J. Impagliazzo, P. Lundin, and B. Wangler (Eds.): HiNC3, IFIP AICT 350, pp. 207–214, 2011.
© IFIP International Federation for Information Processing 2011
Making Business of a Revolutionary New Technology:
The Eckert-Mauchly Company, 1945–1951
Lars Heide
Copenhagen Business School, Porcelænshaven 18B
2000 Frederiksberg, Denmark
heide@cbs.dk
Abstract. The paper analyzes John Presper Eckert and John William Mauchly’s
endeavours to design, sell, and build the revolutionary new technology of the
first large, commercial computers. It discusses how Eckert and Mauchly’s
conceptualization of the computer grew out of their ENIAC and EDVAC
projects at University of Pennsylvania. They incorporated their own business to
gain profit from production and attain the freedom needed to develop their
revolutionary new computer technology through a series of small, separate
computer projects with private and government customers. It approaches
innovation as a chaotic process and uses uncertainty to conceptualize the basic
relations between actors and organizations.
Keywords: Eckert-Mauchly Company, EDVAC, ENIAC, John Presper Eckert,
John William Mauchly, UNIVAC, University of Pennsylvania.
1 Introduction
1
“… did you ever think it was going to turn out like this? … my
colleague Mr. Eckert and I, independently I think, have developed
about the same answer: that, yes, we felt it was going to turn out to be a
big thing. It was just to our disappointment that it took so long. But then
it always takes a long time to change people's minds, and it takes even
longer for us to change an institution.”
2
This was how John W. Mauchly, in 1973, recalled the development of the first
UNIVAC computer, completed in 1951. He acknowledged that he and John Presper
1
The paper is based upon extensive material in the archives of Hagley Museum and Library,
Wilmington, Delaware, USA (Sperry Corporation Records (acc. 1825), Sperry-Rand
Corporation Administrative Records (acc. 1910)) and at the Charles Babbage Institute at
University of Minnesota, Minneapolis, Minnesota, USA (Margaret R. Fox Papers (CBI 45),
C. E. Berkeley Papers (CBI 50), Mauchly talk 1973 (OH 44)) and extensive literature.
Research for this paper was accomplished based upon a grant from Hagley Museum and
Library, Wilmington, Delaware, and an Arthur L. Norberg Travel Grant from the Charles
Babbage Institute at University of Minnesota.
2
John Mauchly, talk to UNIVAC meeting in Rome in 1973, p.1, interview OH 44, Charles
Babbage Institute.
208 L.
Heide
Eckert did not know where the development would end, when it started in the mid-
1940s. In contrasts, the histories of the computer industry subscribed to a linear
narrative from the ENIAC project during World War II to the IBM System/360
mainframe computer closure in the mid-1960s, and to personal computers and
beyond. They discuss either engineering (hardware and software) or business aspects
of the development and they, therefore, miss the shaping interaction between markets
and engineering, and the crucial role of government in both funding projects and
acquiring computers.
An integrated analysis of business, technology, private market, and government
market is essential in order to understand how mainframe computers emerged.
Further, the linear narrative ignores the size of the endeavor that Eckert and Mauchly
faced in the mid-1940s of designing and building a revolutionary new technology, and
it ignores the complexity of subsequent development, until IBM established the
mainframe computer closure by introducing its System 360 in 1964. The development
of revolutionary new technology and establishing its production was never a simple
rational simple process. In the mid-1940s, designing and producing mainframe
computers required revolutionary new design and basic elements, like adders,
memory, storage, and input and output media.
The paper analyzes Eckert and Mauchly’s endeavours between 1945 and 1951 to
design, build, and sell the revolutionary new technology of the first large, commercial
computers in interaction with private and government customers. It approaches
innovation as being chaotic and uses uncertainty to conceptualize the basic relations
between actors and organizations.
2 ENIAC and EDVAC: Technical Feasibility and Design of an
Operational Computer
In late 1945, John William Mauchly and John Presper Eckert started exploring what a
computer should be and its commercial potential. This was based upon the ENIAC
project, which had proved the feasibility of building a large electronic computer, and
the EDVAC project, which had produced a feasible rapid-access memory, the stored
program design and opened for more complex use than calculating the third root of
2,589 raised to the sixteenth power.
In August 1942, John Mauchly wrote a memorandum on a large high-speed
vacuum-tube calculator for the Army Ordnance Department. The proposal outlined
the main features of a digital vacuum-tube calculator and focused on its technical
feasibility. This became the basis for a contract on a large electronic computer, which
the Army awarded to the Moore School. Eckert and Mauchly’s project team designed
new electronic circuits and built a reasonably reliable machine. By June 1944, the
formal design of the computer was completed and it was dedicated in February 1946
as the Electronic Numerical Integrator and Computer (ENIAC).
ENIAC proved the feasibility of building a large electronic computer, but it did not
have reliability, size, moderate energy consumption, and user friendliness needed in
order to make it attractive for a wider range of public and private organizations.
Based on discussions between Moore School and Army Ballistics Research
Laboratory, in October 1944, the Army granted a supplement to the ENlAC contract
Making Business of a Revolutionary New Technology
209
to develop a computer, EDVAC, which was easier to use for calculations. During
1945, this project ran in parallel with the completion of ENIAC, which had priority.
3 Shaping a Computer
In the summer of 1945, the building of large computers became technically feasible.
ENIAC was being completed and had proven the feasibility of large electronic
computers.
In December 1945 – while completing the ENIAC – Eckert and Mauchly started
envisioning commercial possibilities. They approached the Bureau of the Census in
order to sell the idea of a large-scale electronic digital computer as an efficient tool
for census work. For this end Mauchly and Eckert studied census data processing, and
probably magnetic tape was a significant sales argument. In the summer of 1945, the
EDVAC design applied magnetic tape or wire for input and output.
However, it was not simple to accommodate commercial production at a non-profit
university, such as University of Pennsylvania. In the spring of 1946, Eckert and
Mauchly resigned from Moore School because they wanted to exploit the commercial
opportunities of building computers.
The ENIAC and EDVAC projects had been classified because of World War II.
The end of hostilities facilitated declassification, which enabled Eckert and Mauchly
to use their expertise for business. In addition, war-funding disappeared and public
funding would again have to be based upon Congressional appropriations. The new
funding structure was probably a major reason that Eckert and Mauchly since
December 1945 worked to attain a contract to build a computer for the Bureau of the
Census. By the spring of 1946, the Census Bureau was definitely interested in
Mauchly's proposal.
At their resignation from University of Pennsylvania in March 1946, the possible
census contact was the only contract, which they found was at hand. They started to
work for establishing a company to develop a multi-purpose rapid computing
machine. Though they only were negotiating a contract with the Bureau of the
Census, they optimistically envisioned many business opportunities in scientific
computations, bookkeeping, and record management. They cited scientific
calculations at universities, government agencies, and industries, bookkeeping in
large companies, particularly insurance and railroad, and record management in
insurance companies and libraries. This was a revolutionary new way to see
computers compared to ENIAC and EDVAC.
However, the end of World War II caused a fundamental change in public funding
of projects. During the war, funds were amble for war-related projects like ENIAC
and the EDVAC pre-project and the trust needed for funding was accomplished quite
informally, when the project was located at a well-estimated institution like the Moore
School. Now, peacetime appropriation made it essential for a public organization to
establish the trust needed to award a project to an organization. Was the project
feasible? Was the price reasonable? Did the organization have the technological and
financial capabilities needed to complete the contract in due time? It was not simple
answering these questions for funding a project with Eckert and Mauchly of building
a computer for the Census Bureau. They had substantial technological expertise;
210 L.
Heide
however, did it suffice to build a revolutionary new device? Moreover, Eckert and
Mauchly’s business was not yet incorporated and had little assets. It was essential to
find a way to establish the needed trust, which required assessment by experts, public
or private, acting as intermediaries between the Census Bureau and Eckert and
Mauchly. The National Bureau of Standards rose to become the intermediary for
civilian and military government computer projects.
By June of 1946, the National Bureau of Standards had decided to award a study
contract to Eckert and Mauchly. The two inventors then formed a partnership, but the
study contract only became effective in October 1946. Originally, they anticipated
that the research and study phase would last six months. They expected to complete
the design phase within the next six months. This schedule proved to be overly
optimistic. The research and study phase itself lasted a full year. Moreover, it was not
until June 1948, that they signed the actual design contract.
The Census Bureau was prepared to spend $300,000 on the Eckert-Mauchly
computer, of which $55,400 went to the National Bureau of Standards for services as
the intermediary. In June 1946, Eckert and Mauchly's had estimated the development
cost at $400,000. Anyhow, despite their limited assets, the two men were willing to
absorb the anticipated loss, because they believed that if they were successful, additional
machines could be sold to both government and industry at substantial profits.
The Census Bureau needed a different computer than the sophisticated calculator,
which was the objective of building ENIAC and EDVAC. Census processing required
facilities for producing double entry tables and it took advantage of the plan to use
magnetic tape in the EDVAC project. It needed fast and reliable sorting, which Eckert
and Mauchly suggested could be achieved through transmission of data between two
magnet tapes operating at separate stations. This made construction of tape stations
and fast exchange of data between tape and the computer a key element of the Census
Bureau contract. The computer was changing from a fast calculator into a smart
punched card machine. This transformation was more complicated and time
consuming than Eckert and Mauchly anticipated.
The situation of late assignment of government contracts, which were smaller than
expected, and more work in designing and building the computer than anticipated
made Eckert and Mauchly open for approaches for additional contracts from A. C.
Nielsen Company, and Prudential Insurance Company.
The A. C. Nielsen Company was founded in 1923 in Chicago, Illinois, by Arthur
C. Nielsen, Sr., in order to give marketers reliable statistics on impact of marketing
and sales programs. In the mid-1940s, A. C. Nielsen Company used large punched
card installations to process the data for their market statistics. Nielsen was interested
in investing in new development, even in a small company, despite the obvious risk
and started negotiations with Eckert and Mauchly in December 1946.
In January 1947, Eckert and Mauchly offered to sell A. C. Nielsen Company a
computer system equipped with a key-to-tape recorder and a printer, all units to be
completed within a year. This implied changing the basis for Nielsen’s data
processing from punched cards to magnetic tape, which they accepted. But A. C.
Nielsen Company was reluctant to sign a purchase agreement with Eckert and
Mauchly’s company because of its precarious financial position. Nielsen wanted
Eckert and Mauchly to develop their equipment and gain more business in order to
establish the trust, which it needed for awarding a contract for a computer system.
Making Business of a Revolutionary New Technology
211
Over the next year, Eckert and Mauchly reshaped their computer system and
particularly its peripherals based upon their discussions with Prudential Insurance
Company. The A. C. Nielsen Company only needed a computer similar to that
already promised to the Bureau of the Census. In April 1948, Nielsen signed a
contract for a computer system with several peripherals. The system included one
computer, six tape units, six key-to-tape units, and one printer.
Simultaneously with the negotiations with A. C. Nielsen Company, Eckert and
Mauchly negotiated with Prudential Insurance Company of Newark, New Jersey. The
negotiations with Prudential made Eckert and Mauchly extend the scope of their
computer project to encompass alphanumeric data processing, because the Eckert
Mauchly Company realized that Prudential demanded equipment for premium billing,
mortality studies, and group insurance. Premium billing presupposed letters and
numbers.
Prudential’s data processing was based on IBM equipment since the 1920s.
Therefore, it viewed the Eckert and Mauchly’s use of magnetic tape for input and
output with concern. Although the concept seemed attractive, there yet was no
working model to prove its feasibility, and the use of tapes in place of cards would
mean that Prudential’s entire data processing operation would need conversion to
tape. Eckert and Mauchly convinced Prudential that their computer system was
feasible and a superior alternative to its current IBM punch-card systems, which
caused two concerns at Prudential. Eckert and Mauchly’s uncertain financial position
was a serious obstacle. Prudential was unwilling to sign a large contract with a small
company that had serious financial problems, particularly when sizable investment
was required to complete the contract. This was the case, because Eckert and Mauchly
had not yet accomplished designing the statistics computer for the contracts with the
Bureau of the Census and A. C. Nielsen Company, and Prudential’s alphanumeric
requirements would imply additional research and development as well as production
expenses.
While Prudential was not willing to contract for a machine, it signed an agreement
with Eckert and Mauchly, in August 1947, where it funded development of the
computer system in return for an option to buy one later. If the option was exercised,
the money provided for development would be applied to the purchase. The
agreement promised to complete the design and several prototypes of several key
elements of the computer system by the end of 1947. Once more Eckert and
Mauchly were overly optimistic in their time estimate. By the end of 1947, they had
no yet completed the prototype, and Prudential could have insisted that half its funds
be returned. Instead, it agreed to amend the contract several times to allow Eckert and
Mauchly’s company more time to fulfill its obligations. Prudential extended the
deadlines because it was impressed with the progress being made and the
development contract made it dependent on Eckert and Mauchly. In the end,
Prudential signed a contract in December 1948 for building a computer system. The
contract called for one card-to-tape converter and two tape-to-card converters, devices
that would allow Prudential to retain its punch-card data processing systems. This was
essential for Eckert and Mauchly extending their market to encompass non-numeric
data processing, for example, in insurance companies. The Prudential contract
described Eckert and Mauchly’s complete computer system, which they had named
UNIVAC a year earlier, Universal Automatic Computer. It had a central computer,
212 L.
Heide
tape drives, alphanumeric key-to-tape encoders, and line printers. In order to answer
demand, Eckert and Mauchly had developed their original ENIAC and EDVAC
number cruncher design into an alphanumeric design for data processing to succeed
extensive punched card business.
4 Establishing Business of Computers
Eckert and Mauchly left the Moore School of University of Pennsylvania in March
1946, because they saw bright possibilities of computer business and they founded a
partnership as the basis for their work. They focused on developing a computer and
attaining computer-building contracts through 1946 and 1947, and Mauchly travelled
extensively to win over customers. In this period, they established contacts with A. C.
Nielsen Company and Prudential Insurance Company, which contracted for
computers in 1948. In addition, the records show that Mauchly had contacts to about
twenty additional private and public organizations prospects in 1946–1947, which did
not produce contracts.
Eckert and Mauchly only incorporated their business in December 1947 and it was
named Eckert-Mauchly Computer Corporation. Though the company attained several
substantial contracts in 1948, its financial problems became more severe during that
year. Since the start of the year, Mauchly worked hard to attack additional customers
based upon a strategy to produce the company out of its crisis. The company was
desperate to securing adequate capital to assure the government and private
companies that they could complete contracts that they would make, and the company
failed to raise substantial new capital by issuing additional stocks in April 1948.
However, in the summer of 1948, new capital came from American Totalistor
Company of Baltimore. John Straus, its vice president, saw possibilities of computers
applied as totalisators. He convinced his company to invest $500,000 in in return for
40 percent of the voting common stock of Eckert–Mauchly Computer Corporation.
American Totalisator’s support kept Eckert and Mauchly’s company floating for
fourteen months, during which development on UNIVAC continued. In October
1949, Eckert and Mauchly’s company received an additional $100,000 from
American Totalisator. Anyway, the company’s problems were not yet solved. Nine
days later, Henry Straus was killed when his small airplane crashed. Straus had been
the prime force behind Totalisator's support and his death terminated the flow of
funds from this company.
As a direct result of Straus’s death, Eckert and Mauchly spent the remaining
months of 1949 seeking financing from loan companies and research foundations.
Finally, they sought to sell their corporation to a major manufacturer. They
approached producers of calculating equipment, such as Burroughs, IBM, National
Cash Register, and Remington Rand. Subsequently, they approached major industrial
producers, such as General Motors.
Remington Rand was first to act, and on 1 February 1950, it purchased all the
shares of the Eckert-Mauchly Computer Corporation. It became a subsidiary of
Remington Rand and functioned as a separate division. It delivered the first UNIVAC
computer to the Bureau of the Census in March 1931. By October 1954, Remington
Rand had delivered twelve UNIVACs and had orders for four more.
Making Business of a Revolutionary New Technology
213
5 Mess of Making Business of a Revolutionary New Technology
I started the paper by citing John W. Mauchly’s reminiscence, in 1973, of the shaping
of what became the UNIVAC computer. He admitted that the shaping process was
protracted, which he explained by observing, “…it always takes a long time to change
people’s minds, and it takes even longer for us to change an institution.”
He was correct that institutions – which he conceptualized as private and public
organizations – had to change, but this only occurred once computers were installed.
In the late 1940s, the organizational change of customer to demand computers had
short duration compared to the at least five years, which Eckert and Mauchly spent in
designing and building the first large computer. In contrast, Mauchly persuaded the
Census Bureau in five months to order a computer system, A. C. Nielsen Company
used between two and three months to reach a decision, and Prudential Insurance
Company reached their decision in less than a year. However, it took much more time
to establish the trust needed for these organizations to sign contracts. They were
uncertain of Eckert and Mauchly’s company’s financial capability, and they did not
want to be dependent on technology which perhaps never would materialize.
The demand for reduction of uncertainty and dependency gave room for the
National Bureau of Standards to establish itself as the national computer intermediary
facilitating civilian and military contracts.
Further, Mauchly underestimated the extent of developing a revolutionary new
technology twenty-five years after it took place, as he and Eckert did while they
explored the new technology between 1943 and 1951. Before starting computer
production, they had to establish company standards for designing and building
completely new technology, based upon many components with reliability problems.
Through this process, the scope of their computer’s planned applications grew from
the original number cruncher (ENIAC and EDVAC), to the numerical statistics
calculator for the Census Bureau and A. C. Nielsen Company, and to the
alphanumeric data processing machine for Prudential Insurance, which was named
UNIVAC. Each extension of applications added a new element of uncertainty to the
project. However, Eckert and Mauchly rejected an enquiry in 1946 of developing
their computer for totalisator applications, which illustrates that they did not pick all
requests. Their choices of customers remained within the scope they envisioned in the
spring of 1946 and gradually expanded their business opportunities.
At each expansion, Eckert and Mauchly accepted new uncertainty, because they
depend on one more customer to fill their company’s extensive need for funds to
complete their previous assignment. Their perpetual search for new customers, made
Mauchly commit extensive time to locate and persuade new customers. Often, Eckert,
who should have committed all his time to complete their technical project,
accompanied him. The records of Eckert and Mauchly’s company provide a hectic
picture of searches for funding that took time and delayed the project. Already in the
summer of 1947, they had also to borrow money from Prudential Insurance to be able
to keep their computer development project floating. In 1948, they received
substantial capital from American Totalisator. However, only access to the large
financial resources of the Remington Rand conglomerate, in 1950, facilitated
completion and production of UNIVAC computers.
214 L.
Heide
Eckert and Mauchly’s first large computer and their business were shaped through
dependency on a series of customers for expertise on future use of computers and
funds for innovating computers. At first glance, the technical development process
and the search for additional customers may seem chaotic, and, certainly, they held
many chaotic elements. However, Mauchly was correct in 1973 to claim that he and
Eckert all the way went for a general objective, which they accomplished though in
Remington Rand, a different business context than they originally anticipated.
Document Outline - Making Business of a Revolutionary New Technology: The Eckert-Mauchly Company, 1945–1951
- Introduction1
- ENIAC and EDVAC: Technical Feasibility and Design of an Operational Computer
- Shaping a Computer
- Establishing Business of Computers
- Mess of Making Business of a Revolutionary New Technology
Dostları ilə paylaş: |