 1970, p.25
- [170]:
W. David Gardner, “Curtain Act at RCA,” Datamation March 1972, p. 36
- [171]:
Ibid
- [172]:
Datamation July 1970, p.80
- [173]:
W. David Gardner, “Anatomy of a Merger,” Datamation Nov 15, 1970, p.22
- [174]:
Ibid., p. 31
- [175]:
“IBM’s System/370 Surfaces, But is That all There is?, “Datamation Aug 1, 1970, p. 58
- [176]:
Datamation Aug 1, 1970, p. 58
- [177]:
“The single item that most distinguishes the architecture of System/370 from its predecessor, System/360, is the availability of a dynamic-address-translation facility, which allows programming systems to efficiently implement a group of functions which are collectively known as virtual storage.” So say the 370 designers. Datamation May 1978, p. 113
- [178]:
“IBM’s System/370 Surfaces, But is That all There is?,” Datamation Aug 1, 1970, p. 58
- [179]:
FromDatamationSeptember 1, 1970: “If RCA has met its design objectives (its new line is due out soon), the industry may see a more serious competitive attack on IBM than any of the other dwarfs has been able to muster in recent years. An internal design report espoused the RCA line’s suitability for computer utility use, with features as potent as those of Multics .”Datamation, Sept 1, 1970, p. 17
- [180]:
Datamation March 1972, p. 37
- [181]:
“Industry soothsayers who expected word on IBM’s fourth generation in 1975 have now set a later date: 1977 or ‘78. Till then, they say, IBM should be able to hold its own in the market place with enhancements to the 360 line, like the 370’s.
- [182]:
Datamation March 1972, p. 36
- [183]:
Datamation, Jan 1972, p. 53
- [184]:
Irwin, p. 26