Fermat's Library | Electronics, Technology and Computer Science, 1940-1975: A Coevolution annotated/explained version.

Paul Ceruzzi is a historian specializing in the history of technolo...

Ceruzzi is referring to Charles Babbage's designs for his "Analytic...

Vacuum tubes are devices that control the flow of electric current ...

Relays are simple electromechanical switches that control electrica...

Howard Aiken was an American physicist and pioneer in computing who...

A mercury delay line was originally developed during World War II f...

The reason vacuum tubes provided such a dramatic increase in speed ...

The stored-program principle is the idea that a computer's instruct...

The ARC computer was an early British computing machine developed b...

Drum-type computers were early stored-program computers that used a...

In 1883, while working to improve the incandescent light bulb, Thom...

Gottlob Frege, David Hilbert, and Kurt Gödel were mathematicians wh...

IBM's launch of the System/360 in 1964 significantly reinforced the...

Annals H~.st Comput (1989) 10 257-275

G American Federation of Information Processing Socletles

Electronics Technology and

Computer Science,

1940-l 975: A Coevolution

PAUL CERUZZI

This paper explores the relationship between two disciplines: electrical

engineering and computer science, over the past 40 years. The aufhor

argues that it was the technology of electronics-the exploitation of the

properfies of free electrons--that final/y permitted BabbageS concepts of

automatic computing machines to be practically realized. Electrical

Engineering (EE) activities thus “took over” and dominated

the

work of

those involved with computing. Once fhaf had been done (around the mid-

195Os), the reverse takeover happened: the soence of computing then

“took over” the discipline of Electrical Engineering, in

the

sense that its

theory of digital switches and separation of hardware and

software

offered

EE a guide

designing and building ever more complex

arcuits.

Categories and Subject Descriptors: K. 2.

[Computing Milieux]:

History

of Computing-hardware, software, systems, theory. A. 1,

[General

Literature]:

Introductions and Survey.

General Terms: Des/gn, Reliability, Theory.

Additional Terms: Computer

Science.

Electrical Engineering.

Introduction

In 1976, a colorful brochure put out by the IBM

Corporation had a startling title: “It Was to Have

Been the Nuclear Age. It Became The Computer

Age: the Evolution of IBM Computers” (Figure

11.’ Leaving aside the question whether it is proper

to identify any period of time by a piece of tech-

nology, the title does call attention to the fact that

the computer seems to have sprung up suddenly

and unexpectedly. to dominate much of the na-

tion’s technology, economy, and culture.

cades, and not sooner? A complete answer to this

question would include a mix of economic and so-

cial as well as technical factors. This essay fo-

cuses on an aspect of the internal development

of computer technology that was as important as

any: namely that after 1940, Babbage’s concep-

tual formulation of an computer was joined to an-

other technology that was well suited to its re-

alization. That technology was electronics.

Annals of the Hlstory of Computing, Volume 10, Number 4, 1989 . 257

But a sophisticated description of a digital

computer appeared in the writings of Charles

Babbage in the 1830s. Why, then, the appear-

ance

of a “computer age” in the past three de-

‘An earlier version of the paper was presented to the an-

nual meeting of the Society for History of Technology (SHOTi,

October 23, 1986. Pittsburgh, Pennsylvania.

Author’s Address:

Dept. of Space Science and Exploration,

National Air and Space Museum. Smithsonian Institution,

Washington, D.C. 20560. (202) 357-2828.

Electronics emerged as the “technology of

choice” for implementing the concept of a com-

puting machine between 1940 and 1955. As it did

so, it enabled persons not trained in Electrical

Engineering to exploit the power and versatility

of computers. This activity led to the sbudy of

“computing” independently of the technology out

of which “computers” were built. In other words,

it led to the creation of a new science: “Computer

Science.”

The term “coevolution” implies that there was

a continuous and reciprocal interaction between

electronics and computing. Such interaction did,

P. Ceruzzi

Coevolution of Electronics and Computer Science

in fact occur. As computer science matured, it re-

first automatic calculators, finally, after years of

hope and promise, came into existence. But al-

most from the start they were eclipsed by ma-

chines using the much faster vacuum tube as its

computing element. The story of the invention of

the electronic digital computer has been told

elsewhere, and in that story the issue of the vac-

uum tube’s perceived unreliability, as well as its

heavy power demands, are among the difficulties

cited for the initial skepticism as to its practi-

cality. These were indeed serious issues, but they

were addressed. Once they were, vacuum tube

technology, with its higher operating speeds, was

perceived as an alternative to relays.

paid its debt to electronics by offering that en-

gineering discipline a body of theory which served

to unify it above the level of the physics of the

devices themselves. In short, computer science

provided electrical engineering a paradigm, which

I call the “digital approach,” which came to de-

fine the daily activities of electrical engineers in

circuits and systems design.2

Though continuous, the interaction between

Computer Science and Electrical Engineering was

marked by two distinct phases. In the first phase,

between 1940-1955, electronics took over the

practice of computing. In the second, from 1955

to 1975, computing took over electronics. I shall

look at each in turn.

Between 1940 and 1950, a scattered group of

persons, without knowledge of one another, put

Babbage’s ideas into working machinery. These

inventors were interested in building machines

that could carry out a sequence of elementary

arithmetic operations, store intermediate results,

and recall those results automatically as needed,

and display or print the final results of the cal-

culation. They were not, for the most part, con-

cerned with the engineering details of their im-

plementation, except insofar as they wished to

have a machine that worked reliably (Cohen

19851. As things turned out, the first reliable,

working computers-in other words, the first

machines to implement Babbage’s idea of an au-

tomat.ic computing machine-used relays or sim-

ilar electromechanical elements to carry and ma-

nipulate numbers. Using relays (a technology

borrowed from the telephone industry), George

Stibitz of Bell Laboratories and Konrad Zuse of

the Henschel Aircraft Company in Berlin each

built calculators that could carry out three to five

arithmetic operations a second. And using a com-

bination of relays and toothed wheels borrowed

from punched-card accounting machines, How-

ard Aiken at Harvard built a powerful “Auto-

matic Sequence Controlled Calculator” with a

similar operating speed (Ceruzzi 1981).

Relay computers played the vital role of intro-

ducing the concept of automatic, sequential cal-

culation to an often skeptical community. It was

with electromechanical relay technology that the

Throughout this paper I ~111 be concentrating on that

branch of Electrical Engmeenng that IS more accurately de-

scribed as “electronic” engineering This term ~111 be defined

later in the text, but essentially I will not address that branch

of EE that deals with Power Engineering or the so-called

“Heavy Currents.”

One reason for the rapid ascendancy of elec-

tronic devices for computing elements was that

events during the war, mainly unrelated to

building computers, had transformed electronics

itself, raising it above the level considered (and

rejected) by the computer pioneers like Aiken or

Stibitz. One development-radar-was critical,

and became the bridge across which electronics

entered the realm of computing.

The role of radar is not usually considered as

Paul Ceruzzi was born

in Bridgeport,

Connecticut, and

attended Yale University,

where he received a

B.A. in 1970. He

attended graduate

school at the University

of Kansas, from which

he received his Ph.D. in

American Studies in

1981. His graduate

studies included a year as a Fulbright Scholar at

the Institute for the Hisfory of Science in

Hamburg, West Germany, and he received a

Charles Babbage Institute Research Fellowship in

1979. Before joining the staff of the National Air

and Space Museum, he taught History of

Technology at Clemson University in Clemson,

South Carolina.

Dr. Ceruzzi’s main scholarly work has been in

the history of computing since 1935. His has

written a book on this subject (Reckoners, the

Prehistory of The Digital Computer, 1935-

1945, Greenwood Press, 19831, and he is

presently working on a major new gallery at the

National Air and Space Museum about the

computer’s impact on air and space flight.

258

Annals of the History of Computing, Volume 10, Number 4, 1989

P. Ceruzzi

Coevolution of Electronics and Computer Science

IT V&S TO

t-!AVE BEEN THE

NUCLEAR AGE.

IT BECAME...

Figure 1. Brochure from IBM, undated, about 1976

(IBM Corporation).

part of the generational lineage of computer his-

tory, in contrast to, say, the invention and mar-

keting of mechanical adding machines. Yet among

those involved in modern computing’s first de-

cade, there was no question as to its influence.

Radar required vacuum tube circuits that han-

dled discrete pulses of current at high frequen-

cies, in contrast to radio transmitters and receiv-

ers which had been the mainstay of prewar

electronics engineering. Both these requirements

matched the needs of computer engineering. Ra-

dar sets typically contained over a hundred tubes,

again in contrast to the more modest four- or five-

t.ube radio sets of the day.

One link from radar to the computer was the

mercury delay line: an ingenious and tricky de-

vice developed with some difficulty for storing

radar pulses. After the war those who had ex-

perience with it ie.g. Maurice Wilkes at Cam-

bridge University in England and Presper Eckert

in Philadelphia) could adapt it for use as a com-

puter memory device. Those who were less fa-

miliar with it had less success, many (e.g. Aiken)

believing that such a device was so fragile that

it would never work in a computer (Wilkes 1985,

p. 128). Mercury delay lines were indeed difficult

to build and operate; nevertheless they played the

role of being the memory device for four of the

first five stored program computers to be built in

the United States and England:

EDSAC, BINAC,

SEAC,

and Pilot

ACE

(the exception was the

Manchester computer, later called “Mark I,” which

used a Williams-tube memory.3

In 1953, when the

IRE Proceedings

issued a

special “Computer Issue,” Werner Buchholz, the

guest editor, stated that although many com-

puter projects were started during WW II,

. .

Still, the present growth of the computer in-

dustry did not start until the results of the enor-

mous development of electronic technology dur-

ing World War II were brought into the field. It

is interesting to note that many computer proj-

ects started around a nucleus of wartime radar

experts. Electronics not only provided the tech-

nological means for greatly increased speed and

capacity, and thereby enhanced the usefulness of

computers many times, but the availability of

cheap, mass-produced components and of engi-

neers trained to use them made it possible to ex-

periment on a greater scale and at a lower cap-

ital investment than before (IRE 1953, p. 12201.

As a result of that development of electronics

technology between 1939 and 1945, employment

in the American electronics industries had risen

from 110,000

to 560,000 (Electronics

1980, pp. 150-

210).

The vacuum tube ascendancy was not imme-

diate, however. Experience with radar had at-

tacked many of the problems of reliability, but

these problems still remained. Just as serious was

the fact that the much faster operating speeds of

tubes required a rethinking of the overall struc-

ture of a computing machine, especially the way

it received its instructions. High electronic speeds

meant nothing if the computing circuits received

their orders by mechanical devices such as paper

tape or punched card readers. Likewise the high

arithmetic speeds had to be carefully matched to

equally high speeds for storing and retrieving in-

termediate results from memory devices. It was

also recognized that higher arithmetic speeds re-

quired greater memory capacities. Each of the first

electronic calculators (i.e., machines whose com-

puting program was not directed by a stored pro-

“See Table 1 for a full listing of these and other early com-

puters and their memory devices.

Annals of the History of Computing, Volume 10, Number 4, 1989

259

P. Ceruzzi

Coevolution of Electronics and Computer Science

gram), namely the Atanasoff computer, the

ENIAC,

the British Colossus, and the IBM SSEC,

addressed these problems in different, and in

hindsight inelegant, ways. Electronic computing

was held up by the need for a consensus on what

a digital computer ought to look likea

faults) with relays (Stibitz 1945). What that meant

This last bottleneck was broken in 1945, with

the emergence of the concept of the stored pro-

gram principle as the way to organize the various

units of a computer. The origins of this concept

are a matter of controversy, but the informal dis-

tribution, in 1945 and 1946, of a “First Draft of

a Report on the

EDVAC"

by John von Neumann

was what brought the computing community a

general awareness of the concept (van Neumann

1945).

Von Neumann’s report described the

EDVAC

terms of its logical structure, using a notation

borrowed from neurophysiology. The

EDVAC'S

im-

plementation in vacuum tube circuits, though

mentioned, is not the focus of von Neumann’s

energies. Instead he focuses on the main func-

tional units of the computer-its arithmetic unit,

its memory, input and output, and so on. The re-

port also described the idea, and the advantages,

of storing both instructions and data in one, high

speed internal memory.

The “First Draft” had an effect on every aspect

of computing. One effect was to hasten the de-

mise of the relay and assure the place of elec-

tronic circuits as the technology of choice for

building a computer. Once the logical design of

a digital computer was laid out in a way not tied

to a specific technical implementation (as von

Neumann’s was), then it became no more diffi-

cult to construct a computer according to that de-

sign using vacuum tubes than it was to construct

it out of relays or anything else. There were

problems of reliability with tubes, but these were

not overwhelming, nor were they that much

greater than similar problems (e.g. transient

‘It is sometimes argued that those who were skeptical of

vacuum tube technology because of its alleged unreliability

were correct, as evidenced by the fact that vacuum tubes were

eventuallv themselves replaced by the presumably more re-

liable “&d-state” devices such as diodes, transistors, and later

on integrated circuits. Transistors did offer far greater reli-

ability and lower power consumption than tubes, but it was

a full decade after the transistor’s invention that it became

practical to produce transistors in quantity with uniform

characteristics such that they could be used in computers. In

the invervening decade (1948-19581, it was questions first of

logical structure, and then of memory technology, that dom-

inated debates over computer design.

was that for an incremental investment of time

and money to utilize vacuum tube technology, one

got a thousandfold increase in speed. That ad-

vantage was overwhelming, and it meant that the

argument of tubes vs. relays was over before it

had a real chance to begin.

Maurice Wilkes, whose

EDSAC

was among the

first stored program computers to be completed,

in 1949, was among those who saw the Report as

the answer to many of the organizational prob-

lems associated with building computers:

. . .

In [the

EDVAC

Report], clearly laid out, were

the principles on which the development of the

modern digital computer was to be based: the

stored program with the same store for numbers

and instructions, the serial execution of instruc-

tions, and the use of binary switching circuits for

computation and control. I recognized this at once

as the real thing, and from that time on never

had any doubt as to the way computer develop-

ment would go (Wilkes

1985, p. 1091.

As the first stored program electronic com-

puters finally began operating in the early 1950s

their superiority was quickly recognized. The Ab-

erdeen Proving Ground provided a good test en-

vironment-at that facility a variety of mechan-

ical and electronic calculators, relay sequence

calculators, and stored program computers were

installed by 1953. Franz Alt, who was at Aber-

deen at that time, later remarked:

. . I

relay computers were in competition with

them [electronic computers], and they didn’t hold

their own. They were much too slow by compar-

ison . . .

After a few years people lost interest

them. They had been built as an insurance

against the possibility that electronic computing

might not work t Alt 1969).

The range and evolution of machine types, and

the emergence of the stored program approach,

is revealed by the listing in Table 1 of digital

computer installations, broken down by their type

of design. (Analog computers and devices are ex-

amined in a separate section). Table 1 summa-

rizes the various types of automatic computing

machines built and installed between 1940 and

1955. For each machine, a date is given for its

completion or first installation, followed by an

260

Annals of the History of Computing, Volume 10, Number 4, 1989

P. Ceruzzi

Coevolution of Electronics and Computer Science

Name

Table

1, Computer Installations, 1940-l 955.

Year

I. Electromechanical and Electronic Calculators

A. Relay or Mechanical Calculators

No. Installed

Bell Labs Model 1 1940

Zuse Z-3

1941 1

Bell Labs Model 2 1942

Harvard Mark I

1944 1

Bell Labs Model 3

1944 1

Bell Labs Model 4 1945

Zuse Z-4

1945 1

Bell Labs Model 5

1946 2

Harvard Mark II

1947 1

ARC (see text)

1948/52 1

Bell Labs Model 6 1949

ONR/ERA relay computer “Abel”

1950 1

BARK

1950

AURA

1952 1

NEC Mark I (Tokyo)

1952 1

8. Electronic Calculators, General Purpose but Externally Programmed

IBM PSRC

1944 5

ENIAC

1945 1

IBM 603 Multiplier

1946 100

IBM 604 1948

5600 eventually

IBM SSEC 1948 1

Harvard Mark Ill 1949

Northrop Aircraft/IBM CPC 1949

700

Harvard Mark IV i 952

ERA Logistics Computer 1953

Burroughs E-l 01

1955 100

Monrobot

1955 5 approx.

Elecom 50 1955

C. Special Purpose Electronic Calculators

Jaincomp

USAF Fairchild

OMIBAC

Teleregister

SPEEDH

Reservisor

BAEQS

Magnefile

TRADIC

(Bell Labs)

MDP-MSI

Haller Ray & Brown

MIDSAC

1950

1952

1953

1954

1955

D. Digital Differential Analyzers

Northrop

MADDIDA

1949 15

CRC 101 & 105

1951 >5

QUAC

1952

Bendix D-l 2 1954

Wedilog

Annals of the History of Computtng, Volume

10,

Number 4, 1989

261

P Ceruzzr

Coevolution of Electronics and Computer Science

Table i. Computer Installations, 1940-l 955. (continued)

Name Year No. Installed

II. Stored Program Electronic Computers

A. Serial,

“EDVAC

type”

-- ---_.___

BiNAC 1949

EDSAC 1949 1

SEAC 1950 1

Pilot ACE

1951 1

RAYDAC 1951

UNIVAC

1951 46 eventually

C.S.I.R.O. Mark I (Australia)

1952 1

CUBA

(France)

1952

EDVAC 1952

LEO

1952

DYSEAC 1953 1

FLAC 1953 1

MIDAC 1953 1

DEUCE 1954 3

by 1955, 32 eventually

B. Drum Memory, “ERA 1101 type”

ERA 1101

OARAC

Burroughs Laboratory Comp.

CADAC

102

Elecom

100

PTERA

Bendix G-15

CALDIC

CIRCLE

Hughes Airborne

IBM 650

MINIAC

ALWAC

ORDFIAC

WISC

PENNSTAC

LGP 30

READIX

1950

1952

1953

1954

1955

5 approx.

approx.

>400 eventually

5 approx.

>2000 eventually

>lOO after 1955

C. Parallel Memory. “von Neumann type”

Whirlwind

1950

SWAC 1950

Manchester (Ferranti) Mark I

1951 9

AVIDAC

1951 1

IAS (von Neumann)

1951 1

ILLIAC 1952 1

IBM 701

1952

by 1955

MANIAC

1952

NAREC

1952

ORDVAC 1952

ARC (see text)

1948/52 1

ERA 1103

1953 IO

JOHNNIAC 1953

IBM 702 1954

Ferranti Mark II

1954 19

NOW

1954

ORACLE

1954

IBM 704 1955 1 in 1955, many later

262 4 Annals of the Htstory of Computing, Volume IO, Number 4, 1989

P. Ceruzzi

Coevolutlon of Electronics and Computer Science

estimated number of installations iin many cases

form of magnetic cores) until the end of this era.

Perhaps symbolic of this phase of the history

of the computer was the

ARC

computer, built by

A. D. Booth of London’s Birkbeck College, and

first operational in 1948. Booth had followed the

American developments closely, and was con-

vinced early on of the advantages of the von

Neumann, stored program approach to computer

design. He proceeded to design and build a com-

puter along these lines, however using relays in-

stead of vacuum tubes in the interests of saving

money and time. But almost as soon as the

ARC

was completed, he set out replacing the relay cir-

cuits with their equivalent vacuum tube circuits

to implement the same logical functions (Booth

1949).

” this information is approximate).’

The figures verify Alt’s impressions of this era:

relay calculators and computers initiated the dig-

ital era, but they were quickly eclipsed by elec-

tronic devices. Of the electronic machines, gen-

eral purpose calculators having a limited degree

of programmability (e.g. Northrop/IBM CPC, IBM

603/604) were installed in large numbers in the

early part of this period, and served as the work-

horse of digital computation until inexpensive

stored program computers became available, be-

ginning around 1953. Special purpose electronic

machines, including the Digital Differential

Analyzer, likewise fall into this category.

Stored program electronic computers did not

begin to appear in large numbers until around

1953, especially with the introduction of the IBM

650. Those that were installed in large numbers

tended to be the slower, but less expensive, drum

types (such as the 6501. These numbers should be

considered in the context of the greater speed,

memory capacity, and overall computing power

of the large scale machines such as the

UNIVAC

and IBM 701, of which only a few were installed

in the early years.

The “Von-Neumann type” design (stored pro-

gram, parallel memory access), which provided

the fastest performance, and which became the

standard architecture down to the present day,

was slow in being established in the form of in-

stalled machines, even though its advantages were

widely known through a series of reports by von

Neumann and others at the Institute for Ad-

vanced Study after 1946. This was mainly due to

the fact that the viability of the parallel archi-

tecture depended on a reliable, fast, and rela-

tively cheap memory device; something which did

not really become commercially available tin the

‘This table is a summary of a report that has been com-

piled from various sources. primarily the ONR Digital Com-

nuter Newsletter. and the Ballistic Research Laboratorv’s

s~rtqv nf‘Automorzc Conrpulers. three volumes of which were

also published during this interval (Weik 1955) The complete

Itsting. with references for each machine, IS on file at the

Computer Museum. Boston, and the Charles Babbage insti-

tute. Mmneapolis. In lookmg at this table it is important to

recognize that the defmnion of computtng machme was not

constant during that period. to the extent that a machine that

might appear on the table as a “computer” m 1945 would not

qualify as a .‘computer”

by 1955.

However m all cases I base

sought to include machmes that went at least a step beyond

performmg simple arithmetic on a pairs of numbers. but whtch

could with some degree of automatic control evaluate math-

ematical expressions of at least a modest length.

Beginnings of Computer Science,

1955-l 975

Bet.ween 1955 and 1975, a science of computing,

in North America adopting the name “Computer

Science,” emerged. Its focus was on the elec-

tronic, stored program digital computer (with its

magnetic core memory) invented in the previous

decade. As this new science emerged, the elec-

tronics technology from which it sprang changed

in response. The first change was a steady pro-

gression of computer related papers in the Elec-

tronics journals6 Eventually it affected the very

definition of “electronics” itself. Before examin-

ing the emergence of a science of computing after

1955, consider the accepted definition of “elec-

tronics” at that time.

The many changes in the practice of electron-

ics during the Second World War, of which com-

puting was but one, led electrical engineers to

reexamine the nature of the discipline. The orig-

inal definition of electronics was that of the

movement of electrons in a gas or vacuum, and

was intended to distinguish radio and commu-

nications work from the power engineering out

of which Electrical Engineering first emerged

iMcMahon 1984). This definition stemmed from

Edison’s observation in 1883 that an evacuated

tube could carry a current, and from the inven-

There is a steady increase of computer related papers, as

indexed in Science ilhsfracts. B (Electrical Engineering). from

zero in 1945 to 10s; in 1965. In 1960, computer papers were

divided into analog and digital, with analog papers dropping

off to about 25 in 1965. After 1965. as argued later in this

paper, digital computing began to dominate all electronics

papers, so that the frequency of computer papers indexed is

no longer a measure of its dominance in the field.

Annals of the Hlstory of Computtng. Volume 10, Number 4, 1989

263

P. Ceruzzl

Coevolution of Electronics and Computer Scrence

tion of the diode and triode vacuum tubes in the

first decade of the 20th century.7

With the advent of servomechanisms, radar,

computers, and the transistor (which did not in-

volve movement of electronics though a vacuum),

the definition had to change. In a guest editorial

in a 1952 issue of the IRE Proceedings, William

Everitt proposed a new definition:

Electronics is the Science and Technology which

deals primarily with the supplementing of man’s

senses and his brain power by devices which col-

lect and process information, transmit it to the

point needed, and there either control machinery

or present the processed information to human

beings for their direct use (Everitt 1962, p. 899).

In subsequent issues, several readers objected,

saying that the notion of “information” was too

vague. Many felt that Everitt was correct in

broadening it beyond the movement of electrons

in a vacuum, but they suggested that a better,

yet still precise, definition might be something

along the lines of “the movement of electrons, in

solid, gas,

or vacuum (McMahon 1984, pp. 231-

232).”

The increasing awareness of the computer as

a machine that integrates all aspects of infor-

mation handling, including communication, was

implied in a 1959 address by Simon Ramo, of

Hughes Aircraft, to the Fifth National Commu-

nications Symposium, where he proposed a new

term, “Intellectronics,” defined as “the science of

extending man’s intellect by electronics (Ram0

1959).“8 And Zbigniew Brzezinski coined the term

“Technetronics” to describe essentially the same

transformation of society (Brzezinski 19701.

By 1977 the computer-and-information defi-

nition had become accepted, at least as a general

overall definition of electronics, as indicated by

the lead article by John Pierce for a special issue

of Sczence on “the Electronics Revolution”:

‘Shortly after Edison’s observation. J. J. Thompson ex-

plained the effect by hypothesizing that a stream of nega-

tively charged particles carried the current through the vac-

uum. In 1994 these- particles were given the name “electrons”-

hence “electronics”---by the Irish physicist George Stoney.

Alan Turing, whom I shall describe later in this paper as one

of the founders of Computer Science. was a distant relative

of Stoney. (Hodges 1983).

‘Ramo’s term did not catch on. although a decade later a

group of engineers working in what has since become known

as “Silicon Valley” founded a company called “Intel,” a con-

traction of either Ramo’s term or of the words “Integrated

Electronics;” (Hanson 1982, Chapter 51. Today, what in

America is known as “Computer Science” is called “Infor-

matics” in Continental Europe.

mat is electronics? Once we associated elec-

tronics with vacuum tubes, but vacuum tubes are

almost obsolete. Perhaps electronics is semicon-

ductor devices. But then, what of magnetic cores

and bubbles and liquid crystals? I think that

electronics has really come to mean all electrical

devices for communication, information process-

ing, and control . . . (Pierce 19771.’

Recent publications hint at a new definition

that is in the same spirit as Everitt’s 1952 defi-

nition of electronics as a matter of communica-

tion and control. Mainly as a result of the de-

velopment of so-called very large scale integration

tVLSI&--integrated circuits with hundreds of

thousands of elementary devices on one chip-

there is a perception that it is the job of electri-

cal engineers to “manage complexity.” Although

it is still of concern to design the elementary

transistors, resistors, and so on, what is now the

critical issue is how to interconnect thousands,

even millions of similar and fairly simple devices

to one another.

In the final chapter of Karl Wilde’s and Nilo

Lindgren’s book on the history of Electrical En-

gineering and MIT, for example, several current

faculty and administrators are asked to define

what they see as the essence of their depart-

ment. Most agree that the computer, and espe-

cially its implementation in VLSI circuits, had

come to dominate the practice of the electrical

engineer.” For one of the administrators inter-

viewed (Fernando Corbato), “if there is a single

theme . . . it is the problem of complexity (Wildes

and Lofgren 1985).“l’ This last definition, if it

becomes generally accepted, reveals a strong in-

‘Notice that although the definition of electronics-as-in-

formation resolved the question of whether solid-state devices

like the transistor and integrated circuit properly belonged

to electronics, it introduced the confusion of allowing electro-

magnetic relay devices to be included. While in the general

view there is nothing wrong with this inclusion, when applied

to the context of the early digital computer era. it is inap-

propriate.

‘“They further agreed that the decision made in the late

1970s to keep Computer Science as a part of EE and not let

it break off as did many other universities was a wise one;

the name of the department is now Electrical Engineering

and Computer Science. I shall have more to say on this later.

“One example might serve to illustrate the increase in

complexity of eiectronyc circuits that has taken place over the

oast few decades: When the World Trade Center was built m

Lower Manhattan beginning in the late 196Os, it displaced a

block known as “Radio Row”: a group of shops selling surplus

radio parts. mainly of World War II vintage. Though many

lamented the passing of this area, it should be noted that in

terms of active circuits, the entire contents of all the shops

on Radio Row are today contained on one or two VLSI chips.

264

* Annals of the History of Computing. Volume 10, Number 4, 1989

P. Ceruzzi

Coevolutlon of Electronics and Computer Science

fluence from the theory and practice of comput-

ing.

Definition of “Computer Science”

While the definition of electronics had been

changing, a new term appeared, a term whose

definition would also evolve: “Computer Sci-

ence.” Though it is now one of the most popular

subjects taught in universities, “Computer Sci-

ence” has had a variety of definitions.12 Its his-

tory is brief, and any statements as to what it is

become dated quickly. Nevertheless there are

general areas of agreement as to its nature.

One definition, stated in its extreme, is that it

is not a science at all. (In debates found in the

letters column of computer trade and profes-

sional journals, one sometimes reads the aphor-

ism, “Any science that needs the word ‘science’

in its name is by definition not a science.“) A less

extreme form of this statement is that computer

science is driven by electronics technology, and

that computer scientists do little more than ob-

serve and collate into general rules the behavior

of the machines the engineers supply them.

Though useful, such rules are not scientific prin-

ciples such as those on which Electrical Engi-

neering itself was based (e.g. Maxwell’s theory of

electromagnetism, semiconductor physics). As

such, computer science is not a true science but

one of the “engineering sciences,” in Edwin Lay-

ton’s terms, which “took on the qualities of a sci-

ence in their systematic organization, their reli-

ance on experiment, and in the development of

mathematical theory”

(Layton 1971). Computer

science’s rules and laws tend to be about observ-

able and practical things, such as the time or the

memory requirements for a certain program that

sorts a file, and not about the fundamental prop-

erties of computing, whatever they may be.

This perception has its historic roots in the

1945-1955 period, when it took heroic engineer-

ing efforts to get a computer to work at all. The

feeling was that any talk about a theory of com-

puters was premature when it is questionable

whether one could get a stored program com-

puter to run without failure for

than a few

minutes at a time. (The one exception to this was

“A recent survey of colleges and umversttres shows that

in 1983,25.000 bachelor’s degrees were awarded m Computer

Science. compared with

18.000 In Electrical Engmeermg,

11,000 m Chemtstry. 12.600 m Mathemattcs and Statistics.

and 3800 In Physics (Gnes et al. 1986).

of course the “theory” of the stored program as

stated in von Neumann’s

EDVAC

Report.) This at-

titude survives among contemporary computer

engineers, who although often too young to know

of the difficulties of electronic computing’s first

decade, are nonetheless paid only to get a ma-

chine “out the door” of the factory. Other than

having the machine pass certain benchmarks to

validate its performance, they are not concerned

with what the customer (including the computer

scientists) do with it (Kidder 1981).

One of the first general textbooks on elec-

tronic computers stated flatly that “The out-

standing problems involved in making machines

are almost all technological rather than mathe-

matical (Bowden 1953). As computer science ma-

tured, others noted that for at least a century there

had been a tradition in a branch of mathematics

of studying the notion of a mechanistic process,

a tradition that began with George Boole’s In-

vestigation into the Lam of Thought

. . . , pub-

lished in 1854, and which included the work of

Frege, Hilbert Godel and many others on the for-

malization of mathematical expressions (van Hei-

jenoort 1967; Aspray 1980). But this tradition”

. . . was smothered after 1940 by a great tech-

nological explosion” Wegener 19701. In the fa-

mous Moore School lectures held in 1947, shortly

after the public unveiling of the

ENIAC,

there was

almost no mention of any of these men and

women-despite the fact that the title of the lec-

tures was “Theory and Techniques for Design of

Electronic Digital Computers” (Moore School

1947). In particular, the tone of the Moore School

sessions was that any discussion of the theory of

computer design had to take a back seat to the

pressing technical problem of designing a fast and

reliable memory system.‘” (The stored program

computers then being built used for their mem-

ory either mercury delay lines or specially built

television tubes. Both methods had extreme lim-

itations, especially regarding reliability, cost per

“‘For example. Claude Shannon’s work. which showed that

the rules of symbolic logic were well suited as a design tool

in the construction of relay circuits that performed arith-

metic, was hardly mentioned at all. For many of the partic-

ipants at these lectures the important thing was to find out

what kind of hardware was inside the many “block diagrams”

the lecturers kept putting on the blackboards. As noted above.

the one exception was the theory of computer design as de-

scribed by von Neumann.

but

even in this instance there was

a feeling at the sessions that theoretical design for the

EDVAC

was being emphasized too much, in the place of a more nar-

rative description of the

ESIX.

a computer based on an ad

hoc theory but one that was at least functioning in 1937.

Annals Of the HIstory of Computtng,

Volume 10. Number 4, 1989

265

P. Ceruzzi

Coevolution of Electronics and Computer Science

bit of storage, and capacity (Redmond and Smith

1980,.

A few years later, a few companies and uni-

versities had succeeded in building working com-

puters. Among them there was a modest debate

over the value of logical state diagrams as a guide

to computer design, as opposed to straight engi-

neering borrowed from radar circuits that han-

dled electrical pulses. For a while the former was

known as the “West Coast” approach to computer

design, but by the mid 1950s the debate fell by

the wayside as computer design finally estab-

lished itself on a firmer foundation of symbolic

logic (Sprague 1972).

Another indication of how technology drove

perceptions of computing comes from the way the

history of computers was marketed. When IBM

introduced the System/360 series of computers

in 1964, they helped promote the notion of com-

puters belonging to three “generations.” defined

according to the technology by which they were

implemented: vacuum tubes. transistors, and in-

tegrated circuits. This classification, which has

since become accepted and even expanded on (viz.

the Japanese “Fifth Generation” project), had at

least two effects on the perception of the history

of computing: first, it relegated all computer ac-

tivities before the

ENIAC

into a limbo of either

“prehistory” or irrelevant prologue; second, it de-

fined progress in computing strictly in terms of

the technology of its hardware circuits.

Recently this view of computer science being

technology driven has been repeated by C. Gor-

don Bell, for many years chief of engineering for

the Digital Equipment Corporation and one of the

inventors of the minicomputer. Speaking of the

“invention” of the personal computer in the late

1970s he said:

A lot of things are called inventions when, ac-

tually, they were inevitable. I believe technology

is the driving devil. It conspires. and if there’s a

concept

half-there or a computer half-designed.

technology will complete it (Bell 19851.

Elsewhere, he has said of the computer industry:

It is customary when reviewing the history of an

industry

ascribe events to either market pull

or technology push. The history of the com-

puter industry . is almost solely one of tech-

nology push (Bell et al. 1978).

And in Tracy Kidder’s chronicle of the team of

young engineers in 1973 who were bidding a new

computer for Data General:

Some engineers likened the chips to an unassem-

bled collection of children’s building blocks. Some

referred to the entire realm of chip design and

manufacture as ‘technology,’ as if to say that

putting the chips together was something else. A

farmer might feel this way: ‘technology’ is the new

hybrid seeds that come to the farm on the rail-

road, but growing those seeds is a different ac-

tivity-it’s just raising food (Kidder 1978, p. 122).

After 1955 there arose compelling arguments

that computer science was a genuine science, al-

beit one that differed in many ways from the

classical sciences. The first argument to appear

was one that emerged in response to the pressure

that computing activities were putting on tradi-

tional disciplinary boundaries, especially in the

universities. With one exception (Wiesner, noted

below), the first published statements about

“computer science” revealed a perception that a

science was

being born, and it needed to be es-

tablished at least on organizational and admin-

istrative grounds; the question of just what it

“was” could be answered later. By the late 1950s

it was recognized that many topics that had much

in common with each other (and all in common

with the computer1 were being taught in various

departments around most universities. The feel-

ing was that those who were concerned with the

computer aspects of their work in these other de-

partments would perhaps not be recognized and

adequately rewarded by their peers for doing good

work. Establishing a separate department of

“computer science” would address that concern.

By the second volume (1959) of the

Commu-

nications

the ACM

(the flagship journal for the

Association for Computing Machinery), the term

“computer science and engineering” had begun to

appear. That September, an article entitled “The

Role of the

University in Computers, Data Pro-

cessing. and Related Fields,” by Louis Fein, dis-

cussed the need to consolidate, under a single or-

ganizational entity, the various studies orbiting

around the computer in various academic de-

partments such as business and economics,

mathematics, linguistics, library science, phys-

ics, and electrical engineering. After mentioning

a few names for this entity, including “informa-

tion sciences” (which he attributed to Jerome

Wiesner), “intellectronics” (which he says was sug-

gested by Simon Ramo‘l, and “synnoetics” (which

Fein himself had suggested elsewhere) he sug-

gested “the ‘computer sciences”‘; later in the ar-

ticle shortened to “Computer Science,” (singular,

and in quotations). This I believe is the origin of

266 - Annals of the History of Computing, Volume 10, Number 4, 1989

Ceruzzi

Coevolution of Electronics and Computer Science

the term [Fein 1959; Fein 1961; Leech 1986).

Computer Science is concerned with information

processes, with the information structures and

proceduresthatenterinto representations of such

processes and their implementation and mfor-

mation processing systems. . The central role

of the digital computer in the discipline is due to

its near universality as an information process-

ing machine. With enough memory capacity, a

digital computer provides the basis for modeling

any information processing system, provided the

task to be performed by the system can be spec-

ified in some rigorous manner. , the stored

program digital computer . . . provides a meth-

odologically adequate, as well as a realistic, ba-

sis for the exploration and study of a great va-

riety of concepts, schemes, and techniques of

information processing (Amarel 1976).

In describing what this new discipline was, Fein

made the further point about what it was NOT:

“Too much emphasis has been placed on the com-

puter equipment in university programs

that in-

clude fields in the ‘computer sciences’ . . . In-

deed an excellent integrated program in some

selected fields of the computer sciences should be

possible without any computing equipment at all,

just as a first-rate program in certain areas of

physics can exist without a cyclotron” (Fein 1959,

p. 11).

The establishment of Computer Science’s ad-

ministrative and organizational boundaries was

followed five years later by the first attempts to

establish it as a true science based on its internal

nature. These attempts centered on the concept

that, like any other, Computer Science is the sys-

tematic study of certain phenomena, only in this

case the object of study is an artificial, not a nat-

ural one. In other words, Computer Science is

simply the study of computers. It is not to worry

that computers are artificial and not natural

phenomena. This was the argument of Herbert

Simon, author of

The Sciences of the Artificial,

and of his colleagues Allen Newell and Alan Per-

lis, then at Carnegie-Mellon University, whose

letter to the editor of

Science

in 1967 was the first

explicit statement to this effect (Newell et al.

1967).

For Newell, Perlis, and Simon, Computer Sci-

ence is the study of computers, just as Astronomy

is the study of stars. But in Astronomy, stars will

be stars, no matter what the astronomer says

about them. This is not so with computers. If what

the computer scientist says about computers in

theory does not agree with observed behavior, he

or she can always change the computer (more

correctly, get an electrical engineer to change the

computer). For most of the period 1950-1980,

there was sufficient continuity of the von Neu-

mann, stored program model of computer archi-

tecture that this did not present a problem.‘” As

long as the basic architecture remains constant,

so too will the definition emphasize the non-

hardware aspects of computing, such as in the

1976 Encyclopedia of Computer Science:

“Lately. with the development of so-called parallel mul-

tiprocessors, things have changed, and there is a correspond-

ing change in the attitude of computer science that increas-

ingly sees hardware issues as relevant. We may now see

return to the pre-1950 era when hardware issues dominated

discussions of the theory of computing. I discuss this issue

further in a later section of this paper.

Since the publication of the Newell et al. letter

Science,

a different internal definition has

emerged, and it is this one which dominates the

day-to-day practice of computer science today, es-

pecially in the universities. That is the definition

of computer science as the study of algorithms-

effective procedures -and their implementation

by programming languages on digital computer

hardware. Implied in this definition is the notion

that the algorithm is as fundamental to comput-

ing as Newton’s Laws of iMotion are to physics;

thus Computer Science is a true science because

it is concerned with discovering natural laws about

algorithms which are not engineering rules of

thumb or “maxims,”

in Layton’s terms (Layton

1971. p. 566; Vincenti 1979). Computer Science

thus becomes a science because it has a theoret-

ical foundation on which to build, such as: “The

notion of a mechanical process and of an algo-

rithm (a mechanical process that is guaranteed

to terminatei are as fundamental and general as

the concepts that underlie the empirical and

mathematical sciences (Wegener 1970, pp. 70-

781.” It is no coincidence that this theoretical

foundation is essentially based on the work of

Hilbert, Turing, Church, and others whose pre-

1940 work in mathematics was neglected by those

building the first electronic computers of the

1940s.‘”

The event that, more than any other, gave the

algorithmic basis currency was the publication,

in 1968, of a book entitled

Fundamental Algo-

“A few texts in Computer Science assert that the funda-

mental princiole of Cornouter Science is the so-called Tur-

mg-Church gypolhes~s, khlch, informally stated, says that

all algorithmic procedures are equivalent to a class of math-

ematical functions known as general recursive functions (As-

pray 1980, Ch. 2). At the same time it should be noted that

few if any textbooks in Computer Science devote much space

to an elaboration of this hypothesis.

Annals of the History of Computing, Volume 10, Number 4, 1989

267

Ceruzzi

Coevolution

Electronics

and

Computer

Science

rithms

by Donald Knuth (Knuth 1968).” In- The notions of a mechanical process and an al-

tended ai the first of a planned seven-volume se-

ries on “The Art of Computer Programming,”

Knuth’s book was a conscious effort to place com-

puter programming on a foundation of mathe-

matical principles and theorems. The book spar-

kled with wit and erudition. It was a specialized

text that many nonspecialists could read and en-

joy. (One of its most memorable passages was a

13-page, detailed analysis of the algorithms im-

plied in getting an elevator in the Mathematics

Building at the California Institute of Technol-

ogy to work correctly. Although the elevator it-

self worked fine, Knuth later recognized with a

shrug that his algorithmic analysis contained

several fatal “bugs.“) Throughout the book, Knuth

codified and formalized a wealth of computing

techniques and tricks that had been informally

known for years among computer program-

mers. *’

Another of Knuth’s strengths was his com-

mand of history. A large portion of each volume

discussed the historical context of computing; in-

deed, Knuth was one of the first to recognize and

appreciate the advances in mathematics

(com-

putational

mathemat.ics) made by the ancient

Babylonians and by Europeans during the Mid-

dle Ages. These historical passages, far from being

diversions from the main thrust of the text, gave

strong support to the notion that “computer pro-

gramming,” if defined properly, was part of a long

scientific tradition.

Fundamental Algorithms

consciously defined

the study of algorithms as a subject that was in-

dependent of any machine that might implement

them. Recognizing the need for a programming

language to describe algorithms of interest

throughout the book, Knuth deliberately intro-

duced

“MIXAL,”

a language developed especially

for the book and one not implemented on any ma-

chine.”

‘“Volumes 2 and 3 of the series (Semrnumerzcal Algo-

rlthms and Sortzng and Searchzngl appeared m 1969 and 1973,

respectively. Volumes 4 through 7 have yet to appear. In 1976,

Knuth hlnted that more volumes were forthcommg. but as of

this writing (19881. they have not iKnuth 1976; 1982).

“One example from Knuth’s third volume m the series may

serve to Illustrate: ‘.By 1952, many approaches to mternal

sorting were known In the programmmg folklore, but com-

paratively little theory had been developed None of the

computer-related documents mentioned so far actually ap-

peared in the ‘open literature’; in fact, most of the early his-

tory of computing appears in comparatively inaccessible re-

ports“ (Knuth 1973. pp. 386-3871.”

‘“Shortly after the appearance of the book. however, com-

puter programmers wrote programs, called “cross compilers,”

gorithm, whether Computer Science’s first prin-

ciples or not, did form an informal basis for the

way the subject has been taught in universities

since the mid-1960s. But a rigorous analysis of

the Turing-Church Hypothesis or of Turing’s 1936

paper “On Computable Numbers . . . ” was rarely

found in introductory Computer Science Courses.

What was more likely to be found was a smat-

tering of history that included a brief mention of

formal computability, quickly followed by the meat

of the course: programming in one or more high-

level computer languages. In short, Computer

Science t.extbooks adopted Knuth’s strategem of

using a programming language as a medium for

describing algorithms; unlike Knuth, most uni-

versity and college texts introduced actual pro-

gramming

languages

(e.g. FORTRAN, PASCAL),

whose programs could and were executed by the

students on their university computers.

The present curriculum of the majority of

Computer Science programs st.ems from a series

of reports published in the journals of the Asso-

ciation for Computing Machinery, and the cur-

riculum that evolved reveals a trend away from

hardware concerns toware more and more of a

mathematical basis (as taught, however, the

mathematical level of many computer science

courses is not high). The version published in 1968

(the so-called “Curriculum ‘68)” was especially

influential; one person calls it comparable to the

EDVAC

report as a founding document of aca-

demic computer science (Pollack 1977). In its first,

preliminary version, “electronics” appears as an

optional elective under “supporting” courses

(Figure 2) (ACM 1965). In “Curriculum ‘68”

hardware courses are gone completely, to be re-

placed by an algorithmic approach and an em-

phasis on languages and data struct’ures (ACM

1968). In the latest version the algorithmic focus

remains, with more mathematics introduced in

early stages (ACM 1977).

A more recent definition has appeared, one that

echoes the one thing is emerging in electrical en-

gineering. That is the perception of computer sci-

ence as the study of complexity, at all its levels.

The notion that there is a new science of the

management of complexity appeared in one of the

that allowed a range of then-exlstmg computers to execute

MIML. statements. But no

“MIX”

computer, which Knuth called

“the world’s first polyunsaturated computer very much

like nearly every computer now in existence. except that It

is. perhaps, nicer,” was ever built (Knuth 1968. p. 1201.

268

- Annals of the History of Computing, Volume 10, Number 4, 1989

P. Ceruzzi

Coevolution of Electronics and Computer Science

COMPUTER SCIESCE COURSE5

Table of Cour-es for Computer Science Jiajors

T.\BLE 1. PR~LI~IS~RY RECOUXEXD\~-10x3 OF THE CCRRICKLCU COUUITTEE OF hC.\i

Required

FOR \[AJORS 1X &,SIPCTER $ClESCE

0. CO\~BISATORICS

.\SD GRAPH TUEORT

5. .iLGURITHMIC 3. SCliERlC.,L

L.kSGU.\GES .\SD

CALWLUS (or

COMPILERS

Course ;j

3. COSSTRUCTIVE

LOGIC

1. ISrRoDrnxo~

TO ACTO!dATA

THEORT

5. FOR\IAL

11. SYSTEUS

SIMULlTlONS

12. NATHE~UTIC+L

OPTI\IIZ.~+IO?I

TECHM~~ES

16. HEURISTIC

LINCUAGES

.4lgebraic Structures

Statistical 1Iethods

DiRerential

Equations

.\dvsnced Calculus

Physics (6 cr.)

Aaslog Computers

Electronics

Probability and Sta-

tlstlcs Theory

IAn~ulstxs

Logx

Pbilosophp and Phi-

losophv of Science

Figure 2. Preliminary curriculum from the ACM’s Committee on Computer Science, 1965. [Note the single course

on “electronics” under the heading of “supporting” and “other electives.” ~from ACM Curriculum Committee on

Computer Science, “An Undergraduate Program in Computer Science: Some Recommendations,

Communications

ACM 1965, p.

5461.1

first published statements about what we now call

“computer science.” In an address at a ceremony

opening the IBM San Jose Laboratory in 1958.

Jerome Wiesner made the following comments:

Information processing systems are but one facet

of an evolving field of intellectual activity called

communication sciences. This is a general term

which is applied to those areas of study in which

interest centers on the properties of a system or

the properties of arrays of symbols which come

from their organization or structure rather than

from their physical properties; that is, the study

of what one MIT colleague calls ‘the problems of

organized complexity’ (Wiesner 1958 1. I9

Currently this definition’s most vocal proponent

has been Edsger W. Dijkstra. Beginning in the

‘“The “MIT colleague” is not identified.

late 196Os, DiJkstra consistently argued that de-

spite an apparent basis on algorithms, Computer

Science departments were teaching only engi-

neering rules of thumb about programming lan-

guages. For Dijkstra it was (and is) imperative

that Computer Science distance itself from not

only hardware issues but also from the mastery

of programming languages as its principal activ-

ity in the schools. His writings, which often take

the form of brief notes. serially numbered and

privately circulated to his colleagues, echo Wies-

ner’s statement as well as those of MIT’s Elec-

trical Engineering Department, for example:

. . .

[Nlow the teaching of programming com-

prises the teaching of facts-facts about sys-

tems. machines, programming languages etc.-

and

it is very easy to be explicit about them, but

the trouble is that these facts represent about 10

Annals of the History of Computing, Volume 10, Number 4, 1989 -

269

P. Ceruzzb

Coevolutlon of Electronics and Computer Science

percent of what has to be taught; the remaining

90 percent is problem solving and how to avoid

unmastered complexity. (Dijkstra 1982, p. 1071

. . But programming, when stripped of all its

circumstantial irrelevancies, boils down to no

more and no less than the very effective thinking

so as to avoid unmastered complexity (Dijkstra

1982, p. 163).

For him, the goal of computer science was to con-

cern itself with the attempt “to define program-

ming semantics independently of any underlying

computational model . . . ” or, in other words, to

“forget that program texts can also be inter-

preted as executable code” (Dijkstra 1982, p. 275).

But although Dijkstra is held in high esteem by

academic computer scientists, Computer Science

as it is taught (especially in the United States)

emphasizes the study and mastery of existing

programming languages (and operating systems)

that can be executed on existing digital com-

puters.

Based on these observations, Computer Sci-

ence, as it was formally recognized and taught

between 1955 and 1975, was an engineering sci-

ence, according to Layton’s term. It also fits Wal-

ter Vincenti’s criteria for engineering science in

that in its first two decades. progress in Com-

puter Science occurred in the absence of any for-

mal or useful theory (Vincenti 1979, pp. 742-746.”

But in the context of its roots in formal mathe-

matics iand in its ever increasing levels of ab-

straction and formality since 1975) it is now a

pure science, albeit one that is still groping for

an agreed upon set of fundamental principles and

one that has a different character from classical

physics or chemistry.

The issue of what is Computer Science ironi-

cally has little to do with its having been shaped

by administrative, government, military, and

university policies-indeed, the same sort of pol-

icy factors are characteristic of nearly all post-

World War II science, including (even especially)

physics. Computer Science concerns the system-

atic study of algorithms, especially in the expres-

sion of those algorithms in the form of computer

programs that can be executed on commercially

‘“In one aspect Computer Science represents a departure

from Vincenti‘s thesis. That is his assertion (Vincenti 1979,

p. 7461. that “. the use of working scale models is

peculiar to technologp. Scientists rarely. if ever, have the pos-

sibility of building a working model of their object of con-

cern.” As noted bv Newell et al. above. the object of study for

Computer Scienchis precisely such a model-a universal model

at that.

sold digital computers. Computer hardware, and

hence electrical engineering, are part of com-

puter science, but at present the university struc-

ture of computer science departments treats com-

puting hardware as a given, and the more one

can ignore purely hardware issues the more

progress can be made on the study of algorithms.

To a lesser extent there is a trend to look at ex-

isting programming languages in the same way.

Analog vs. Digital

Despite the many pieces of common ground be-

tween computing and electronics. the two activ-

ities remained distinct. One reason was due to a

fundamental difference in the ways each disci-

pline approached the handling of signals of elec-

tron currents, a difference that had characterized

the evolution of each discipline from its earliest

days. Electrical Engineering evolved as a study

of using devices (like the vacuum tube) to am-

plify continuous signals (McMahon 1984). Com-

puting Engineering, later on Computer Science,

was concerned with using electrons to count and

switch. The one was analog, the other digital.

Analog computing devices, electronic or other-

wise, belong to the history of computers (the

ENIAC

owed as much a debt to wartime analog comput-

ing projects as it did to radar or to digital me-

chanical computing). But analog computers do not

belong to Computer

Science,

as the discipline es-

tablished itself in the late 1950s. The reason is

simple: Computer Science centers on the pro-

grams that execute algorithmic procedures; but

whatever advantages analog have over digital

machines, their inability to be programmed eas-

ily put them forever at a disadvantage, and pre-

clude their being part of the discipline. Several

examples illustrate this difference.

The first concerns the fate of the various Dif-

ferential Analyzer projects, centered at MIT un-

der the leadership of Vannevar Bush. These ma-

chines were certainly among the world’s first

“computers,” in the sense that they were the first

machines capable of automatically evaluating

fairly complicated mathematical expressions. Yet

they never really fulfilled their promise, and of

all the reasons this was so, it was the difficulties

involved in reprogramming them for different

tasks that was decisive (Owens 19861.”

“Owens argues. for example, that it was the Rockefeller

Analyzer’s inability of to be reprogrammed easily that was

270 - Annals of the HIstory of Computing, Volume 10, Number 4, 1989

0, Ceruzzi

Coevolution of Electronics and Computer Science

Another example concerns the introduction of

digital computing to the aviation industry. This

community, consistently one of the largest cus-

tomers for computer equipment, had a long tra-

dition of using analog computing devices for air-

craft stability and control, as well as for ground

based applications such as missile tracking and

guidance. But at the same time this industry was

among the first to adopt the digital approach, as

soon as it felt that the problems of reliability and

size could be managed. They did so, despite nu-

merous engineering difficulties, again because of

the digital computer’s greater flexibility ice-

ruzzi, 1989).

lished the notion that the study of the software

side of computing was a valid activity, and it val-

idated the departments of Computer Science (not

Electrical Engineering) as the places where this

study would take place. Computer Science thus

emerged and was split from Electrical Engineer-

ing between 1940 and 1970 as a result of the res-

olution of tha analog/digital split in favor of dig-

ita1.22

And at MIT in the early 195Os, a strong re-

search program had developed on automatic con-

trol of machinery, especially for automating fac-

tory processes and production. This work had its

roots in the (essentially analog) engineering of

MIT’s Servomechanisms Laboratory. By 1952,

when a special issue of Scientific American on

“Automatic Control” appeared, it was recognized

that digital techniques were preferable in all but

one aspect. That was the ability of analog devices

to operate in “real time.” But it was also noted

that the steady progress of digital computing in-

dicated that these machines would soon have the

speeds necessary for such operations. And when

they did, automatic control would be done by dig-

ital computers (Ridenour 1952).

Throughout the 1950s and 196Os, advances in

digital computer programming permitted its in-

cursion into areas where analog devices had for-

merly held sway: consider the replacement of the

slide rule by the electronic calculator, or the re-

placement of the engineer’s drafting table and

machinist’s jigs by CAD/CAM (Noble 19841. The

final blow, and the event which completed Com-

puter Science’s break with Electrical Engineer-

ing, was the discovery of the Fast-Fourier Trans-

form, an algorithm which permitted digital

computers to tackle signal processing and anal-

ysis, a discovery which “thus penetrated the ma-

jor bastion of analog computation” (Newell 1982).

In short, analog computing faded because no

one was able to build an analog computer that

had the property of being universally program-

mable in the sense of a Turing Machine. And in

triumphing over analog, digital computing estab-

the main cause for Its quick demise. Iromcally, in grappling

with this problem for the Rockefeller Analyzer, Perry Craw-

ford and others at MIT developed their plugboard system of

programming for it. a techmque which was then apphed to

the programming of the

ESIAC.

Computer Science Takes over Electronics

Electronics technology took over computing; af-

ter Computer Science established itself, it repaid

the debt by taking over electronics. That is, the

notion of using electronic components as digital

switches, to perform functions that are specified

not by their circuit wiring but by “software,” came

to dominate the activities of the electronics en-

gineer. This notion had its origins in von Neu-

mann’s

EDVAC

Report. and by the end of this pe-

riod has been formalized by computer scientists.

They in turn furnished a paradigm that became

an organizing force for the practice of electronics

engineering above the physics and engineering

of the device level. In other words, in the early

decades of computing (1940-19601, the theory of

computing drew its conceptual framework from

electronics. In the next two decades (1960-1980)

computing supplied to electronics a paradigm,

namely the digital approach, that has reshaped

that discipline.

Consider the following example: the 25 Octo-

ber, 1973 issue of Electronics was devoted to “The

Great Takeover.” The magazine’s publisher in-

troduced the issue by saying:

The proliferation of electronics’ multifarious

technologies into new products. new applica-

tions. and new markets-indeed, into new ser-

vices never before considered possible-is the

theme of this special issue of

Electronics.

On the

cover, we have called the pervasive movement of

electronics into just about every area of human

endeavor ‘The Great Takeover.” And, in many

ways it was inevitable, as the cost advantages of

“The comblmng of the two m MIT’s admimstratlve struc-

ture IS an isolated case. To the extent 1’~ points to a future

trend. It is because dlg7tal computing has run up against some

fundamental lm-uts. including the basic quantum granularity

of materials. Therefore progress in computing may require a

turnmg away from the basic principles on which Computer

Science has been founded-mcludlng. among others, the su-

perlorlty of digital over analog.

Annals of the History of Computing, Volume 10. Number 4, 1989

271

P. Ceruzzl

Coevolutlon of Electronw and Computer Science

electronic technologies and cost-effectiveness of

electronic solutions took over more and more jobs

from the venerable mechanical and electrome-

chanical technologies.

(Electronics 1973,

ing techniques, and their expression in the mi-

Inside. a loo-page section described in detail how

electronic devices were rapidly sending older

technologies to the scrap-heap. Examples in-

cluded retail sales, (where point-of-sale terminals

were replacing cash registers), pocket calculators

replacing slide rules, electronic circuits replacing

mechanical clockwork in watches, computers in

banking replacing older posting and accounting

techniques, and many others.

Although nowhere was it explicitly stated in

this issue, the reason electronics was taking over

the world was that digital circuits were taking

over electronics. Every example given described

a digital circuit. Buried at the end of the section

on “technology” was a half-page piece entitled

“Don’t Forget Linear” (p. 84). The implication was

that “linear”-that is, analog-circuits were in-

deed all but forgotten. And of the linear circuits

described, a large percentage were those that

performed a conversion between analog and dig-

ital.

After about 1973, Electronics Engineering be-

came digital computer engineering. Radio, and

communications applications. from which elec-

tronics sprang and which dominated it in its ear-

lier period, were still there, but insofar as they

were, they were treated as a subset of digital

techniques. Even the humble radio, from which

modern electronics engineering grew, has now lost

its tuning dial to a calculator style digital key-

pad. Thanks to the mass produced microproces-

sor, it has become easier to take a computer and

program it to “act like a radio,” than it is to de-

sign and build a radio from scratch.

Analog circuits, now called “linear applica-

tions.” are still found of course, but they occupy

an inferior position. Many do believe however that

progress in computer engineering will hinge on

a redefinition and breakthrough in analog cir-

cuits, as digital circuits approach the physical

limits of the ultimate granularity of matter

(Sutherland and Mead 1977).” Digital comput-

‘“In recent years

computer

science has had to return to a

closer look at technological questions. This phenomenon is

outside the scope of this paper, but briefly it can be sum-

marized as a reaction to the introduction of the microproces-

sor in 1974. which has driven the cost of computing to near

zero. Ivan Sutherland, a founder of computer graphics, and

Carver Mead. one of the founders of a theory of VLSI, summed

croprocessor, offer overwhelming advantages over

any other approaches to, say, building a radio or

an automatic control system or whatever. Thus

it has become not only possible but compelling to

recast as much of electronics practice into a dig-

ital computing mold. Increasingly, digital com-

puting appears as a natural extension of the very

properties of electronics that have always been

part of its appealing characteristics.

Conclusion

Electronics took over computing in the late 1940s

because of its inherent advantages over other

techniques. Digital computing, the theory for

which grew out of Computer Science, took over

electronics because it provided a path for those

inherent advantages to progress. In sum, those

advantages are as follows:

Like electronics in general, digital computing

offered speed. The “instant” communications of-

fered by the Morse telegraph was matched by the

relentless drive by the computer engineer for

faster switching circuits, and by the computer

scientist’s development of algorithms that do not

“blow up”-take up exponentially greater num-

bers of cycles as the complexity of the problem

increases by a small increment.

Like electronics in general, digital computing

offered leverage. One early notion of electronics

(still used in Europe) was that it concerned the

applications of “weak currents,” in contrast to the

“strong currents”

of traditional electrical engi-

neering. Weak currents of electricity, carried on

thin and light wires or as weightless signals

through the ether, do the heavy work of carrying

messages, motion pictures, and signals that con-

trol heavy machinery. With computers is it the

same: a tiny chip and its accompanying ethereal

software do the heavy work of “crunching num-

bers” and moving and processing huge quantities

of data.

up this phenomenon as follows: “Computer science has grown

up in an era of computer technologies in which wires were

cheap and switches were expensive. Integrated circuit tech-

nology reverses the cost situation. making switching ele-

ments essentially free and leaving wires as the only expen-

sive component.” (Sutherland and Mead 1977. pp. 210-228).

As a result, the theory of computing has to be revised. but as

of their writing (1977) this revision had only begun. With the

advent of the microprocessor it is now more practical to stamp

out very complex computers on a chip, and then deliberately

hobble them to do a more mundane task. than it is to design

and build from scratch the simpler circuit to do that mundane

task.

272

Annals of the History of Computing, Volume 10. Number 4, 1989

P. Ceruzzl

Coevolutlon of Electronics and Computer Science

Finally the digital approach extended the con-

cept in electrical engineering of the separation of

a machine’s structural design from its implemen-

tation. When electric power was introduced into

the factory, one of its major advant.ages was that

it allowed the machinery of production to be ar-

ranged on the factory floor according to the logic

of production, without regard to the logic of the

distribution of power as had been the case with

mechanical transmission. The design tool of the

electronics engineer has been the schematic dia-

gram-unlike, say, the architectural drawing-

a diagram that enjoys the luxury of postponing

the physical and spatial details of implementa-

tion until later. During the Second World War,

aviation electronics gave us the “black box”-ra-

dio, radar, and control systems whose internal

organization were invisible, and irrelevant, to

those using it.

From Computer Science came an explicit di-

vision of a technical process into “hardware” and

“software.” Progress in the latter depends on the

fact that programmers need not be concerned with

the details of the former. It is this separation that

electrical engineering has siezed upon. By defer-

ring the thorny issues of applications to a later

phase of software development, the electrical en-

gineer is better able to cope with the physical

problems of constructing circuits having hundreds

of thousands of individual active devices, each of

which has to mesh with all the others. It is in

this sense that the computer scientist has repaid

the debt they owed to electrical engineering, by

providing a way for the engineer to construct cir-

cuits several orders of magnitude more complex

than were possible otherwise. Digital engineer-

ing has become a paradigm for electrical engi-

neering, in Kuhn’s sense of general organization

of a body of theory, whose acceptance by a com-

munitv facilitates their day-to-day work (Kuhn

197oG”

“Sly discussion of the “digital paradigm” for modern Elec-

trical Engineering is based on Kuhn’s earliest writings on the

subject. The subsequent elaboration and refutation of Kuhn’s

thesis. by numerous writers including Kuhn himself. does not

in my view affect the validity of his original insight as a way

of understanding the practice of what he calls “normal sci-

ence.”

in this case normal

engineering

science. The distinc-

tion between hardware and software is present in all tech-

nology. though usually never so easy to discern as in a

computer. It‘is furthermore a distinction which exists in a

hierarchy of levels-the microprocessor. for example. is such

a versatile circuit because it does not require that its func-

tions be fixed as it leaves the factory. It leaves the electronics

engineer free to concentrate on the job of fabricating the chip

itself, without having to worry at that time about precisely

what it will

used for.

But as the paradigm of the stored program

digital computer came to pervade the nature of

electronics, at the same time important distinc-

tions remained between computer science and

electronics technology. The two did not become

synonymous, and with the exception of a few in-

stitutional settings, they are likely to remain

separate academic disciplines (Figure 31. Those

who called themselves computer scientists, though

their livelihood depended on the existence of

computer technology, distanced themselves from

hardware issues insofar as possible. They did so

in part for institutional reasons, but also because

they felt a need to focus their energies on estab-

lishing a foundation, based on scientific princi-

ples, for the construction and analysis of algo-

rithms that a computer implements. This was an

activity they felt would never be considered a

proper branch of electronics engineering.

By the end of the 1950s electronics engineers

agreed with computer scientists that their work

concerned the processing, storage, and transmis-

sion of information, although the definition of

“information” was neither precise nor the same

for each group. Twenty years later, both groups

have begun to see their work as the “manage-

ELECTRICAL ENGINEERING

COMPUTER SCIENCE

lDlglta1 c:rcuits

/, \

:osi5;es )

,;lr;ults I

j fbrlng-Church i

j r;y;c!w;!s

Figure 3.

The Merging of Computer

Science and Elec-

trical Engineering?

Annals of the History of Computing, Volume IO, Number 4, 1989 *

273

P. Ceruzzi

Coevoiutlon of Electronics and Computer Sctence

ment of complexity,” a term once again not pre-

cisely defined. Both disciplines continue to evolve

and change rapidly. For both Electronics Engi-

neering and Computer Science, the present def-

initions, institutional settings, and research pro-

grams will evolve, and coevolve, in the future.

Acknowledgments

An earlier draft of this paper was presented at

the Annual Meeting of the Society for History of

Technology (SHOT), October 1986. The author

wishes to thank W. David Lewis, Edwin T. Lay-

ton Jr., William Aspray, Michael Williams, Mi-

chael Dennis, and Paul Forman for their critical

reading and comments on that draft. The Smith-

sonian Institution’s Research Opportunities Fund

contributed financial support toward the prepa-

ration of this paper.

References

ACM Curriculum Committee on Computer Science.

1965. “An Undergraduate Program in Computer

Science: Some Recommendations,”

Communications

ACM 8 tseptember), pp. 543-552.

ACM Curriculum Committee on Computer Science.

1968. “Curriculum 68: Recommendations for Aca-

demic Programs in Computer Science.”

Commune-

cations

ilCM 11 (March), pp. 151-197.

ACM Curriculum Committee on Computer Science.

1977. ACM SIGCSE Bulletin. 9.

pp. l-16.

Alt. F. 1969. Interview. Smithsonian Computer His-

tory Project.

Amarel. S 19T6. “Computer Science.” m -4nthony Ral-

ston fed.\.

Encyclopedia

Computer

Sczence. New

York: Van Nostrand, pp. 314-318.

Aspray, W. 1980. “From Mathematical Constructivity

to Computer Science: ,4lan Turing, John von Neu-

mann. and the Origins of Computer Science in

Mathematical Logic,” Dissertation, University of

Wisconsin, 1980,

Bell,

C. G. 1985. Computerrcorld,

Oct. l-1, 1985, p. 19.

Bell. C. G.. J. C. 1/Iudge. and J. McNamara. 1978.

Computer Engineering: .4 DEC Vzeu of Hardware

Systems

Design.

Bedford, Massachusetts: Digital

Press, p, 27.

Booth. A.D. 1949. “The Physical Realization of an

Electronic Digital Computer.”

Electronzc Engzneer-

ing 21

(19491. p. 234: 22 119501, pp. 492-498.

Bowden, B.V. 1953.

Faster Than Thought.

London, Pit-

man, p.

30.

Brzezinski,

1970.

Between Two Ages: America’s Role

in the Technetronx Era,

New York: Viking.

Ceruzzi, P. E. 1981.

Reckoners: The Prehtstory of the

Digltal Computer, 1935-1945.

Westport, Connecti-

cut: Greenwood Press.

Ceruzzi,

P. E.

1989.

Beyond the Limits: Flight Enters

the Computer

Age, Cambridge, Massachusetts: MIT

Press.

Cohen, I. B. 1985. Foreword to the 1985 reprint of A

Manual of Operatzon for the Automatic Sequence

Controlled Calculator,

Cambridge, Massachusetts:

MIT Press 1985, p. x.

Dijkstra, E. 1982.

Selected Writings on Computing: A

Perspective,

New York: Springer.

Electronics,

1980. “Fifty Years of Achievement: A His-

tory,”

Electronics,

Special Commemorative Issue, 17

April, 1980.

Electronics,

1973. “The Great Takeover,” Special Issue

Electronics,

25 October.

Everitt, W. 1952. “Guest Editorial,”

IRE Proceedings

40.

899.

Fein, L. 1959. “The Role of the University in Com-

puters, Data Processing, and Related Fields,”

CACM

2 (Sept.). pp. 7-14;

Fein. L. 1961. ‘The Computer-related Sciences (Syn-

noetics) at a University in the Year

1975.” American

Scientist,

49/2 (June), pp. 149-168.

Gries, D., R. Miller, R. Richie, and P. Young 1986.

“Imbalance between Growth and Funding in Aca-

demic Computing Science; Two Trends Colliding,”

Comm.

ACM, 29 (September), pp. 870-876.

Hanson, D.

1982. The New Alchemists: Silicon Valley

and the Mzcroelectronics Revolution,

Boston: Little,

Brown.

Hodges, A.

1983., Alan Turing, the Emgma. New

York,

Simon & Schuster.

IRE 1953. “Computer Issue.”

IRE Proceedings 41 (Oc-

tober).

Kidder, T. 1981.

The Soul of a Neu: Machine,

Boston:

Little. Brown. 1981, pp. 245 ff.

Knuth. D E. 1968.

The Art of Computer Program-

mung; Vol. 1: Fundamental Algortthms,

Reading,

Massachusetts: Addison-Wesley.

Knuth, D. E. 1973.

The Art of

Computer

Program-

mung; Volume Three: Sorting and Searching,

Read-

ing, Massachusetts: Addison Wesley.

Knuth, D, E 1976. ‘*The State of the ‘Art of Computer

Programming,“’ Stanford University, Computer Sci-

ence Dept., Report STAN-CS-76-551, May.

Knuth. D. E. 1982. “A Conversation with Don Knuth,”

Annals of the HEstory of Computing. 4/3

(July), pp.

257-274.

Kuhn, T.

1970. The Structure of Sczentzfic Revolutions.

2nd Ed., University of Chicago Press.

Layton, E. T. 1971. ‘Mirror-Image Twins: The Com-

munities of Science and Technology in 19th Century

America,”

Technology and Culture,

12 ioctober), pp.

562-580.

274

Annals of

the History of

Computing, Volume

10,

Number

4, 1989

P. Ceruzzi *

Coevolution of Electronics and Computer Science

Leech, J. 1986. Letter to the author.

entific American,

Special Issue on “Automatic Con-

McMahon, A. Michal 1984.

The Making

a Profes-

trol,” 187, (September), pp. 116-130.

soon: A Century

Electrwal Engineering in Amer-

Sprague, R. E. 1972. “A Western View of Computer

ica, New York: IEEE Press.

History,”

Comm. ACM 15

(July), pp. 686-692.

Moore School of Electrical Engineering, 1947. “Theory

Sutherland, I. and C. Mead. 1977. “Microelectronics and

and Techniques for Design of Electronic Digital Computer Science.”

Screntific Amerwan, 237

~Sep-

Computers: Lectures Given at the Moore School, 8 tember), pp. 210-228.

July 1946-31 August, 1946.” Philadelphia: Moore Stibitz, G. 1945. “Relay Computers,” National Defense

School of Electrical Engineering. Univ. of Penn., Research Committee, Applied Mathematics Panel,

1947; reprinted 1985, MIT Press, Cambridge, Mas-

AMP Report 17l.lR. February 1945. pp. 21-22.

sachusetts. van Heijenoort, J. 1967.

From Frege to Gddel: A Source

Newell, A., A. Perlis, and H. Simon. 1967. Letter to

Book in Mathematical Logic, 1879-1931,

Cam-

the Editor of

Science 157 122

Sept.), pp. 1373-1374. bridge, Massachusetts: Harvard Univ. Press.

Newell, A. 1982. “Intellectual Issues in the History of Vincenti, W. 1979. “The Air-Propeller Tests of W. F.

Artificial Intelligence,” Carnegie-Mellon Univer- Durand and E. P. Lesley: A Case Study in Techno-

sity, Department of Computer Science, Report CMU- logical Methodology,”

Technology and Culture, 20

CS-82-142. (October), pp. 712-751.

Noble, D. F.

1984. Forces

Production: A Social His-

von Neumann, J. 1945. “First Draft of a Report on the

tory

Industrtal Automation

(New York: Knopf).

EDVAC,"

Philadelphia: Moore School of Electrical En-

Owens, L. 1986. “Vannevar Bush and the Differential gineering, 30 June.

Analyzer: The Text and Context of an Early Com- Wegener, P. 1970. “Three Computer Cultures: Com-

puter,”

Technology and Culture,

27 (January), pp. 63- puter Technology, Computer Mathematics, and

95. Computer Science.”

Advances In Computing. 10,

Pierce, J. R. 1977. “Electronics: Past, Present, and Fu- p. 9.

ture.”

Science, 195

(18 March), pp. 1092-1095. Weik,

1955.

A Survey

Domestic Electronic Dig-

Pollack, S. 1977. “The Development of Computer Sci-

ltal Computing Systems,

Aberdeen, Maryland: Bal-

ence, ” in S. Pollack (ed.),

Studtes tn Computer Set-

hstic Research Laboratory, Report No. 971.

ence,

Washington, D.C. 1982, pp. 1-51. Wiesner, J. 1958. Communication Sciences in a Uni-

Ramo, S. 1959. Address to 5th National Communica- versity Environment.

IBM J. Res. Dev. 2/4

(Octo-

tions Symposium. Utica, NY, 7 Oct. 1959. Quoted in her), pp. 268-275.

Computers and Automation,

9 (Jan. 19601, p. 6.

Wildes, K. L. and N. A. Lofg-ren 1985.

A Century

Redmond, K. C. and T. M. Smith 1980.

Project Whirl-

Electrical Engtneertng at MIT, 1882-1982.

Cam-

wind: The History

a Pioneering Computer.

Bed- bridge, Massachusetts: MIT Press, 1985.

ford. Massachusetts: Digital Press.

Wilkes. M. V. 1985.

Memoirs

a Computer Pioneer,

Ridenour, L. 1952. “The Role of the Computer,”

Sci-

Cambridge, Massachusetts: MIT Press.

Annals of the History of Computing. Volume 10, Number 4, 1989

275

Discussion

Paul Ceruzzi is a historian specializing in the history of technology and computing. He's best known for exploring how technological developments and societal factors influence each other. As curator emeritus at the Smithsonian National Air and Space Museum, Ceruzzi researched, curated exhibits, and lectured on aerospace electronics and computing, highlighting how these fields transformed air and space flight. ![](https://i.imgur.com/aShOfZY.jpeg) *Paul Ceruzzi* Relays are simple electromechanical switches that control electrical circuits. They work by using an electromagnet: when electric current flows through a coil, it creates a magnetic field that mechanically opens or closes a separate circuit. In early computing machines, relays served as digital switches to represent and manipulate numbers, effectively acting like the binary logic elements (1s and 0s) used in modern computers. The relay technology was initially developed for the telephone industry, where it was used to automate switching and routing of telephone calls, connecting lines as needed. ![](https://i.imgur.com/IhHmupA.gif) *Demonstration of the relay mechanism* Howard Aiken was an American physicist and pioneer in computing who designed and built the Automatic Sequence Controlled Calculator (ASCC), also known as the Harvard Mark I, completed in 1944. This was one of the earliest automatic computers, capable of performing long sequences of mathematical operations without human intervention. Like other early computers, the Mark I used electromechanical components such as relays, switches, and rotating mechanical wheels. While its speed - approximately three to five arithmetic operations per second - was limited compared to later electronic computers, the Mark I was extremely reliable and demonstrated the feasibility of automatic, programmable computation. ![](https://i.imgur.com/BgPWe5P.png) *Harvard Mark I* IBM's launch of the System/360 in 1964 significantly reinforced the idea of dividing computer history into distinct technological "generations." This emphasis served as a marketing strategy, clearly positioning the new integrated-circuit-based System/360 as more advanced and efficient than earlier vacuum-tube or transistor-based systems. Drum-type computers were early stored-program computers that used a rotating metal drum as their main memory. Data and instructions were stored magnetically on the surface of the drum. As the drum spun, read/write heads accessed the information sequentially. Although relatively inexpensive and reliable, these drum-based machines were slower compared to larger systems like UNIVAC or IBM 701, which used faster but more costly memory technologies. Because of their affordability, drum-type computers, such as the popular IBM 650, were installed widely and helped make digital computing accessible to a larger number of users in the 1950s. ![](https://i.imgur.com/PHzz6Qq.jpeg) *Drum Type Memory * Ceruzzi is referring to Charles Babbage's designs for his "Analytical Engine”, which he conceived in the 1830s. Focused on the immense design challenges and the struggle to secure funding, Babbage did not formally publish his work, which was first widely described in an 1842 paper by Luigi Menabrea based on Babbage's lectures. The Analytical Engine was a revolutionary concept that anticipated the core components of modern computers. Babbage's design featured an arithmetic processing unit (which he called the "mill"), a separate memory (the "store"), and a system for input using punched cards. ![](https://i.imgur.com/jY6kQpy.jpeg) *A representation of the Analytical Engine* The ARC computer was an early British computing machine developed by Andrew D. Booth at Birkbeck College, London, in the late 1940s. "ARC" stood for Automatic Relay Calculator, reflecting its original electromechanical relay-based design. Booth initially built the machine using relays to save cost and time, but quickly transitioned it to faster vacuum tubes. Influenced by John von Neumann’s stored-program principle, the ARC was among Britain's earliest stored-program computers. Booth used it primarily for research in crystallography, performing complex mathematical calculations needed to analyze molecular structures. The reason vacuum tubes provided such a dramatic increase in speed compared to relays is that vacuum tubes switch electronically, without physical moving parts. Relays, which rely on mechanical movements, were limited to just a few arithmetic operations per second. Vacuum tubes, by contrast, could perform thousands of operations per second - roughly a thousand times faster. Vacuum tubes are devices that control the flow of electric current within a sealed, airless glass or metal container. Inside, electrons are emitted from a heated cathode and flow toward a positively charged anode. The introduction of a metal grid between the cathode and anode (forming devices such as triodes) allows precise control of this electron flow: changing the grid's voltage controls whether electrons pass through, effectively switching or amplifying electrical signals. Vacuum tubes, first invented by John Ambrose Fleming in 1904, were widely used before WWII in radio, telephone, and radar technologies. Their ability to rapidly switch electronic signals made them ideal components in early digital computers like the ENIAC (1945), enabling dramatically faster calculations than previous electromechanical technologies. ![](https://i.imgur.com/i94eN0j.jpeg) *Vaccum Tube* Gottlob Frege, David Hilbert, and Kurt Gödel were mathematicians whose work significantly influenced the theoretical foundations of computer science by advancing our understanding of formal logic and mathematical reasoning. In the late 19th century, Frege developed a precise logical notation to rigorously represent mathematical statements, laying the groundwork for symbolic logic. In the early 20th century, Hilbert proposed the idea of formalizing all mathematics into a logical system with clearly defined rules, known as Hilbert’s program. Later, in the 1930s, Gödel demonstrated, through his incompleteness theorem, that in any sufficiently powerful mathematical system, some statements can neither be proved nor disproved within that system. Although their work was primarily theoretical, these foundational insights became crucial to the later development of computer science, especially in fields such as computability, algorithm theory, and programming language design. The stored-program principle is the idea that a computer's instructions and data should both be stored together inside the computer’s memory. Before this, computers were "programmed" externally, often using switches, plugs, or separate punched paper tapes, making changes slow and cumbersome. By storing instructions internally, a computer could switch tasks rapidly and flexibly, allowing it to become much more versatile. While John von Neumann popularized this idea in his influential 1945 report, known as the "First Draft of a Report on the EDVAC," the exact origins of the stored-program concept remain debated. Some historians credit pioneers such as J. Presper Eckert, John Mauchly, and even Alan Turing with independently developing or significantly contributing to the same idea, leading to ongoing discussions about who deserves primary recognition. A mercury delay line was originally developed during World War II for radar systems. Radar detects objects by sending out a radio pulse and measuring the time it takes for the pulse to return after reflecting off an object. To accurately determine this time, radar equipment needed to briefly store (or delay) the transmitted pulse. Mercury delay lines did exactly this, temporarily holding the original radar pulse long enough for it to be directly compared with the reflected echo. By precisely measuring the time delay between these two signals, the radar system could calculate how far away the detected object was. After the war, mercury delay lines were adapted as memory devices in early digital computers, such as EDSAC and UNIVAC I. In these computers, digital data, including instructions and intermediate calculation results, were temporarily stored as pulses traveling slowly through tubes of mercury. The data pulses were continuously refreshed, enabling the computer to reliably store and retrieve information before more advanced electronic memory became available. ![](https://i.imgur.com/oW0VTN8.jpeg) *How a mercury delay line works* In 1883, while working to improve the incandescent light bulb, Thomas Edison observed that electricity could flow between a heated filament and a metal plate placed inside a vacuum-sealed bulb, even though they weren’t physically connected. Edison noticed this effect but didn't fully understand its significance at the time. He did, however, patent the discovery, calling it the Edison Effect. Although Edison initially considered it just a curiosity, it became a key insight, leading to the development of vacuum tubes and laying the foundation for the modern discipline of electronics.

Comments

Products

Project