9

The Social Dániel Bell

Framework of
the Information
Society

The endless cycle of idea and action,

Endless invention, endtess experiment,
Brings knowledge of motion, bút nőt of stíllness... ■
Where is the Life we have lost in living?
W h e re is th e w is d o m w e ha ve lo s t in k n o w le d g e ?
Where is the knowledge we have lost in information?
T. S. Eliot; Choruses from “ The Rock'1

INFORMATION AND TELECOMMUNICATIONS IN THE POSTINDUSTRIAL

SOCIETY

In the Corning century, the emergence of a new social framework based on

telecommunications may be decisive fór the way in whích economic and
social exchanges are conducted, the way knowledge is created and re-
trieved, and the character of the occupations and work in which mén
engage.This revolution in the organizationand Processing of information
and knowledge, in which the computer plays a Central role, has as its
context the development of what I have called the postindustrial society.1
Three dimensions of the postindustrial society are relevant to the discus-
sion of telecommunications:
a. The char.ge from a goods-producing to a service society
b. The centrality of the codification of theoretical knowledge fór innova-
tion in technology
c. The creation of a new “ intellectual technology" as a key tool of Systems
analysis and decision theory
The change from a goods-producing to a service society can be indi-
cated briefly. In the United States in 1970, sixty-five out of every 100

Dániel Bell is professor o f sociology at Harvard University. He is widely known fór

his víews on the postindustrial society—a term that he coined. In his numerous
books and articles he has examined the interplay between technology and society
and the way in which the infrastructures of transportatíon. energy, and com m unica-
tion tie societies together.
Dániel Bell 164

persons in the labor force were engaged in Services, about 30 percent in

the production of goods and construction and under 5 percent in agricul-
ture. The word Services of course covers a large multitude of activities. In
preindustrial societies a sizable proportion of the labor force is engaged
in household or domestic service. (In England until the 1870s the single
largest occupational eláss was servants.) In an industrial society Services
are auxiliary to the production of goods, such as transportation (rail and
truck), Utilities (power and light), banking, and factoring. Postindustrial
Services are of a different kind. They are humán Services and professional
Services. The humán Services are teaching, health, and the large array of
social Services; professional Services are those of Systems analysis and
design and the programming and Processing of information. In the last
two decades, the net new growth in employment has been entirely in the
area of postindustrial Services, and while the rate ol growth has slowed
(particularly because of the financial costs of education and the cutbacks
in social Services in urban communities), the generál trend continues.
The axial principle of the postindustrial society, however, is the cen-
trality of theoretical knowledge and its new role, when codified, as the
director of social change. Every society has functioned on the basis
of knowledge, bút only in the last half century have we seen a fusion of
Science and engineering that has begun to transform the character of
technology itself. As Cyril Stanley Smith, the distinguished metallurgist,
has observed, "In only a small part of history has industry been helped by
Science. The development of a suitable science began when chemists pút
intő rational order facts that had been discovered íong before by people
who enjoyed empirical diverse experiment.” 2
The industries that still dominate society—Steel, autó, electricity, tele­
phoné, aviation—are all "nineteenth-century" industries (though steel
began in the eighteenth century with the coking process of Abraham
Darby, and aviation in the twentieth with the Wright Brothers) in that they
werecreated by "talented tinkerers" who worked independently of orwere
ignorant of contemporary science. Alexander Graham Bell, who invented
the telephoné about one hundred years ago (though the actual fact is in
somé dispute), was an elocution teacher who was looking fór somé means
to amplify sound in order to help the deaf. Bessemer, who created the
open-hearth process (to win a prize offered by Napóleon III fór a better
means of casting cannon) did notknow the scientific work of Henry Clifton
Sorby on metallurgical processes. And Thomas Alva Edison, who was
probably the most prolific and talented of these tinkerers (he invented,
among other things, the electric light bulb, the phonograph, and the
motion picture), was a mathematical ílliterate who knew little and cared
less about the theoretical equations of Clerk-Maxwell on electromagnetic
properties.
The Inform ation Society 165

Nineteenth-century inventing was trial-and-error empiricism, often

guided by brilliant intuitions. Bút the natúré of advanced technology is its
intimate relation with Science, where the primary interest is nőt in the
product itself bút in the diverse properties of materials together with the
underlying principles of order that allow fór combination, substitution, or
transmutation. According to Cyril Smith, "All materials came to be seen in
competition, with the emphasis only on the properties that were needed.
Thereafter every new development in advanced technology—radar, nu-
clear reactors, jet aircraft, computers and satellite Communications to
name a few— has served to break the earlier close association of materials
research with a single type of manufacture, and the modern materials
engineer has emerged.”
The natúré of this change, in technology and in science, has been to
enlarge the “ field of relation" and the rangé of theory so as to permit a
systematic synergism in the discovery and extension of new products and
theories. A Science, at bottom, is a set of axioms linked topologically to
form a unified scheme. Bút as Bronowski has observed, “ A new theory
changes the system of axioms and sets up new connections at the joints
which changes the topology. And when two Sciences are linked to form
one (electricity and magnetism, fór instance, or evolution with genetics),
the new network is richer in its articulation than the sum of its two parts."3
While modern science, like almost all humán activities, has moved to-
ward a greater degree of specialization in its pursuit of more detailed
knowledge, the more important and crucial outcome of its association
with technology is the integration of diverse fields or observations intő
single conceptual and theoretical frameworks offering much greater ex-
planatory power. Norbert Wiener, in his autobiographical / Am a
Mathematician, points out that his first mathematical papers were on
Brownian motion and that at the same time electrical engineering work
was being done on the so-called shot effects, or the movement of electric
current through a wire. The two topics were unrelated; yet twenty years
later the situation had changed dramatically.
In 1920 very little electrical apparátus was loaded to the point at which the
shot effect became critical. However the later development—first of
broadcasting and then of radar and television—brought the shot effect to
the point where it became the immediate concern of every Communica­
tions engineer. The shot effect was nőt only similar in origin to the Brown­
ian movement, fór it was a result of the discreteness of the universe, bút
had essentially the same mathematical theory. Thus, my work on the
Brownian motion became somé twenty years later a vitai tool fór the
electrical engineer.4
Wiener's theory of cybernetics joins a variety of fields in the common
framework of statistical Information theory. "The development of ideas on
the structure of synthetic polymers,” Cyril Smith writes, "eventually came
Dániel Bell 165

to bridge the gap between the nineteenth century chemists molecule and
the early twentieth-century crystal, so paving the way fór the unified struc-
tural view of all materials which we see taking shape today."5 The de-
velopment of solid-state physics, which is the foundation of the electronic
revolution, árosé out of the work of metallurgists and physicists on the
structure of conductor devices.
The methodological promise of the second half of the twentieth century
is the management of organized complexity: the complexity of theories
with a large number of variables and the complexity of large organizations
and Systems which invoíve the coordination of hundreds of thousands
and even miilions of persons. Since 1940 there has been a remarkable
efflorescence of new fields and methods whose concern is with the prob-
lems of organized complexity: information theory, cybernetics, decision
theory, game theory, utility theory, stochastic processes. From these have
come specific techniques such as linear programming, statistical decision
theory, Markov chain applications, Monté Carlo randomizing, and mini-
max strategies, which allow fór sampling from large numbers, alterna-

Tabte 9.1
The Postindustrial Society: A Comparative Schema

Preindustrial
Mode of Production Extractive

Economic sector Primary

A griculture
Mining
Fishing
Tim ber
Oil and gas

Transform ing resource Natural power

W ind, water, draft animal, humán muscle

Strategic resource Raw materials

Technology Craft
Skill base Artisan, manual worker, farmer

M ethodology Commonsense, trial and error;

experience
Time perspective O rientation to the pást

Design Game against natúré

Axial principle Traditionalism
The Inform ation Society 167

tive optimál outcomes of different choices, or definitions of rational action

under conditions of uncertainty.
Since technology is the instrumental mode of rational action, I have
called this new development "intellectual technology," fór these methods
seek to substitute an algorithm (i.e., decision rules) fór intuitíve judg-
ments. These algorithms may be embodied in an automatíc machine or a
computer program, or a set of instructions based on somé statistical or
mathematical formula, and represent a "form alization" of judgments and
their routine application to many varied situations. To the extent that
intellectual technology is becoming predominant in the management of
organizations and enterprises, one can say that it is as Central a feature of
postindustrial society as machine technology is in industrial society,

A Knowledge Theory of Value

If one compares the formai properties of postindustrial society with those
of industrial and preindustrial society (see table 9.1), the crucial variables
of the postindustrial society are information and knowledge.

Industrial Postindustrial
Fabrication Processing; Recycling

Secondary Services
Goods-producing
Tertiary Quarternary
M anufacturing
Transportation Trade
Du rables
Utilities Fináncé
Nondurables
Insurance
Heavy construction Quinary Reál estate
Health, education
Research, government,
Recreation

Created energy Information

E lectricity—oií, gas. coal, nuclear Com puter and data-transm ission Systems
power

Financial Capital Knowledge

Machine technology Intellectual technology
Engineer, sem iskilled worker Scientist, technical and professional
occupations
Empíricism, experimentation Abstract theory, models, sim ulations,
decision theory, systems analysis

Ad hoc adaptiveness, experimentation Future orientation: forecasting and

planning

Game against fabricated future Game between persons

Econom ic growth C odification of theoretical knowledge
Dániel Bell 168

By information I mean data Processing in the broadest sense; the stor-

age. retrieval. and Processing of data becomes the essential resource tor
all economic and social exchanges. These include:
a. Data Processing of records: payrolls, government benefits (e.g., Social
Security), bank clearances, credit clearances, and the like
b. Data Processing fór scheduling: airline reservations, production
scheduling. inventory analysts, product-mix information, and the like
c. Data bases: characteristics of populations as shown by census data,
markét research, opinion surveys, election data, and the like
By knowledge, I mean an organized set of statements of fact or ideas,
presenting a reasoned judgment or an experimental result, which is
transmitted to others through somé communication médium in somé
systematic form. Thus, I distinguish knowledge from news or entertain-
ment. Knowledge consists of new judgments (textbook, teaching, and
library and archive materials).
In the "production of knowledge,” what is produced is an intellectual
property, attached to a name or a group of names and certified by
copyright or somé other form of social recognition (like publication). This
knowledge is paid fór—in the time spent in writing and research, in the
monetary compensation by the Communications and educational média.
The response of the markét, along with administrative and political deci-
sions of superiors or peers, judge the worth of the result and any further
claim on social resources that might be made in its behalf. In this sense,
knowledge is part of social overhead. More than that, when knowledge
becomes involved in somé systematic form in the applied transformation
of resources (through invention or social design), then one can say that
knowledge, nőt labor, is the source of value.
Economists, in their formai schemes to explain production and ex-
change, use as key variables "land, labor and Capital,” though institution-
ally minded economists such as Werner Sombart and Joseph Schumpeler
added the notion of an acquisitive spirit or entrepreneurial initiative. The
analytical mode used by economists, the “ production function," sets forth
the economic mix only as Capital and labor—a system that lends itself
easily to a labor theory of value, with surplus labor value as congealed
Capital, bút neglects almost entirely the role of knowledge or of organiza-
tional innovation and management. Yet with the shortening of labor time
and the diminution of the production worker (who in Marxist theory is the
source of value, since most Services are classified as nonproductive
labor), it becomes clear that knowledge and its applications replace labor
as the source of "added value" in the national product. In that sense, just
as Capital and labor have been the Central variables of industrial society,
so information and knowledge are the crucial variables of postindustrial
society.
The Information Society 169

INTELLECTUAL FOUNDATIONS OF THE REVOLUTION IN

COMMUNICATIONS

Fór Goethe, the basis of the humán community was communication.

Decades before other persons spoke of such projects, he envisaged a
Panama Canal, a Suez Canal, and a canal between the Rhine and the
Danube as the means by which the humán community might become more
closely intertwined. Bút it was the Canadian economic histórián Harold
Innis, more than any other person, who saw changes in the modes of
communication, rather than production and property relations, as the key
to transitions from one stage of society to another.
Western civilization has been profoundly influenced by communication
. . . (and can bel divided intő the following periods in relation to média of
communication: clay, the Stylus and cuneiíorm script from the beginnings
of civilization in Mesopotamia: papyrus, the brush and hieroglyphics and
hieratic to the Graeco-Roman period, and the reed pen and the alphabet to
the retreat of the Empire from the west; parchment and pen to the tenth
century of the dark ages; and overlapping with paper, the latter becoming
more important with the invention of printing; paper and the brush in
China, and paper and the pen in Europe before the invention of printing or
the Renaissance; paper and the printing press under handicraft methods
to the beginning of the nineteenth century, or from the Reformation to the
French Revolution; paper produced by machinery and the application of
power to the printing press since the beginning of the nineteenth century
to paper manufactured from wood in the second half of the century;
celluloid in the growth of the cinema; and finally the rádió in the second
quarter of the present century. In each period I have attempted to trace the
implications of the média of communication fór the character of knowl-
edge and to suggest that a monopoly or an oligopoly of knowledge is built
up to the poínt that equilibrium is disturbed.6
Innis was a technological determinist. He thought that the technology
of communication was basic to all other technology, fór if tool technology
was an extension of man’s physical powers, communication technology,
as the extension of perception and knowledge, was the enlargement of
consciousness. He argued nőt only that each stage of Western civilization
was dominated by a particular médium of communication bút that the rise
of a new mode was invariably foilowed by cultural disturbances.7
One can say that the new média of communication today are television
or the computer, or the variant modes of storage, retríeval, and transmis-
sion that wiil arise through the "fusing" of technologies. Bút the core of
the present Communications revolution is nőt a specifíc technology bút
the set of concepts represented by the term Information íheory.

The Statistics of Language

Information theory árosé from the work of Claude Shannon on switching

circuits to increase “ channel capacity," the design fór which he derived
Dániel Bell 170

from the algebra of logic. The algebra of logic is an algebra of choice and
deals with the rangé of choices in a determinate sequence of alternative
possibílities in the routing of a message. The parlor game of "Twenty
Questions” is often taken as a conventional illustration of how one nar-
rows a rangé of possibílities by asking a series of yes or no questions. As
Shannon points out in the article on Information theory that he wrote fór
the Encyclopaedia Britannica, “ The writing of English sentences can be
thought of as a process of choice: choosing a first word from possible first
words with various probabilities; then a second with probabilities depend-
ing on the first: etc. This kind of statistical process is called a stochastic
process, and information sources are thought of, in information theory, as
stochastic processes."
The information rate of written English can be translated intő bits (£?/-
nary digifs 1 and 0), so that if each letter occurred with equal frequency,
there would be 4.76 bits per letter. Bút since the frequencies are unequal
(E is common, Z. Q, and X are nőt), the actual rate is one bit per letter.
Technically, English is said to be 80 percent "redundant," a fact that one
can immediately ascertain by "deciphering" a sentence from which vari­
ous vowels or consonants have been deleted. By knowing the statistical
structure of a language, one can dérivé a generál formula that determines
the rate at which information can be produced statistícally and create
huge savings in transmission time. Bút though transmission was the im-
petus to the formulatíon of information theory, the heart of the concept is
the idea of coding. Messages have to go through “ channels” ; inevitably,
they are distorted by "nőise" and other forms of “ resistance" that arise
from the physical properties of the channel. What Shannon found was that
it is possible to encode a message that can be accurately transmitted even
if the channel of communication is faulty, so long as there is enough
capacity ín that channel.
Shannon's mathematical theory had immediate application to industry.
The theoretical and statistical underpinnings seemed to confirm the more
generál theory of Wiener's Cybemetics, a work that had been commis-
sioned by an obscure publisher in Francé after the war and became an
immediate best-seller on its publication by Wiley in 1948. What Shannon's
and Wiener’s work seemed to promise was the move toward somé generál
unified theory of physics and humán behavior (at least in physíology, psy-
chology, and linguistics) through the concept of information. As Shannon
himself wrote in his Britannica essay,
A basic idea in communication theory is that information can be treated
very much like a physical quantity such as mass or energy. . . .
The formula fór the amount of information is identical in form with
equations representing entropy in statistical mechanics, and suggeststhat
there may be deep-lying connections between thermodynamics and in-
The Information Society 171

formation theory. Somé scientists believe that a proper statement ot the

second law of thermodynamics reqtiires a term relating to information.
These connections with physics, however, do nőt have to be considered in
the engineering and other applications of information theory.8
Bút this is a confusion of realms—compounded by the facile use of the
word entropy to equate the degree of disorder or nőise (i.e., the loss of
accuracy) in communication with the loss of heat orenergy in transforma-
tional activities in physics. As Wiener pút it in his Cybernetics, resisting the
easy comparisons of living with mechanical organisms, "Information is
information, nőt matter or energy. No materialism which does nőt admit
this can survive at the present day."9
However true it may be as a statistical concept that information is a
quantity, in its broadest sense—to distinguish between information and
fabrication—information is a pattern or design that rearranges data fór
instrumental purposes, while knowledge is the set of reasoned judgments
that evaluates the adequacy of the pattern fór the purposes fór which the
information is designed. Information is thus pattern recognition.subject to
reorganization by the knower, in accordance with specified purposes.
What is common to this and to all intellectual enterprises is the concept of
relevant structure. This concept is what underlies the shift, in the works of
Cyril Stanley Smith, from “ matter to materials," from theclassificatory and
even combinational arrangements of elementary properties of matter that
began with the pre-Socratics to our present-day understanding of the
structural relations of the properties of materials.
These structural relations—in science, as in the economy—fali intő two
separate domains. The first is the transformation of matter and energy,
from one matéria! form intő another. The second is the transformation of
information from one pattern intő another. As Anthony Oettinger puts it in
an aphorism, “ Without matter there is nothing; without energy matter is
inért; and without information, matter and energy are disorganized, hence
useless."

The Use of Models

Technological revolutions, even if intellectual in their foundations, be-

come symbolized if nőt embodied in somé tangible "thing," and in the
postindustrial society that “ thing" is the computer. If, as Paul Valery said,
electricity was the agent that transformed the second half of the
nineteenth century, in a similar vein the computer has been the "analytical
engine" that has transformed the second half of the twentieth century.
What electricity did—as the source of light, power, and communication—
was to create the "mass society"; that is, to extend the rangé of social ties
and the interaction between persons and so magnify what Durkheim called
Dániel Bell 172

the social density of society. In that respect, the computer is a tool fór
managing the mass society, since it is the mechanism that orders and
processes the transactions whose huge number has been mounting al­
most exponentially because of the increase in social interactions.
The major sociopolitical question facing the mass society is whether we
can manage the economy effectively enough to achieve our social goals.
The development of computere has allowed us to construct detailed mod-
els of the economy. Wassily Leontieff recently described the extraordinary
expansion of the input-output system:
The first input-output tables describing the flow of goods and Services
between the different sectors of the American economy in census years
1919 through 1929 were published in 1936. They were based on a rather
gross segregation of all economic activities in 44 sectors. Because of the
lack of computing facilities, these had to be further grouped intő only 10
sectors, fór the purposes of actual analytic calculations.
The data base, the computing facilities, and the analytical techniques have
advanced much further than could have been anticipated forty years ago.
National input-output tables containing up to 700 distinct sectors are
being compiled on a current basis, as are tables fór individual, régiónál,
State and metropolitan areas. Priváté enterprise has now entered the
input-output business. Fór a fee one can now purchase a single row of a
table showing the deliveries of a particular product, say, coated laminated
fabrics or farming machine tools, nőt only to different Industries bút to
individual plants within each industry segregated by zip code areas.10
Though it is clearthat economists are able to model the economy and do
computer simulations of alternative policies to test their consequences, it
is much less clear whether such models allow us to manage the economy.
The critical point is that the crucial decisions fór any society are the
political ones, and these are nőt derivative from economic factors.
Can one model a society? One immediate problem is that we do nőt
have any persuasive theories of how a society hangs together, though
paradoxically, because of our understanding of technology, we have a
better idea of how societies change. One can only model a closed or finíte
system; the econometric models operate within a closed system. Yet
society is increasingly open and indeterminate, and as mén become more
conscious of goals there is greater debate about decisions. Decisions on
social policy become more and more a matter within the purview of the
political System rather than of aggregate markét decisions, and this, too,
weakens our ability to model a society.
Beyond this there may be reasons intrinsic to the structure of "large
numbers” that could prevent the computer from becoming the instrument
fór the modeling and prediction of any complex system. John von
Neumann, one of the pioneers in the development of the theory of elec-
tronic computing, thought that the prediction of weather would be possi-
The Information Society 173

ble once computers became sophisticated enough to handle all the

numerous interacting variables in the atmosphere. Yet as Tjalling Koop-
mans and others have pointed out, beyond a certain threshotd introducing
added complexity results in answers that are less and less reliable. Thus,
the effort to optimize an objective by seeking fór complete Information
may be self-defeating. The social world is nőt a Laplacean universe where
one can plot, from the initial values, the determinate rates of change of
other phenomena. If so many parts of the physical world now require us to
deal wíth a calculus of possibilíty ratherthan determined regularities, this
is even more true in a social world where mén are less and less willing
passively to accept existing arrangements bút instead work actively to
remake them. By letting us know the risks and probabilities, the computer
has become a powerful tool tor exploring the permutations and combina-
tions of different choices and fór calculating their consequences. the odds
of success or failure. The computer does this by using a binary code that
with high speed can answer a question with a yes or a no. What it cannot
do, obviously. is to decide like a roulette wheel whether to stop on the yes
or on the no.

The Economics o f Information

Information is Central to all economic transactions—indeed, perfect in-

formation isthe indispensable condition fór perfect competition in generál
equilibrium theory. Yet we have no economic theory of information, and
the character of information, as distinct from the character of goods,
poses somé növel problems fór economic theorists.
In a price and markét economy, the condition fór efficiency, or optimál
use of resources, is complete information among buyers and sellers, so
that one can obtain the “ best” price fór one's goods or Services. Bút with
the widening of markets and the reduction of distances by transportation
and communication—which a!so enlarges the sphere of competition—
efficiency increasingly demands nőt only a knowledge of contemporary
alternatives bút of the likely future ones as well, since political decisions or
new technologies may radically altér prices. A political embargó may cut
off the supplies of a resource, A tax cut or a tax rise will affect the level of
spending. New technologies may sharply cut the price of a product (wit-
ness theextraordinary changes in two years in the price of small electronic
calculators), leaving firms with large inventories or committed to older
production techniques at a great disadvantage.
Information, as Kenneth Arrow puts it, reduces uncertainty." The
random-walk theory that one cannot "beat the stock markét" is based on
the assumption that stock prices reflect new information about companies
so quickly that investors have little chance to earn better-than-average
Daniéi Bell 174

returns on their money. Therefore the wíser strategy is to piacé one's

money in an index fund that reflects the average prices of the markét as a
whole. The job search in the labor markét is enhanced by access to a wider
pool of information. Accurate crop reporting Controls the vagaries of the
futures markét in commodities. One can multiply the illustrations indefi-
nitely.
Bút information is nőt a commodity, at least nőt in the way the term is
used in neoclassical economics or understood in industrial society. Indus-
trial commodities are produced in discrete, identifiable units, exchanged and
sold, consumed and used up, like a loaf of bread or an automobilé. One
buys the product from a seller and takes physical possession of it; the
exchange is governed by legal rules of contract. In the manufacture of
industrial goods, one can set up a “ production function" (i.e., the relatíve
proportions of Capital and labor to be employed) and determine the ap-
propriate mix relatíve to the costs of each factor.
Information, or knowledge, even when it is sold, remains with the pro­
ducer. It is a “ collective good” in that once it has been created, it is by its
natúré available to all.12 In fact, the character of Science itself, as a
cooperative venture of knowledge, depends on the open and complete
transmission of all new experiments and discoveries to others in the field.
Multiple discoveries of the same theory űr experimental result or tech-
nique, which Róbert Merton argues is a more dominant pattern in science
than the image of the lonely genius or schotar, are one result of this
openness and the rapid spread of knowledge.13
If knowledge is a collective good there is little incentive fór any indi-
vidual enterprise to pay fór the search fór such knowledge, unless it can
obtain a proprietary advantage, such as a patent or a copyright. Bút
increasingly, patents no longer guarantee exclusiveness, and many firms
lose out in spending money on research only to find that a competitor
(particularly one overseas) can quickly modify the product and circumvent
the patent; similarly, the question of copyright becomes increasingly dif-
ficult to police when individuals or libraries can Xerox whatever pages
they need from technical journals or books or when individuals and
schools can tape music off the air or record a television performance on
videó discs. Bút more generally, the results of investing in information (i.e.,
doing research), are themselves uncertain. Because firms are averse to
risk, they tend to undervalue such investments from the social point of
view, and this leads to underinvestment in priváté research and develop-
ment.
If there is less and less incentive fór individual persons or priváté enter-
prises to produce knowledge without particular gain, then the need and
effort falls increasingly on somé social unit, be it university orgovernment,
to underwrite the costs. And since there is no ready markét test (how does
The Information Society 175

one estimate the value of basic research?), it is a challenge fór economic

theory to design a socially optimál policy of investment in knowledge
(íncluding how much money should be spent fór basic research; what
allocations should be made fór education, and fór what fields; in what
areas of health do we obtain the “ better returns"; and so on) and to
determine how to “ price” information and knowledge to users.14

The Merging o f Technologies

Through the níneteenth and up to the mídtwentieth century, communica-

tion could be divided roughly intő two distinct realms. One was mail,
newspapers, magazines, and books, printed on paper and delivered by
physical transport or stored in libraries. The other was telegraph, tele­
phoné, rádió, and television, coded message image or voice sentby rádió
signals or through cables írom person to person. Technology, which once
made fór separate Industries, is now erasing these distinctions, so that a
variety of new alternatives are now available to information users, posing,
fó r that very reason, a major set of policy decisions fó r the lawmakers of
the country.
Inevitably, Earge vested interests are involved. Just as the substitution of
oil fór coal and energy and the competition of truck, pipeline, and railroad
in transportation created vast economic dislocations in corporate power,
occupational structures, trade unions, geographical concentrations, and
the like, so the huge changes taking piacé in Communications technology
will affect the major industries that are involved in the Communications
aréna.
Broadly, there are five major problem areas:
1. The meshing of the telephoné and computer Systems, of telecommuni-
cations and teleprocessing, intő a single mode. A corollary problem is
whether transmission w ill be primarily over telephone-controlled wires or
whether there will be independent data-transmission Systems. Equally,
there is the question of the relatíve use of microwave relay, satellite
transmission, and coaxial cables as transmission Systems.
2. The substitution of electronic média fór paper Processing. This includes
electronic banking to eliminate the use of checks; the electronic delivery
of mail; the delivery of newspapers or magazines by facsimile rather than
by physical transport; and the long-distance copying of documents.
3. .The expansion of television through cable systems, to allow fór multiple
channels and specialized Services, and the linkage to home terminals fór
direct response to the consumer or home frűm local or Central stations. A
corollary is the substitution of telecommunication fór transportation
through videophone, closed-circuit television, and the like.
4. The reorganization of information storage and retrieval systems based
Danié/ Bell 176

on the computer to allow fór interactive network communication in team

research and direct retrieval from data banks to library or home terminals.
5. The expansion of the education System through computer-aíded in-
struction, the use of satellite Communications systems in rural areas,
especially in the underdeveloped countries, and the use of videó discs
both fór entertaínment and ínstruction in the home.15
Technologically, telecommunications and teleprocessing are merging
in a mode that Anthony Oettinger has called "compunications” (see fig.
9.1). As computers come increasingly to be used as switching devices in
Communications networks and electronic Communications facilities be-
come intrínsic elements in computer data-processing Services, the distinc-
tion between Processing and communication becomes indistinguishable.
The major questions are legal and economic. Should the industry be
regufated or competitive? Should it be dominated, in effect,byA T& Torby
IBM?*
The entry of specialized carriers intő the business field, undercutting
AT&T prices, threatens its consumer rate structure as well, and would cre-
ate large political upheavals. Yet the '‘computer" proponents have argued
that technological innovation in the telephoné field has been stodgy,
whereas the energetic and bustling computer field has demonstrated its
ability to innovate rapidly and reduce costs and prices, sothat competition
in transmission, in the end, would serve the country as a whole.
The questions I have been raising about the tusion of Communications
technologies—the ríse of compunications—are nőt only technological and
economic bút, most important, political. Information is power. Control
over Communications Services is a source of power. Access to communi­
cation is a condition of freedom. There are legal questions that dérivé
directly from this. The electronic média, such as television, are regulated,
with explicit rules about "fairness" in the presentation of views. access to
reply to editorials, and the like. Bút ultimately the power is governmental.
Decisions about the station's future lie with the Federal Communications
Cornmission. The telephoné industry is regulated on its rates and condi-
tions of service. The computer industry is unregulated and operates in an
open markét. The print média are unregulated, and their rights on free
speech are zealously guarded by the First Amendment and the courts.

* In 1976, AT&T introduced a bili in Congress to allow it to buy out its microwave
competitors. and it wants Congress to require anyone plugging specialized Services
intő its lines to buy a connecling device from the phone company. IBM has entered
intő a direct challenge to AT&T by setting up The Satellite Business Systems
pany jointly with Aetna Insurance and Comsat General to operate a satellite Com­
munications service that would transmit the full rangé of ''compunications" by
1981.
The Inform ation Society 177

Computer Satellite
Terminál (WeatherTIntel! igencefetc.)

Teletypesetter Telemetry Telecontrol

9.1
The Changing Telecom m unications Network. As of 1974, the 144 m illión piain old
ielephones still predom inated, bút many other devices are now attached to a
network that has become an infrastructure basic to most social functions, including
many that reach directly intő the home. As com puters and com puter term ínals have
become increasingly pervasive over the last two decades, the network has
developed toward an integrated com puter Communications or ,,com p un icatio ns,,
network. Front Paul J. Berman and Anthony G. O ettinger, The M édium and the
Telephoné: The P olitics o f Inform ation Resources, W orking Paper 75-8. December
15. 1975, Harvard Program on Inform ation Technologies and Public Policy.
Cambridge, Mass.

Libraries have largely been priváté or locally controlled; now great data
banks are being assembled by government agencies and by priváté corpo-
rations. Are they to be under government supervision or unregulated? All
of these are major questions fór the future of the free society and bear on
the problem of a national information policy.

THE QUANTITATIVE DIMENSIONS OF THE INFORMATION SOCIETY

In 1940, Colin Clark, the Australian economist, wrote his path-breaking

Conditions of Economic Progress, in which he divided economic activity
inlo three sectors, primary (principally extractive), secondary (primarily
manufacturing), and tertiary (Services). Any economy is a mixture of al)
three sectors, bút their relatíve weights are a function of the degree of
productivity (output per capita) in each sector. Economic progress is
defined as the rate of transfer of labor from one sector to another, as a
function of differential productivity. As national incomes rise, the expan-
sion of the manufacturing sector is followed by a greater demand fór
Daniéi Bell 178

Services and a íurther corresponding shift in the slope of employment. In

thisfashion, Clark was able to chart the rate ofchange from a preindustrial
intő an industrial society and then intő a service society.
The difficulty remains the definition of Services. In classical economics.
beginning with Adam Smith, Services were thought of as unproductive
labor. Marx. accepting that distinction. had based one of his theories on
the crisis of capitalism. that of the falling rate o f profit, on the proposition
that as a higher proportíon of output shifted from "variable capital” (pro-
ductive labor) to "constant capital” (machinery. fór example), the rate of
profit would fali since the base on which surplus value was produced
would be shrinking (unless overcome by more intensive exploitation, such
as lengthening the working day or speeding up the pace of work). As the
notion that Services were unproductive became increasingly dubious,
economists were (aced with a double problem of redefinition: first. deter-
mining which Services were unproductive (e.g.. domestic servants) and
which were productive {e.g., education, by increasing the skill of labor, or
medicine, by making persons healthier or prolonging working life); sec-
ond, developing a more adequate set of distinctions within the Services
category. Somé writers sought to restrict the tertiary sector to auxiliary
blue-collar work, such as transportatíon, Utilities, repair (e.g., autó
mechanics), and personal Services (laundry, barbers, and so on), and to
define a quaternary sector made up essentialíy of the white-collar indus-
tries, such as banking, insurance, and reál estate, and a quinary sector,
made up of knowledge activities like scientific and technical research,
education, and medicine, While such distinctions are useful fór indicating
the complexity of occupational distributions, with them one loses the
thrust implicit in the origínal Colin Clark scheme, with its emphasis on
differential productivity as the mechanism fór the transition from one type
of society to another.
Without pretending to be exhaustive, I have adopted a scheme fór the
postindustrial society of classifying economic sectors as extractive, fabri-
cation, and information activities. The underlying sociological rationale is
that it seeks to look at the character of work as a shaper of the character of
individuals. The scheme is based on the distinction that somé societies are
primarily engaged ín games against natúré, others in games against fabrí-
cated natúré (things), and others in games between persons. It alsó de-
rives from the propositions I have pút forward regardíng the centrality of
knowledge in the postindustrial society, the primacy of a knowledge
theory of value as against a fabor theory of value. and the growth of
information Processing within the traditional sectors, such as agriculture,
manufacture, and Services, which is beginning to transform the character
of those sectors as well.
The Information Society 179

The Measurement of Knowledge

In 1958. Fritz Machlup, then at Princeton Universíty, made the first efforts
to measure the production and distribution of Knowledge. The definition
of knowledge was somewhat unsatisfactory, fór Machlup rejected “ an
objective ínterpretation according to what is known," as against a subjec-
tive Ínterpretation derived from what a knower designates as being
known.16 And Machlup worked from the standard national accounts, al-
though in important details he varied from standard usages.17
S1Í1I, Machlup'spainstaking work was crucial. In his accounting scheme,
he grouped thírty industries intő five major ctasses o( knowledge produc­
tion, processing, and distribution: (1) education, (2) research and de-
velopment, (3) média of communication, (4) ínformatíon machines, and (5)
information Services. The categories were broad, Education, fór example,
included education in the home, job, and church as well as in school.
Communications média included all commercial printing, stationery, and
Office supplies. Information machines included musical instruments, sig-
naling devices. and typewriters. Information Services included monies
spent fór securities brokers, real-estate agents. and the like.
Machlup estimated that $136,436 millión was spent tor knowledge, or29
percent of gross national product (GNP),18 and that 31 percent of the labor
force was engaged ín that sector. Of equal importance. he estimated that
between 1947 and 1958, the knowledge industries expanded at a com-
pound growth rate of 10.6 percent a year, which was double that of the
GNP itself duríng the same period. In 1963, Gilbert Burok, an editor of
Fortune, replicated Machlup's estimates and calculated that in that year
knowledge produced a value added of $159 billión, or 33 percent of the
GNP.19 Five years later, Professor Jacob Marschak, one of the most emi-
nent economists in the United States, in computations made in 1968, said
that the knowledge industries would approach 40 percent of the GNP in
the 1970s.20
The last decade has in fact seen enormous growth in the "information
economy,” which includes various fields. In education, whíle the rate of
growth of college education has slowed down, there has been a continu-
íng íncrease in aduit education which, in fact, has maintained its rise. In
health, the expansíon of health Services continues, particularly with the
multiplication of federal legislation. Information and data processing con-
tinue to rise, particularly as the volume of transactions and record keeping
increases. Telecommunications finds its major area of growth in interna-
tíonal Communications, particularly with the launching of new satellites.
Televisíon is on the threshold of a number of major changes with the
growth of both cable television and videó discs.
Still, if one wanted to measure the actual economic magnitudes of the
Dániel Bell 160

information economy, the difficulty is that there is no comprehensive

conceptual scheme that can divide the sector logically intő neatly distinct
units, making it possible to measure the trends in each unit over time. A
logical set of categories might consist of the followíng: knowledge (which
would include situses such as education, research and development, li-
braries, and occupations that apply knowledge, such as lawyers, doctors,
and accountants); entertainment (which would include motion pictures.
television, the music industry); economic transactions and records (bank­
ing, insurance, brokerage); and infrastructure Services (telecommunica-
tions, computers and programs. and so on).
Two somewhat different approaches have been adopted. Anthony Oet-
tinger and his cotleagues have taken the “ information industries" from the
Standard Industrial Classification used by the U.S. Census and listed their
gross revenucs in order to provide somé crude baselines to measure
changes. The difficulty here is that merging technologies and double
counting defeat such efforts. The second approach, a more difficult and
pioneering effort, is that of ('/arc Porát, which is to use the National
Income Accounts to define a primary sector, the direct sale of information
Services (like education, banking, advertising) to consumers, and then to
define a secondary sector—the planning, programming, and information
activities of priváté and public bureaucracies in enterprises and
government—and impute the value added by such activities to the national
product and national income.

The Information Economy

Marc Porát has broken down the National Income Accounts fór 1967 in
order to see what portions may be attributable, directly and indirectly, to
information activities. In doing this, he has used three measures to com-
pute gross national product. One is “ finai demand" (which eliminates the
intermediate transactions that would add up to double counting), the
second is “ value added,” which is the actual value added by a specific
industry or component o f an industry to the product, and the third is the
income or compensatíon received by those who create these goods and
Services. Theoretically, the totals of all three figures should be equal; in
fact, tor statistical reasons, in part owing to different methods of collec-
tion, the figures do nőt always dovetail exactly. Bút the virtue of using all
three is that one can make different analytical distinctions. Por my pur-
poses. the most important measure is that of value added. fór with it one
can seek to determine the actual Services provided by information activi­
ties and then check these figures against the income or compensations
received by those engaged in providing the Services.
Porat's work is the first empirical demonstration of the scope of irform a-
The Information Society 181

tion activities since Machlup, bút it goes far beyond Machlup's work, nőt
only because it uses finer categories and makes three different kinds of
estimations, bút alsó because it seeks to establish an input-output mátrix
that would permit, once the accounts were complete, an estimation of the
impact on other parts of the economy of a change, say, írom a "paper
economy” to an "electronic transmission” economy or from books to
videó discs as modes of instruction, along with hundreds of similar ques-
tions. Here, however, I am interested primarily in Porat's findings on the
value of information activities in the economy.21
Porát sets up a six-sector economy. There is a primary information
sector which includes all industries that produce information machines or
markét information Services as a commodity. (This includes the priváté
sector, which contributes about 90 percent of the primary information
Products and Services, and the government, which accounts fór the re-
maining 10 percent.) There is a secondary information sector with two
segments, the public bureaucracy and those priváté bureaucracies whose
activities are nőt directly counted in the national accounts as information
Services—such as the planning, programming, scheduling, and marketing
of goods or Services—yet who are actually engaged in information and
knowledge work. The value of these activities has to be imputed (fór
example, by factoring out the income or compensation of those persons
within a manufacturing firm who are engaged in such work). The three
remaining sectors consist of the priváté productive sector, producing
goods; the public productive sectors (building roads, dams, and so on);
and the household sector,
The primary information sector isthe one that is most easily measurable,
since it sells its products in a markét. It includes industries and activities as
diverse as computer manufacturing and Services, telecommunications,
printing, média, advertising, accounting, and education; it is the produc­
tive locus of an information-based economy.* In 1967, sales of information
goods and Services in the primary information sector to the four major
sectors of fínal demand amounted to $174,6 billión, or 21.9 percent of

* Porát divides the sector intő eight major classes of industries: (1) the knowledge
production and inventive industries; (2) inform ation distributíon and communica-
tion industries; (3) risk-management industries. including components of fináncé
and insurance; (4)search and coordination industries, including all markét inform a­
tion and advertising vendors; (5) inform ation Processing and transm ission Services,
both electronic and nonelectronic; (6) inform ation goods industries, including in­
form ation machines; (7) selected government activities that have direct markét
analogs in the primary information sector, including the Postai Service and educa­
tion ; and (8) support facilities such as Office and education buildings.
These eight major groups a re fu rth e r subdivided in tő 116 industries, which can be
located in the Standard Industrial Classification; the monetary figures can be lo-
cated in the National Income Accounts.
Dániel Bell 182

GNP. In other words, seventeen cents of every consumer dollár repre-

sented dlrect purchase of information goods and servíces. If one looks at
the íncome sídé, ín 1967 nearly 27 percent of all income originated with
information goods and Services. The civitian government was the most
information-intensive— almost 43 percent of all federal, State, and local
wages were paid to federal primary ínformation-creating personnel such
as Postai Service workers or education workers.
Strikingly, as Porát points out, over 43 percent of all corporate profíts
originated with the primary Information industries. All corporations in the
United States earned somé $79.3 billión in profits in 1967; the primary
information industries earned $33.7 billión. After removing the govern-
ment's share of the primary information sector's national income ($37.2
billión), the information industries alone accounted fór 21 percent of
national income bút 42 percent of corporate profits. Each dollár of em-
ployee compensation generated thirty-four cents in profits. as against a
ratio in the overall economy of twenty-one cents—a difference that Porát
attributes to the targe profits earned by the telephoné and banking indus­
tries with their high profit-to-labor ratios. Calculating value added. about
25 percent of totál GNP originated in the primary information industries. In
all, over $200 billión of the totál GNP o f S795.4 billión originated in infor­
mát! on goods and Services.
The most interesting and növel aspect of Porat's work is the definition
and measurement of the secondary information sector, a sector that Porát
deríves from Galbraith's notion of the “ technostructure." This is the sec-
tion of an industry that is dírectly engaged in information work bút whose
activíties are nőt measured as such, fór whíle the goods produced may be
sold in the markét (and thus are reflected in the GNP as manufactured
goods like automobíles or transportatlon activíties like aírline flights), the
information components ín those enterprises—the planning, scheduling,
and marketing activíties in automobiles; the computerized reservation
processes in aírline flights—are nőt counted dírectly in the GNP.
The secondary information sector expands fór several reasons, One is
the inherent tendency fór bureaucracíes to grow, whích whíle true is a
quite simpte-minded explanation sínce there are always constraints of
costs. A second, more serious reason is the multiplication of technical
activíties that comes with size, complexity, and advanced technology—
such as research, planning, qualíty controf, marketing, and the like. And
third is the fact that firms íntegrate or coordinate to economize on Infor­
mation costs. Thus a group of independent, high-quality hotels in different
cities recently banded together to createa common reservation service as
a means of competing with the large hotel chaíns by saving on Communi­
cations costs. In fact. as Porát points out, there are quasi-industries híd-
den within the secondary sector that under somé círcumstances could
The Information Society 183

become independent, primary (i.e., directly measurable) industries. One is

the hypothetical “ reservations industry." This "industry" sells its Services
to airlines, trains, hotels. theater box offices, and automobilé rental com-
panies through computerized data networks. In actual fact, each of the
industries or firms maintains its own reservations Systems, so the Informa­
tion costs are counted within the product cost. Yet if a single company
created an efficient reservations network that it could sell to all these
industries to replace the in-house Services they maintain themselves, then
these information activities would be measured in the "final demand" of
GNP.
Other than these quasi-information industries, the búik of the secondary
information sector consists of planning and financial control, the adminis-
trative superstructure that organizes and manages the activities of firms
or government agencies—in short, the priváté and public bureaucracies.
In 1967, according to Porát, 21 percent of GNP originated in the secondary
information sector—18.8 percent in the priváté bureaucracies and 2.4
percent in the public bureaucracies. Of the 168.1 billión in value added,
somé 83 percent ($139.4 billión) originated in compensation to information
workers. somé 3.5 percent ($5.8 billión) represented depreciation charges
on information machines, and the balance wasearned by proprietors per-
forming information tasks. Insum, nearly 50 percent of GNP, and more than
50 percent of wages and salaries, dérivé from the production, Processing,
and distribution of information goods and Services. It is in that sense that we
have become an information economy.
The growth of the secondary sector is, of course, the growth of the
bureaucratic society. In 1929, somé 13 percent of the national income
originated in the secondary sector, bút by 1933 it had fallen to 9 percent.
During the depression, the secondary sector shrank from 72 percent of the
sizeof the primary information sector to 40 percent. Bút it is in the war and
postwaryears, with the expansion of government and the growth of corpo-
rate size, that the secondary information sector begins to swell, so that by
1974, about 25 percent of the national income could be attributed to the
secondary information sector and about 29 percent of the national income
to the primary sector.
The final necessary component is the change in the composition of the
work force itself over time. From 1860 to about 1906, the largest single
group in the work force was in agriculture. In the next period, until about
1954, the predominant group was industrial. Currently, the predominant
group consists of information workers. By 1975, the information workers
had surpassed the noninformation group as a whole. On the basis of
income received, the crossover came earlier, since those in information
occupations, on the average, earn a higher income. By 1967, somé 53
percent of the totál compensation was paid to information workers.
Dániel Bell 184

Figure 9.2 and tables 92 and 9.3 illustrate the change. In 1930t there
were 12 miílion workers in the information sector, almost 10.5 millión ín
agriculture, 18 millión in industry, and 10 millión in Services. By 1970, there
were 37 millión in the information sector, less than 2.5 millión in agricul­
ture, 22.9 millión in industry, and 17.5 millión in Services. In percentage
terms, the labor force in the information sector today is over 46 percent; in
agriculture, 3 percent; in industry 28.6 percent; and in Services 21.9 per­
cent.
What of the future? Extrapolations can be deceptive. The information
sector has grown hugely in the last decade and a hatf, bút that has been a
result of both the rapid introduction of new technology in computers and
telecommunications and the economic growth rate that financed it. In
many sectors, such as education, pubíic polícy is the decisive variable.
Although the cohort of younger people wilí begin to shrink—in absolute
numbers it is still growing, bút the rate is slowing rapidly—there is an
evident desire on the part of many in the aduit population to undertake
continuing education. Thus many community colleges are finding them-
selves transformed intő aduit schools. Whether or nőt society can afford
these costs or wants to pay them is a different question. Bút asíde from
issues of pubíic policy the expansion of the information economy will
largely depend on two developments. One is automation—in industry and
in the white-colfar occupations. The second is the growth of information
and its retrieval—data bases, scientific information networks, and the
explosion of international Communications.

9.2
Four-Sector Aggregation of the U.S. W ork Force, 1860-1980 (using médián
estimates o f inform ation workers)
The Inform ation Society 185

Table 9.2
Four-Sector Aggregation of the U.S. Labor Force (Medián Definition)
Experienced Civílían Work Force
Inform ation A gricuiture Industry Service
Year Sector Sector Sector Sector Totál

1860 480,604 3,364,230 3,065,924 1.375,525 8,286,283

1870 601,018 5,884,971 4,006.789 2,028.438 12,521,216
1880 1,131,415 7,606,590 4,386,409 4,281,970 17,406,384
1890 2,821,500 8.464,500 6,393,883 5,074,149 22,754,032
1900 3,732,371 10.293,179 7,814,652 7,318,947 29,159,149
1910 5,930,193 12,377,785 14.447,382 7.044,592 39,799,952
1920 8,016,054 14,718,742 14,492,300 8,061,342 45,288,438
1930 12,508,959 10,415,623 18.023.113 10.109,284 51,056,979
1940 13,337,958 8,233,624 19.928.422 12.082,376 53,582,380
1950 17,815,978 6,883,446 22.154,285 10.990,378 57,844,087
1960 28,478,317 4,068,511 23,597,364 11,661,326 67,805.518
1970 37,167,513 2,466,883 22.925,095 17,511,639 80,071,130
1980:l 44,650,721 2,012,157 21,558,824 27,595.297 95,816,999
Percentages
1860 5.8 40.6 37.0 16.6 100
1870 4.8 47.0 32.0 16.2 100
1880 6,5 43.7 25.2 24.6 100
1890 12.4 37.2 28.1 22.3 100
1900 12.8 35.3 26.8 25.1 100
1910 14.9 31.1 36.3 17.7 100
1920 17.7 32.5 32.0 17.8 100
1930 24.5 20.4 35.3 19.8 100
1940 24.9 15.4 37.2 22.5 100
1950 30.8 11.9 38.3 19.0 100
1960 42.0 6.0 34.8 17.2 100
1970 46.4 3.1 28.6 21.9 100
1980* 46.6 2.1 22.5 28.8 100
J,Bureau of Labor Statistics projection
Daniéi Beli 186

Tablt 9.3
Two-Sector Aggregation ol the U S. Labor Force
Experienced Civilian Work Force
Inclusive Definition Restrictive Definition
Non- Non-
Information information Information information Totál U.S.
Vear Workers Workers Workers Workers Labor Force

1860 580.040 7,706,243 372,883 7,913.400 8.286.283

1870 788,837 11,732,379 500.849 12,020,367 12.521,216
1880 1,340,292 16.066,092 887,726 16,518,658 17,406.384
1890 2,980,778 19,773,254 2.480.189 20.273.843 22,754,032
1900 4.286.395 24,872,754 3.120,029 26,039,120 29,159,149
1910 7.283.391 32,516.561 4,537,196 35,262,756 39,799,952
1920 9,963.456 35,324,982 6,023,362 39,265.076 45,288.438
1930 16.031.889 35,025,090 8,883,914 42,173,065 51,056,979
1940 16.470,313 37.112,067 9.883.428 43,698.952 53.582,380
1950 21.691.532 36,152.555 13,940,424 43,903.663 57,844,087
1960 30.851,510 36.954.008 19,256.767 48,548,751 67,805.518
1970 40.529.588 39.541,542 29.464.497 50.606,633 00,071,130
1980a 49,154,120 46,662,879 39,955.688 55.861.311 95.816,999
Percentages
1860 7.0 93.0 4.5 95.5 100
1870 6.3 93.7 4.0 96.0 100
1880 7.7 92.3 5,1 94.9 100
1890 13.1 86.9 10.9 89.1 100
1900 14.7 85.3 10.7 89.3 100
1910 10.3 81.7 11.4 88.6 100
1920 22.0 78,0 13.3 86.7 100
1930 31.4 68.6 17.4 82.6 100
1940 30.7 69.3 18.4 816 100
1950 37.5 62.5 24,1 75.9 100
1960 45.5 54.5 28.4 71.6 100
1970 50.6 49.4 36.8 63.2 100
1980* 51.3 48.7 41.7 58-3 100
aBureau of Labor Statistics projection
The Information Society 187

FUTURE PROBLEMS: THE RETRIEVAL OF INFORMATION

In his Sartor Resartus, Thomas Carlyle wrote ironically, "He who first
shortened the labour of the Copyists by the device of movable type was
disbanding híred Armies. . . He was, of course. referring to Johann
Gutenberg (and praising hím as well fór “ cashiering most Kings and
Senates and creating a whole new Democratíc world: he had invented the
art of ponting"). Yet such “ technological” displacement, characteristi-
cally, had contradictory results. While old-fashioned calligraphers no
longer could practice their skill and thus were relegated to the artisan
scrap heap, more jobs were created by the increased demand fór printed
materials, and newer, less artistíc bút differently skilled mén found em-
ployment.
And yet initialty the pace of change was nőt so abrupt and rapid as to
create Wholesale turnoversin the print trade of the time. The printing press
of the eighteenth century was little different than that used by Gutenberg
three hundred years before. It was a wooden handpress on which a fiat
ptate was Iáid upon a fiat piece of paper with pressure created by the
tightening of screws. Wood was eventually replaced by metál and the
screw by a double lever, which allowed the speed of printing to be in­
creased by half. By 1800 a radically new method of printing using a
rotating cylinder—the basis of the modern press until the development of
photographic technologies—was invented and with its greater speed
began gradually to dísplace the fiat press. The double rotary cylinder,
developed fór newspapers in the 1850s, made it possible to print two sides
of a piece of paper at once. By 1893, the New York World's octuple rotary
press was printing 96,000 copies of eight pages in a single hour, whereas
seventy years before the average was 2,500 pages an hour.22
Such developments, understandably, went hand in hand with com-
plementary technologies. The linotype, developed by Mergenthater in
1868, replaced monotype by selecting and casting type by keyboard, re-
ducing composition costs by half while quintupling the speed of typeset-
ting. The paper industry, which until the early nineteenth century was a
time-consuming hand process using rags, was transformed in the middle
of the century by the Fourdrinier process which mechanized the produc-
tion of paper with the use of wire webs and cylinders. At the same time the
development of wood pulp and a practical pulping process displaced rags,
so that paper which had cost almost $350 a tón at míd-century had
come down to $36 a tón by the end of the century. Each of these develop­
ments was sped by new sources of energy. Printing presses, originally
turnéd by hand and briefly even by horse (in America at least), became
powered by steam and then by electricity. Papermaking, dependent in-
Dániel Bell 188

itially on waterpower, came to use hydraulic power accelerated by eíectric

turbines.
Bút what is so striking is how long it took, írom the time of Gutenberg,
fór all this to develop. It is only in the twentieth century that one finds the
mass production of newspapers (with míllions of copies of a síngle íssue
prínted overnight), magazines (set and printed in widely dispersed places
using common tapes), and books. And now, with the revolution in Com­
munications, all this will change. The information explosion is a set of
reciprocal relations between the expansion of Science, the hitching of that
Science to a new technology, and the growing demand fór news, enter-
tainment, and instrumental knowledge, all in the context of a rapidly
increasing population, more iiterate and more educated, living in a vastly
enlarged world that is now tied together, almost in reál time, by cable,
telephoné, and international satellite, whose inhabitants are made aware
of each other by the vivid pictorial imagery of televísion, and that has at its
disposal large data banks of computerized information.
Given this huge exptosion in news, statistical data, and information, it is
almost impossibte to provide any set of measurements to chart its growth.
Vet there is one area—the growth of scientific information—where somé
reconstruction of historical trends has been carried out, and I will use that
as a baseline fór understanding the problems of the next twenty years.
The historical picture of the knowledge explosion was first formulated
statistically by Derek de Solla Price in 1963, in his workL/ff/e Science, Big
Science. The first two scientific journals appeared in the mid-seventeenth
century, the Journal des savants in Paris and the Philosophical Transac-
tions o f the Roya! Society in London. By the middle of the eighteenth
century, there were only ten scientific journals, by 1800about 100, by 1850
perhaps 1,000. Today? There are no exact statistics on the number of
scientific journals being published in the world. Estimates rangé between
30,000 and 100,000, which itself is an indication of both the dífficulty of
definition and the difficulty of keeping track of new and disappearing
journals. In 1963, Price estimated that 50,000 journals had been founded,
of which 30,000 were still surviving. A UNESCO report in 1971 pút the
figure between 50,000 and 70,000. Ulrich'sInternationalPeríodicals Direc-
tory (a standard library source) in 1971-72 listed 56,000 titles in 220 sub-
jects. of which more than half were in the Sciences, medicine, and
technology; bul these were only of periodicals in the Latin script and
excluded most Slavic, Arabic, Orientál, and African languages.
Perhaps the most directly measurable indicators are university library
holdings. The Johns Hopkins University in 1900 had 100,000 books and
ranked tenth among American university libraries. By 1970, it had over 11/2
millión volumes, a growth of 3.9 percent per year, although ithaddropped
to twentieth piacé. In that same period, the eighty-five major American
The Inform ation Society 189

universities were doubling the number of books in their libraries every

seventeen years, (or an annual growth rate of 4.1 percent. (The difference
between 3.9 and 4.1 percent may seem slight, yet it relegated the Johns
Hopkins Library to the bottom of the second decile.)
A 1973 OECD survey of all the extant studies of the growth in scientific
knowledge came to the following conclusions:
1. In all the case studies, growth follows a geometric progression, the
curve being exponential.
2. However, the growth rates varied considerably, the lo west one being 3.5
percent yearly, the highest 14.4 percent.
3. The lowest growth rates are shown by the number of scientific periodi-
cals published, covering a 300-year period, and the number of specialized
bíbliographical periodicals involved in indexing and abstracting over a
140-year period. In the case of scientific journals, the annual growth rate
has been 3.5, 3.7, or 3.9 percent, depending whether the number pub­
lished in 1972 is taken as 30,000, 50,000, or 100,000. The growth rate fór
indexing and abstracting organizations has been 5.5 percent a year. In
1972, there were 1,800 such Services in science.
4. A recent series reporting the number of articles by engineers in civil
engineeríng journals (írom 3,000 pages of technical articles in three
specialized periodicals in 1946 to 30,000 pages in forty-two specialized
periodicals in 1966) shows growth rates of 12.3 percent a year.
5. The growth rate in the number of international scientific and technical
congresses increased almost fourfold in twenty years, rísing from 1,000 in
1950 to over 3,500 in 1968.23
The multiplication ín the number of scientific reports and documents
has naturally led to the conclusion that such progression cannot continue
indefinitely, that at somé point a slowdown would take piacé, probably in
theform of a fogistic curve that would symmetrically match the exponen­
tial rise of the ascent. The crucial question has been to identify the point of
inflection where the reverse trend would begin. Derek de Solla Price
argued in 1963 that "at somé time, undetermined as yet bút probably
during the 1940s or 1950s, we passed through the míd-períod in generál
logistic curve of Sciences body politic." In fact, he concluded, saturation
may have already arrived.24
Yet as Anderla noted in his study fór the OECD, "Today it is absolutely
certain that these forecasts, repeated without number and echoed almost
universally, have failed to materialize, at any rate so far." As evídence, he
assembled the number of abstracts published between 1957 and 1971 fór
nineteen scientific disciplines and demonstrated that between 1957 and
1967 the output increased by nearly two and a half times, fór an annual
growth rate of 9.5 percent. Over the fourteen years from 1957 to 1971, the
Dániel Bell 190

volume increased more than fourfold, fór a growth rate of 10,6 percent, so
that there was an escalation ín growth rather than the predicted reverse.26
The major reason fór this continued escalation is the tendency fór
Science to generate more and more subspecíaltíes, each of which creates
íts own journals and research reports system. At the same time, cross-
dísciplínary movements ariseto bridge somé of the subspecialties, extend-
ing the proliferation process even further.
What then of the future? The production of scientific líterature is deter-
mined ín the first instance by the projected rate of increase in the scien­
tific population. It is calculated that in 1970 the scientific population
represented about 2 percent of the totál labor force. The rate of increase
has been estimated variously at between 4.7 and 7.2 percent a year (a
fifteen-year and a ten-year doubling time, respectively), although certain
categories, such as computer scientists, have been increasing by more
than 10 percent annually. Taking 1970 as a base, one can estimate the
likely size of the scientific population in 1985 by making three assump-
tions: an unyielding exponential increase to the horizon year of 1985; a
break occurring in 1980, with the logistic curve begínning to slow
down at that time; or the point of inflection coming as early as 1975. Given
these assumptions, the number of scientists, engineers, and other lechni-
cians in 1985 could account fór a low of 3.8 percent to a high of 7.2 percent
of the totál labor force. If one takes the midpoínts, between 4 percent and
5.7 percent of the totál working population would be scientists and en­
gineers in 1985.
In order to project the volume of Information that is likely to be pro-
duced, we can take as a base a survey of the U.S. National Academy of
Science which revealed that in the early 1970s about 2,000,000 scientific
wrítings of all kinds were issued each year, or between 6,000 and 7,000
articles and reports each working day. Fór an internally consistent time
seríes, the most relíable intiicators are the statístícs of abstracts of articles
ín the leading specialized revíews, which from 1957 to 1971 increased
exponentially at a rate of more than 10 percent a year. As with the growth
rates ín the number of scientists, one can assume breaks in the logistic
curves at 1975, 1980, or 1985 and then take a médián figure. According to
these computatíons, there is every indícation that projections to within a
year or two of the 1985 horizon might well lie within the index rangé of
somé 300 to 400, In other words, the number of scientific and technical
abstracts would be three or four times the present number.

The End of the Alexandrian Library

Clearly, if the explosion in information continues, it cannot be handled by

present means. If by 1985 the volume of information is four (low estimate)
The Information Society 191

or seven times (high estimate) that of 1970, then somé other ways must be
found to organize this onslaught of babéi. In one of these pleasant exer-
cises that statisticians like to undertake, it iá estimated that under present
projections, the Yale University Library would need a permanent staff of
6,000 persons in theyear2040to cope with thebooksand research reports
that would be coming annually intő the library. (Such projections recall
earlier ones that if the U.S. telephoné System had to handle the current
volume of calls solely through operator-assisted methods, then every
female in the labor force—a sexist remark obviously made before women's
lib—would now be working fó r AT&T.)
Obviously, the information explosion can only be handled through the
expansion of computerized and subsequently automated information Sys­
tems. The major advance to date has been the computerization of abstract-
ing and indexing Services. Most of the printed abstract index bulletins in
research libraries are prepared from computer tape. The Chemical
Abstract Service (CAS), the largest in the field, is a case in point. Before
computerization, it took the CAS about twenty months to produce an
annual index; these are now available twice a year, while the unit costs fór
indexing have decreased from $18.50 to $10.54. Moreover, as the new
substances are recorded in the Chemical Registry System—thereare now
3,000,000 items in the fíles—it is possible to store, recreate, and display
structure diagrams on videó terminals from the computer-readable struc-
ture records stored in the system. A further development is the rise of
computer-based searching services, drawn from the tape initially used to
expedite the printing of indexes. Two American firms, the Systems De­
velopment Corporation and Lockheed Information Systems, provide on­
line searching to over thirty bibliographic data bases. Together they pro­
vide immediate access to over 15 millión citations, with an annual increase
of approximately 3.5 millión citations.26
The logic of all this is that the image of the Alexandrian Library—the
single building like the Bibliothéque Nationale, the British Museum, or the
Library of Congress—where all the world s recorded knowledge is housed
in one building, may become a sad monument of the printed pást. Data-
based Stores of information, especially in the scientific and technical field,
w ill come from specialized information centers. transmitted through com­
puter printouts, facsimile, or videó display to the user, who will have
consulted an index through on-line searching to locate items of interest
and then order them on demand.
All this supposes two things. One, the creation o f large-scale networks
in which a national system is built through the linkage of specialized
centers. And two, the automation of data banks so that basic scientific and
technical data, from industrial patents to detailed medical information,
can be retrieved directly from computers and transmitted to the user. Bút
Dániel Bell 192

both suppositions raise two very different problems. One is the intellectual
questíon raised by Winograd in chapler 4, the distínction between pro-
gramming a data base, and constructing a program fór use as a knowl-
edge base. Retrieving somé census ítems from a data base is a simple
matter; bút finding kindred and analogous conceptual terms—the han-
dling of ideas—raises all the problems that were first encountered, and
never successfully solved, in the effort to achieve sophisticated machine
translation of languages.
As early as the pre-Socratics, when philosophy was first becoming
setf-conscious, there was an awareness of the ambiguitíes of language
and the hope, as with the Pythagoreans, that certainty could be expressed
through mathematical relations. Descartes, in creating his analytical
geometry, thought he could substitute the "universal language of logic"
fór the messy imprecisions of ordinary language, as Spinoza felt he could
create a "morál geometry" to deal wíth ethical questions. In each genera-
tion that hope has arisen anew. In 1661 a Scotsman, George Dalgarno,
published his Ars Signorum in which he proposed to group all humán
knowledge intő seventeen sections (such as "politics" and "natural ob-
jects") and to label each wíth a Latin consonant. Vowels would be used to
label the subsections intő which each section was to be divided, and the
process of subdivision was to be continued with consonants and vowels
alternating. In thís way, any item of knowledge would have a specific
reference and identification.27
In the twentieth century we have had the effort of Whítehead and Russell
to formaiize all logic using a mathematical notation, the effort of the
logical positivists such as Carnap to construct (in theory) a language that
would avoid the ambiguitíes of ordinary díscourse and to propose (in
practice) a verifiability principle that would specify which propositlons
were testable and could be held to "make sense," as against those that
were (pejoratively) metaphysical, emotive, or theological and could nőt,
given the natúré of language, be “ proved." And most recentty, in the
Britannica 3, Mortimer Adler has proposed a new scholastic ordering of
knowledge. the Propaedia, that would guide encyclopedia users to interre-
lated sets of relevant terms, as his earlier Syntopicon sought to be an
intellectual index to the 101 major “ ideas" of humán thought.
The attempts to discipline humán knowledge and create a vast and
unifíed edífice. as Dalgarno and even Leibniz sought to do, were bound to
fai I. The effort to formaiize knowledge or construct artificial languages has
proved inadequate. The scholastic orderíngs of Mortimer Adler may help
an individual to tra ce th e bibliographic cross-relationships of ideas, bút if
the purpose of a library, or a knowledge-based computer program, is to
help a histórián to assemble evidence or a scholar to ‘'reorder" ideas, then
the ambiguity of language itself must be confronted. Terms necessarily
The Inform ation Society 193

vary in different contexts and lend themseives to different interpretations,

and historica! usages shift over time (consider ihe problem of defining an
intellectual, or the natúré of ídeology), making the probtem of designing a
“ knowledge" program quíte different from designing an “ information"
program.
The process of creating new knowledge (reasoned judgments) proceeds
by what Léon Walras. the great mathematical economist. called tátonne-
ment, trial-and-error tapping, by taking fragments of intellectual mosaics
whose larger shapes cannot be predicted in advance and fitting them
together in different ways or by regarding large conceptual structures
from a new angle, which opens up wholly new prisms of selection and
focus. A sophisticated reader, studyi ng a philosophical text, may make use
of the existi ng index at the back of the book, bút if he is to absorb and use
the ideas in a truitful way, he has necessarily to create his own index by
regrouping and recategorizing the terms that are employed. As John
Dewey pointed out in Art as Experience, the natúré of creativity is to
rearrange perceptions, experiences, and ideas intő new shapes and
modes of consciousness. In this process, no mechanical ordering, no
exhaustive set of permutations and combinations, can do the task. Des­
cartes once thought that the geometer with a compass could draw a
círcle more exactly than an artist could freehand. Bút a perfect circle, or
even a set of interlocking circles, is nőt art without somé larger conceptual
context that "redesigns" an older or different way of arranging shapes.
Art, and thought, as modes of exploration, remain primarily heuristic.
A more mundane yet sociologically important problem is the lack of a
national information policy on science and technical information, let alone
on library resources generally. Should there be a national scientific and
technical computer network? Should there be a government Corporation
or utility with direct responsibility to scientific and technical users or
simply a major, governmentally organized data base (tike the census)
made available to commercial Services that meet specifíc consumer
needs? Such questions have been raised since the creation of the Office of
Science Information within the National Science Foundation in 1958, and
they have been asked over and over again in a number of governmental
and National Academy of Science studies in subsequent years. No an-
swers have been forthcoming; no policy exists. Yet if science information
is the end product of the S35 billión annual investment that the nation
makes in research and development and information, broadly defined,
accounts fór almost 50 percent of the gross national product, then somé
coherent national policy is in order.

THE POLICY QUESTIONS OF THE INFORMATION SOCIETY

My basic premíse has been that knowledge and information are becoming
Dániel Bell 194

the strategic resource and transforming agent of the postindustrial soci-

ety. Inevitably, the onset of far-reaching social changes, especially when
they proceed, as these do, through the médium of specific technologies,
confronts a society with major policy questions. Here I can only schemati-
cally indicate somé of the questions society wíTI face in the next two
decades.

The New Infrastructure

Every society is connected by diverse channels that permit trade and

discourse between its members. These modes, or infrastructures, have
usually been the responsibility of government—as buitder, financier, main-
tainer, or regulator. The first infrastructure was transportation—roads,
canals, railroads, airways—which breaks down the segmentatior, of soci­
ety and allows fór the movement of people and goods. Caravans and trade
routes formed the social framework of older humán societies. The second
infrastructure has been the energy Utilities—waterpower, sleam pipes,
gas, electricity, oil pipelines—fór the transmission of power. By mobilizing
technological ratherthan natural sources of energy and linking them intő
power grids, nőt onty have we transformed the lives of cities through
lighting, bút we have provided power fór the fabrication of goods and
the use of consumer appliances. The third infrastructure has been Com­
munications—first the mails and newspapers, then telegraph and tele­
phoné, now rádió and television—as média fór the mounting explosion
of messages, the bombardment of sensory experiences, and the increased
degree of social and psychic interaction between persons that is now
accelerating exponentially.
In the next two decades. there is little likelihood of any major develop-
ments in the first infrastructure, that of transportation. The adoption of the
Concorde or other supersonic airplanes, if it comes, may halve the time fór
Crossing the óceán, bút the effect will be minor compared to the reduction
in the time needed to cross the Atlantic in the last hundred years, from
several weeks by steamshíp to six days by fást boát, to sixteen hours by
propeller pláne, to seven hours by jet. Mass transit in the cities. if it returns,
is unlikely to replace the automobilé or other modes of personal move­
ment unless fuel prices rise so high as to overthrow the hedonistic way of
life that has become entrenched in advanced industrial societies. The
rising demand fór personal transportation in the newer developing coun-
tries and increases in congestion may lead to new combinations of taxis,
leasing, and motor Utilities (in which one shares in a common pool). Bút
much of the vaunted experimental innovations, such as monorails or
automated elevated speedways or even hovercraft, have proved to be
either uneconomic or technologically too complicated.
In the second infrastructure, energy, there are clearly major new de-
The Information Society 195

velopments requiring large Capital expenditures. involving conservation

(insulated housing), better extractive techniques fór coal and its gasifica-
tion, potential uses of nuclear energy, research in tapping solar sources of
energy, and more efficient modes of electricity transmission, such as
superconductivity. These efforts, if made, will stimulate a huge expansion
in the areas of research and development (and of engineers and techni-
cally trained personnel), and, if successful, will establish new energy grids
that will supply a steady source of renewable power and once again bring
down the price of energy relatíve to other goods. Bút such changes, large
as they may be, are primarily substitutes fór existing energy sources and
modes of transmission. They do nőt presage huge upheavals in the role
energy plays in the society.
The really major social change of the next two decades will come in the
third major infrastructure, as the merging technologies of telephoné,
computer, facsimile, cable tefevision, and videó discs lead to a vast reor-
ganization in the modes of communication between persons; the trans­
mission of data; the reduction if nőt the elimination of paper in transac-
tions and exchanges; new modes of transmitting news, entertainment, and
knowledge; and the reorganization of learning that may follow the expan-
sion of computer-assisted instruction and the spread of videó discs.
One may be skeptical, as l am, about extravagant claims regarding the
quantum leaps in level of education that computer-assisted instruction
and videó discs wili bring. Learning, as I think we have learned, is a
functíon of both the abílíty to learn and the cultural milieu; any technology
is only instrumental, and its impact depends upon other social and cultural
factors, Bút in the realm of data transmission (especially in the world of
business) and in the development of knowledge networks (particularly in
Science and research), what Anthony Oettinger has called compunications
certaínly will stimulate vast social changes.
This upheaval ín telecommunications and knowledge poses two
economic-politícal policy problems, one structural, the other intellectual.
Thestructural question is what kind of technical-economic organization is
best desígned to be efficient, meet consumer (í.e„ industrial, commercial.
financial, scientific, library) use, and remain flexible enough to allow fór
continuíng technological development. One proposal is fór a single com­
puter utility that would centralize and provide a single source fór informa-
tion and transmission of data fór consumer use, either government-owned
(as are the telephoné and broadcasting systems in many European coun-
tries) or privately owned bút government-regulated, Like AT&T and the
major broadcast networks in the United States. Among different versions
of the computer utility idea, there is a proposal fór diverse sources of
information (i.e., different data banks operated publicly or privately) based
on a single transmitting system (such as the present telephoné quasi-
Dániel Bell 196

monopoly) or, conversely, a centralized set of data bases with diverse

means of transmission. Against these are the proposals fór a complefely
unregulated, competitive markét System, in which different “ producers"
would be free to set up diverse informational Services and transmission
would be through cable, microwave, or satellite communication operated
by different combines, each competing fór the business. These are the
issues whose economic aspects Noll has addressed in chapter 12.
It has been argued that a single national computing service, intercon-
necting all user terminals írom geographícally dispersed data banks,
would achieve vast economies of scale, and if run as a government utility
(like TVA) would avoid the concentration of vast power in the hands of a
single priváté enterprise. Against this, as Noll points out, computer Sys­
tems sell nőt merely computational power or data Processing bút ''infor­
mation,” and the large and varied needs of thousands of different kinds of
users fór different kinds of information—medical, technical, economic,
marketing—would best be served by specific firms that would be respon-
sive, in the way efficient markets can be, to the diverse needs of consum-
ers. Others have argued that government control cou Id be as dangerous, if
nőt more so, than priváté concentration since it could be more easily
misused fór political purposes. And there is the further question of
whether a competitive decentratized system would nőt be more flexibfe
technologically, and more innovative, than a large monopoly system,
either public or priváté. The record so far, in the instance of the computer
versus the telephoné, would indicate that technological innovation has
come more rapidly and more responsively in an unregulated and competi­
tive atmosphere than in the government-regulated sphere.
On the traditional grounds of economic efficiency and technological
responsiveness, it seems to me that Noll makes a convincing case fór the
primacy of the markét and fór a markét system. Yet he alsó points out that
regulators fend to see prices as taxes to be levied according to somé
calculus of social worth, favoring one group over another, rather than
seeing prices as signal-conveying information about costs that induce
buyers to make economically efficient decisions. He is, I believe. right in
his observation. Yet is the policy itself so wrong? Where markets are open
and competitive, the allocation of resources does respond most efficiently
to the preferences and demands of consumers, and this is the justifiable
defense, theoretically, of the markét as the arbiter of economic activity. Yet
if in the institutional world income distribution is grossly distorted, or
various social groups are discriminated against, then redress through
subsidy may be one means of achieving equity, even if sometimes at the
expense of efficiency. Second, there is the growing realization that mar­
kets do nőt often reflect the larger rangé of social costs that are generated
in the process, and these may be unfairly distributed. As Arthur Okun has
The I ntor mát ion Society 197

pointed out, the trade-off between efficiency and equity presents a reál
problem. The point is nőt to dísguise the issue bút to make it as explicit as
possible, so that one knows the relatíve gains and losses in equity and
efficiency that result from markét and regulatory decisions.
The second polícy problem posed by the upheaval in telecommunica-
tions is intellectual rather than structural and concerns the question of a
national information policy, particularly the dissemination of Science and
technical information. The government is obviously committed to the
furtherance of research and development, Increases in productivity de-
pend increasingly on the more efficient distribution of necessary knowl-
edge, bút so far there is no unified government policy or an organized
System to bring scientifíc and technical information to diverse users, to
speed the process of innovation, and shorten the time of development and
diffusion.
After Sputnik, there was a flurry of studies reviewing the problem. A
report by William C. Baker of Bell Laboratories stated the unexceptionable
principle that the flow of scientific information was necessary. A second
report in 1962 by J. H. Crawford fór the president's Office of Science and
Technology recommended that each agency of government set up a
specific Office to produce scientific information, and these were created in
the Department of Defense, the Atomic Energy Commission, and the Na­
tional Aeronautic and Space Agency. In 1963, a report by Alvin Weinberg
of the Oak Ridge National Laboratory argued that the government had the
further responsibility to organize the dissemination of research informa­
tion in order to avoid costly duplication of effort. The government did
create a coordinating body called COASTI (Committee on Scientific and
Technical Information) to implement this effort.
Yet the odd if nőt surprising fact is that little has been done. During the
Nixon administration, COASTI, the Office of Science and Technology, and
the Science Information Council were dismantled. Inevitabty the number
of hortatory studies multiplied. In 1969, the National Academy of Sciences
and the National Academy of Engineering brought forth the SATCOM
(Committee on Scientific and Technical Communication) report, which
involved more than 200 scientists, calling fór a national policymaking body
to deal with information policy. In 1972 the Federal Council on Science
and Technology and the National Science Foundation commissioned yet
another report, by Dr. Martin Greenberger of Johns Hopkins University,
which concluded, unsurprisingty, that the government was nőt well or­
ganized to deal with the problems of scientific and technical information
facing the country.
It still is nőt. Meanwhile, the number of scientific papers and the volume
of scientific information continueto rise. There is a growing trend toward
cross-disciplinary information which the single-disciplinary systems (such
Dániel Bell 198

as abstracting and indexing) are nőt equipped to handle. The proliferation

of diverse types of matériái, stored in different ways from books, films,
computer tapes, videó tapes, and so on, makes it difficult to keep track of
everything. And ftnally, the number of users continues to increase.
All trends pose a large variety of policy issues. Should there be, as
Fernbach suggests in chapter 8, a national Library of Data, like the Library
of Congress, to store all basic data and programs in giant memories?
Should thís library—if such a Library of Bábel as Jorge Luis Borges
envisaged ever comes about—alsó concern itself with the dissemination
of dala, asthe government’s Medlars system does fór medical information,
or should it be available fó r priváté companies, such as Lockheed or
Systems Development Corporation or the New York Times, to provide
specialized Services fór subscribers through proprietary Communications
and terminál Systems?
The growth of shared Communications Systems and on-líne terminals
makes a national scientific and technical information network a tangible
possibility, In chapter 17 Denicoff describes the development of the I n ­
teractive computer network invented in 1968 by Dr, L. G. Roberts fór the
Advanced Research Projects Agency (ARPA), which was first employed by
the Defense Communications Agency in 1976. Its most valued result,
according to Denicoff, was the emergence of a “ user community." The
operationa! reality of such a community, he writes, is the proof of the gains
we have made in scientific cooperatíon. In the same vein, Joseph Becker
has argued that
a national scientific and technical information network implies the inter-
connection of discipline-oriented and mission-oriented information Sys­
tems fór remote use through standard Communications. Unless cohesive
development takes piacé, the separate Systems wíll remain insulated from
one another and from their users. Bút, if maximum communication can be
established among them, the array can be converted intő a national re-
source of immense value to America's scientific enterprise.28
H. G. Wells, in one of his megalomaniacal visions of the future, proposed
a "world brain" that like a vast computer would bring together in one
piacé all organized scientific knowledge and make it available through
communication networks to the "new samurai," the coming scientific elite
of the world. Is such a technological phantasmagoria feasible (as somé
computer scientists claim it is) or desirable (as others do), or is it simply
one of those marvelously simple visions (like that of Sidney Webb) of tidily
and neatly organized bundles of knowledge that can be separated and
reassembled by pressing the right button? If the last, it is a deceptive
Vision, which misunderstands the way the mind actually works, and which
makes the sociological error of assuming that somé Central knowledge
System can function better than the decentralized, self-organizing system
in which demand specifies the organizational and markét response to the
The Information Society 199

needs of the users. This is an issue that should remain open to extended
debate, fór it is too serious and too costly to be settled on purely ideologi-
cal grounds.
And finally, on a more mundane level, there is the legal and economic
question o f what is an ‘'intellectual property"—at least where the intellec-
tual product is clearly defined (such as a book or a journal article), let
alone where the boundaries are blurred, as in the instance of a computer
program. How does one balance the rights of fair use as demanded by
libraries against the economic rights o f authors and publishers? As books
become stored in computer memories and can be retrieved on tapes and
printed by attached photocopying devices, who is to p a yfo rw h a t? Should
Xerox and IBM récéivé financial returns while the intellectual producers
gain only the psychic satisfaction of the widespread reproduction of their
words?
The courtsand the Congress have been struggling with these questions
fór years. Clearly no solution will completely satisfy those who press fór
the widest possible dissemination of intellectual matéria! under somé
fair-useand information-need concept, or those who demand payment fór
any use of copyright matéria!. Bút we need a clarification of the legal and
philosophical issues at stake.

Social and Economic Transformations

The major determínant of policy issues. as l have indicated, is the ques­

tion of what kind of infrastructure will be created out of the merging
technologies of computers and Communications. Inevitably this will give
rise to more diffuse policy issues deriving from the economic and social
transformations that may come in their wake. I will conclude by examining
five Central issues of this kind.
1. The location o f cities Historically, all cities were formed at the cross-
roads of overland caravan routes, at the strategic confluence of rivers. or
at large, protected ports on seaways and oceans as entrepőts and trading
centers. Almost all the major cities of the world have been located on
rivers, lakes, and oceans since transportation—and particularly waterways
fór heavy barge loads—tied areas together in the first infrastructure.
In the industrial age, cities were located near major resource bases,
such as coal and iron, as one sees in the English Midlands or the Germán
Ruhr and most strikingly in the great industrial heartland of the United
States, where the great iron-ore resources of the Mesabi Rangé in upper
Minnesota were connected to the great coal regions in Southern Illinois
and western Pennsylvania through a network of lakes and rivers. In this
way the great industrial cities of Chicago, Detroit, Cleveland, Buffalo, and
Pittsburgh were intricately linked in one huge complex.
In the transition to a service economy, the metropolitan cities became
Dániel Bell 200

the major fínancial centers and headquartersforthe greatenterprises.The

histories of New York and London form striking parallels. Both began as
port cities through which goods could be sent overseas or transported
inland. New York was a large, íce-free port, protected by two great bays,
yet connected through the Hudson River and the Erié Canal system to the
midwestern Great Lakes complex. As trade increased, banking, factoring,
and insurance árosé as auxiliary Services to commerce; later, wíth the rise
of industry, they became nerve centers fór financíal and stock transac-
tions. In íts third phase, New York became a large headquarters city, where
the major corporations located their head offices to take advantage of the
external economies offered by the concentration of banking, legal, pub-
lishing, and Communications Services.
In economic geography, the resource base was the decisive locational
factor up to the last forty years, when all this began to change. In the
United States ín the postwar years, the economic map of the country was
reworked largely through poiitics, since the new large aircraft, space, and
missile companies were created entirely by government contracts and the
decisions to locate them in areas like the Pacific Northwest, Southern
Caiifornia, and Southwest Texas were made on political grounds. With the
rise of air cargo, we have witnessed a phenomenon in which new “ airplane
cities,'’ such as Dallas-Fort Worth, Houston, Denver, and Atlanta, rather
than water and rail cities, serve as régiónál hubs fór industrial and com-
mercial spokes. And now, as the increasing spread (and cheapness) of
telecommunications reduces the former external economies of physical
proximity, we see the dispersal of corporate headquarters and major
white-collar concentrations like the insurance industry from the decaying
Central cities to the suburbs. The location of research laboratories, new
universities, and large hospital complexes is less dependent on the tra-
ditional factors of economic geography and more influenced by the near-
ness of educational facilities, easier life-styles, and political factors.
Phenomena like the development of “ Silicon V a lié /' in California—the
ribbon of electronics and computer firms stretching from San Francisco
to San Jose—and Route 128 around Boston were a response to the
availability of university research facilities, plus more pleasant space fór
the smaller-sized physical plants and offices than the industrial areas
could provide.
C. A. Doxíades has envisaged the growth of línear cities wíthout the
older foca) piazzas and markét centers of the classical European towns.
B. F. Skínner has suggested that ín an age of advanced communicatíon,
networks of towns will replace the large, increasingly ungovernable cities.
The question of whether these apocalyptic vísions will be realized is moot;
the life and death of cities is a long historical process. Bút what is chang-
ing is the concept of “ urbanism” itself. Thirty years ago Louis Wirth wrote
T/Je Information Society 201

a famous essay entitled "Urbanism as a Way of Life,” in which he summed

up the characteristics of urbanism as a híghly Interactive, heavily mobile,
culturally and politically attentive mode, as against the older sma!l-town
and rural patterns centered on the church and the family. What is happen­
ing today is that the entire nation (if nőt large parts of the wortd) is
becoming urbanized in the psychological sense, though increasingly more
dispersed geographically.
The changes in the character and pattern of telecommunications poses
probtems of national land use, of the social costs of dispersíons and
concentrations, the management of the decay of otd cities, and the control
of the sprawl of new ones. Inevitably, the decisions will ref lect the interplay
of markét and political forces, since neither one can be decisive in itself.
Bút it is the exact mix of the tw o th a t remains as the interesting sociologi-
cal question fór the next decades,
2. The possibilities of national planning León Trotsky once said that a
capitalist economy is one where each mán thinks fór himself and no one
thinks fór all. That a single "one" can thinkfor "a ll" is probably impossibte
and, if so, would be monstrous, since the "one" would be somé giant
bureaucracy and the "a ll" a putative single interest equally applicable to
all citizens in the society. As Alán Altshuler of MIT has remarked, “ those
who contend that comprehensive planning should play a large role in the
future evolution of societies must argue that the common interests of
society's members are their most important interests and constitute a
large proportion of all their interests, They must assert that conflicts of
interests in society are íllusory, that they are about minor matters, or that
they can be foreseen and resolved in advance by just arbiters Iplanners]
who understand the totál interests o f all the parties."
In this respect, Altshuler is probably correct, yet stich a view unduly
restrícts the meaning of planning in all its possible varietíes. The different
kinds of planning can be arrayed in a simple logical ladder:
a. Coordinated information Almost al! major enterpríses make five- and
even ten-year plans (fór product development, Capital needs, manpower
requirements, new plants) as a necessary component of their own plan­
ning. And various Services, such as the McGraw-Hill survey of Capital
spending budgets or the federally financed Universíty of Michigan surveys
of consumer intentíons, seek to provide more comprehensive Information
(or firms about these trends to aid them in their planning, A national
computerízed information service, through the Bureau of the Census or
somé similar government body, could bring together all such relevant
information—just as the various econometric models now in use make
forecasts of the annual GNP and its major components, which become the
basis of both governmenta! and priváté policies. To this extent. the idea of
a coordinated information system is simply an extension of the planning
Dániel Bell 202

process that is now so extensive in the corporate and governmental sec-

tors.29
b. Modeling and simulation Using an input-output mátrix, such as that
developed by Wassily Leontieff, one could test aJternative economic
polícies in order to weigh the effects of different government policies on
different sectors of the economy. In a more radical version, the Russian
economist Leonid Kantorovich has argued that a national computerized
economic System, registering the different prices and allocations of items,
could spot items that devíate from planned or targeted goals or the dis-
proportionate use of resources in various sectors.
c. Indicative pianning In this model. which is used by the French Com-
missariat du Plán, several thousand industry committees coordinate their
pfans regarding economic activities, and these plans become the basis of
governmental decisions to stímulate or inhibit certain sectors, largely by
easier credit facilities or credit restrictions.
d. National goals In this scheme, the government would stipulate cer­
tain major goals—the expansion of housing or levels of economic
growth—and monitor the economy to see whether such goals were being
achieved as a guide to which further measures (tax cuts, investment
credits, credit allocations, preferred sections such as housing) might be
necessary to achieve them.
e. Mobilized targets This is, in effect, a “ war economy,” such as that
exemplified by the War Production Board in the United States during
World War II or the British Ministry of Supply; in practice, it is the actual
natúré of Soviet "pianning,” In this system, certain key targets are
specified (level of Steel output, kind of machine tools, number of tanks and
aircraft, and so on), and the government physically allocates, by a priority
system, the key materials and manpower to designated factories. In this
respect, the entire economy is nőt planned, bút key sectors are controlled.
These different modes of pianning rangé from direct Controls and polic-
ing at one end to “ simple" informalion coordination at the other. Which
kind of pianning society wiII adoptis a political question. Given thedegree
of interdependence and the spillover effects of various individual deci­
sions, somé larger degree of pianning—analogous to the rise of environ-
menta! monitoring and reguládon pianning—than we now have is proba-
b!y inevitable. The computer and the large-scale information systems that
are being developed will make it feasible; bút how one reconciles pianning
with various kinds o f individual freedom is a very different and more
difficult question.
3. Centralization and prívacy Police and political surveillance of indi-
viduals is much more possible and pervasive because of sophisticated
advances in the information process. In a survey of federal agencies' use
of computerized data banks, former Senator Sam Ervin wrote in the pref-
The Intormation Society 203

ace to a report by the Senate Judiciary Committee's Subcommittee on

Constítutional Rights: "The sub-committee has discovered numerous in-
stances of agencies starting out with a worthy purpose bút going so far
beyond what was needed in the way of infOrmatlon that the individuals
privacy and right to due process of law are threatened by the very exis-
tence of files. . . . The most sígnificant finding is that there are immense
numbers of government data banks, cluttered with diverse information on
just about every Citizen in the country. The 54 agencies surveyed were
willing to report 858 [data banks] containing more than VA billión records
on individuals."
Government demand fór Information can be highly costly to enterprises
and institutions. Derek Bök, the president of Harvard, reported that the
demand of the governmental agency enforcing the affirmatíve action pro­
gram fór detailed information on every aspect of employment practices
and the need to keep records of all job searches fór applicants to teaching
and other positions cost the university over a millión dollars a year. What
information is necessary and what is nőt is a difficult question to decide,
particularly in the abstract. Yet the tendency of almost every bureaucracy.
reflecting an aspect of Parkinsons Law, is to enlarge its demands on the
principle that (a) "a ll" information might conceivably be necessary; and
(b) it is easier to ask fór everything than to make discriminations.
The simple point, fór it is one of the otdest and m ostim portanttruism sof
politics, is that there is an inherent potential fór abuse when any agency
with power sets up bureaucratic rules and proceeds without restraint to
enforce them. The other, equally simple point, is that control over informa­
tion lends itself more readily to abuse—from withholding information at
one end to untawful disclosure at the other, both processes exemplified by
Watergate— and that institutional restraints are necessary, particularly in
the area of information, to check such abuses.
4. Elité andmass Every society we have known has been divided, on one
axis or another. intő elite and mass. On a different axis, a society may be
designated as open or closed. In the pást, most societies have been elite
and closed in that aristocracies have been hereditary. Evén when there has
beenan examination systemfór choosing mandarins.as in Imperial China,
the selection process has been limited to a small eláss of persons.
In the West the major elites have traditionally been landed and proper-
tied elites. Evén in an occupation like the military, which requires somé
technical skill, until about a hundred years ago (in Britain, fór example)
commissions could be purchased. The older ladders of social mobility
were "the red and the black." the army and the church. Modern capitalist
and industrial society began to break open those molds. In business, there
was the rise of the entrepreneur, the engineer, and the manager. With the
succeeding breakdown of "family capitalism," the managerial elites were
Dániel Bell 204

no longer children of previous owners bút mén who earned their way up by
technical competence. In government, there was the expansion of the
administrative bureaucracy, in which top positions were achieved, as in
Francé, through a rigorous selection system by rites of passage through
the grandes écoles, or by patronage, as was usual in the United States.
Modern societies, in contrast with the pást, have become more open
societies. bút at the same time, as knowledge and technical competence
have become the requirement fór elite positions. the selection process has
fallen more and more onto the educational System as thesluice gates that
determine who shall get ahead. The result has been increasing pressure
On the educational system to provide "credentials" tor those who want to
move up the escalator of social mobility. In the postindustrial society, the
technical elite is a knowledge elite. Such an elite has power within intel-
lectual institutions—research organizations, hospital complexes. universi-
ties, and the like— bút only influence in the larger world in which policy is
made. Inasmuch as political questions become more and more intricately
meshed with technical issues (from military technology to economic pol­
icy), the knowledge elites can define the problems, initiate new questions,
and provide the technical basesfor answers; bút theydo nőt have the power
to say yes or no. That is a political power that belongs, inevitably, to the
politician rather than to the scientist or economist. In this sense, the idea
that the knowledge elite will become a new power elite seems to me to be
exaggerated.
Bút what is equally true is that in contemporary society there is a
growing egalitarianism fostered in large measure by sectors of the knowl­
edge elite, especially the younger ones, and given the most vocal support
by those in marginal positions and marginal occupations in the knowledge
sector. Within institutions, this has taken the form of attacks on "author-
ity" and “ professionalism" as elitist and demands that all groups have
somé share in the decision-making power. In certain European universi-
ties, fór example, even the nonprofessional staffs are given a voice and
representation in university affairs, while on academic issues, from cur­
riculum to tenuredecisions, the three "estates" of students, juniorfaculty,
and senior faculty have equal corporate rights. How far this egalitarianism
will go remains to be seen.
The fear that a knowledge elite could become the technocratic rulers of
the society is quite far-fetched and expresses more an ideological thrust
by radical groups against the growing influence of technical personnel in
decision making. Nor is it likely, at least in the foreseeable future, that the
knowledge elites will become a cohesive “ eláss" with common eláss
interests, on the model of the bourgeoisie rising out of the ruins of
feudálisai to become the dominant eláss in industrial society. The knowl­
edge eláss is too large and diffuse, and there seems little likelihood, either
The Information Society 205

in economic or status terms, that a set of corporate interests could de-

velop so as to fuse this stratum intő a new eláss. What is more likely to
happen, as I have argued previously, is that the different situses in which
the knowledge elites are located will become the units of corporate action.
One can identify functional situses, such as scientific, technological
(applied skills like engineering, medicine, and economics), administrative,
and cultural, as well as institutional situses, such as economic enterprises,
govemment bureaus, universities, research organizations, social service
complexes (like hospitals), and the military. The competition fór monies
and influence will be between these various situses, just as in the com-
munist world the major political units are nőt classes bút situses such as
the party, the government machine, the Central planners, factory manag-
ers. collective farms, research institutes, cultural organizations, and the
like.
What one sees in contemporary society is the muItiplication of con-
stituencies and consequently the multiplication of elites; and the problem
of coordinating these elites and their coatitions becomes increasingly
complex.
5. International organization The problems of creating a new infrastruc-
ture fór telecommunications (or compunications) on a national scale are
magnified when the questions are projected on the international scene.
Just as within the last thirty years the United States has become a "na-
tional society,” so in the next thirty years we will have an international
society—nőt as a political order, bút at least within the space-time
framework of Communications. Here nőt only is the scale enormously
larger, bút more importantly there is no common political framework fór
legislating and organizing the creation of a worldwide infrastructure.
International telephoné traffic, fór example, has been growing by about
20 percent a year, and international Communications is handled by In-
telstat, an international commercial satellite organization with ninety-odd
member countríes. Yet Intelstat has been largely dependent on one Ameri­
can aerospace company (Hughes Aircraft) to build the satellites and on the
American space agency to launch its satellites intő orbit. The day-to-day
financial and technical management of Intelstat has been in the hands of
an American Corporation, Comsat, whose ownership is distributed half
among ordinary shareholders and hatf by the large Communications com-
panies, among which AT&T has a prominent voice. The question of such
dominance is bound to become more and more of an international politi­
cal issue in the next decades.
On a different level, the creation of worldwide knowledge data banks
and Services becomes an important issue as more and more countries and
their scientific, technical, and medical organizations seek to share in the
Dániel Bell 206

enlarged computerized systems and on-line networks that are being de-
veloped in tbe advanced industrial societies.
And finaily—although this is only a sampling of the international issues
that will play a role in the transformation of contemporary society—there is
the question of the spread of computers. specificatly the sharing of ad­
vanced computer knowledge and the creation of international computer
data-transmission systems. In the period before World War I. steel produc-
tion was the chief index of the strength of nations. and when Germany
began to overtake Great Britain and Francé as a Steel producer, it was a
tangible sign of Ihe growth of her economic and military power. A few
years ago, the Soviet Union overtook the United States in steel output, a
fact that received only passing mention in the back pages of the New York
Times. Vet the Soviet Union is far behind the United States in the produc-
tion of computers and their degree of sophistication. The export of
computers—to the Soviet Union and to China—are still political, nőt com-
mercial questions, fór one of the chief uses of computers has been fór
military planning, the design of military hardware, and most importantly
the creation of guided missiles and ' ‘smart” bombs.

Turning Points and Promises

I have been arguing that Information and theoretical knowledge are the
strategic resources of the postindustrial society, just as the combination of
energy, resources, and machine technology were the transforming agen-
cies of industrial society. In addition—is the claim extravagant?—they
represent turning points in modern history.
D. S. L. Cardwell has identified four major turning points in the rise of
scientific technofogy.30 One was the éra of invention at the close of the laté
Middle Ages, signaled by the development of the clock and the printing
press. The second, the scientific revolution, was symbolized by Galileo.
with his emphasis on quantitative measurement and his technical analyses
of the strength of materials and the structure of machines (fór example,
the square-cube law on the natúré of size and growth). The third, the
industrial revolution of Newcomen and Watt, was the effort to realize a
Baconian program fór the social benefits of science. The fourth is repre-
sented in the work of Carnot and Faraday, nőt only because it produced
new conceptions of thermodynamics and field theory bút alsó because it
provided the bridge to a more integrál relationship between science and
technology.
The new turning points are of ’ wo kinds. One lies in the changing
character of science. The transmutation of materials made possible by
knowledge of the underlying structure of the properties of matter and the
reorganization of Information intő different patterns through the use of the
new communication technologies, particularly the computer, are trans-
The Information Society 207

forming the social organization of Science. On the one hand they create
Big Science and on the other enhanced communícation through on-line
networks, cooperative ventures in the discovery of new knowledge and the
experimenta! testing of results. Science as a "collective good” has be-
come the major productive force in society.
The second turning point is the freeing of technology from its "impera-
tive" character to become almost entirely instrumental. It was—and
remains—a fear of humanists that technology would more and more “ de-
termine” social organization because the standardization of production or
the interdependence of skills or the natúré of engineering design forces
the acceptance of one. and only one, “ best" wayof doing things—a theme
that itself was fostered by prophets of the industrial age like Frederick W.
Taylor. Bút the natúré of modern technology frees location from resource
site and opens the way to alternative modes of achieving individuality and
variety within a vastly increased output of goods. This is the promise— the
fateful question is whether that promise will be realized.

EDITORS’ POSTSCRIPT

McCarthy's critique of Bell's essay rests on the argument that the new
technology is nőt the main cause of change in our society today, given our
reliance primarily on p re - rather than post-World War II inventions. In
addition, the course projected fór technology may veer off in unexpected
dírections. If, fór example, firms and individuals keep progressively more
of their records in computers, then automated paper handling may indeed
reverse the growth trend forecast fór the “ knowledge industries.'' Further
comments and discussion on Bell's views can be found in Weizenbaum’s
essay.

NOTES

1
Fór an elaboration of this concept. see my book, The Corning of Post-Industrial
Society (New York: Basic Books, 1973). A paperback edition with a new introduc-
tion appeared in 1976 (New York: Harper & Row. Colophon Books).
2
Cyril Stanley Smith, "M etállurgy as a Humán Experience," Metailurgical Transac-
tions A, 64, no. 4 (April 1975):604. Professor Smith adds. “ As an undergraduate (a
half century ago) I had to decide whether to enroll as a ferrous or a non-ferrous
m etallurgíst; l heard little about ceramics and nothing whatever ab ou t polymers.
The curriculum , though refined in detaü. had pretty much the same aim as the
eighteenth century courses in the mining academy in Freiberg and the Ecole de
Mines in Paris." (Ibid., p. 604.)
3
Jacob Bronowski. ‘ Humanism and the G rowth of Knowledge." in The Philosophy of
Kari Popper, ed. Paul A. Schlipp (LaSalle, III.: Open C ourt Publishing Company.
1974). p. 628.
Dániel Bell 206

4
Norbert Wiener, I Am a Mathematicían (Cambridge, Mass.: MIT Press, 1970), p. 40.
(The book was first published in 1956 by Doubleday, New York.)
5
Smith. "M etallurgy as a Humán Experience.” pp. 620-621.
6
Harold A. Innis. "M inerva s Owl,” in The Bias ot Com m unication (Toronto: Univer-
sity of Toronto Press, 1951), p. 3, given as Ihe presidential address to the Royal
Society of Canada in 1947.
7
Fór example,
The use of clay favored a dom inant role fó r the temples with an emphasis on
priesthood and religion. Libraries were bulit up in Babylon and Nineveh to
strengthen the power of monarchy. Papyrus and a sim plified form o f w riiing in the
alphabet supported the growth o f dem ocratic organization, literature, and philoso-
phy in Greece. Following Alexander empires returned with centers at Alexandria
and elsewhere and libraries continued as sources of strength to monarchies. Romé
extended the political organization of Greece in its emphasis on law and eventually
on empire. Establishment of a new Capital at Constantmople was followed by
ímperial organization on the orientál model particularly after official recognition of
Christianity. Improvement o f Scripts and wider dissem ination of knowledge enabled
the Jews to survive by emphasis on the scriptures and the book. In turn Christianity
exploited the advanlages of parchment and the codex ín the Bibié. With access to
paper the Mohammedans at Baghdad and later in Spain and Sicily provided a
médium tor the transmíssion of Greek science to the Western w orld. Greek Science
and paper with tineencouragem ent of writing in the vernacular provided the wedge
between the temporal and the spiritual power and destroyed the Holy Román
Empire. The decline of Constantinople meant a stim ulus to Greek literature and
philosophy as the decline of Mohammedanism had meant a stim ulus to Science.
Printing brought renewed emphasis on the book and the rise of the Reformation. In
turn new methods of com m unication weakened the w orship o f the book and
opened the way fór new ideologies. Monopolies or oligopolies of knowledge have
been built up in relation to the demands of force chietly on the defensive, bút
improved technology has strengthened the position of force on the offensive and
compelled realignments favoring the vernacular. (Ibid., pp. 31-32.)
Marshall McLuhan, as is evident, was a disciple of Harold Innis (he wrote the
introúuction to the paperback edition of The Bias of Com m unication) and derived
most of his major ideasfrom hím. Bu1 McLuhan nőt only "hyped up" and vulgarized
Innis's ideas. he alsó reversed the thrust of his argument, fór Innisfeared that the
tendency of new média was to extend centralization and concentrate power while
McLuhan, though propagaling the notion o f a "global viliágé," argued that the
newer média would encourage decentralization and participatíon.
8
Encyclopaedia Britannica, 1970 ed., s.v, "inform atio n theory.”
9
Norbert Wiener, Cybernetics (New York: Wiley, 1948), p. 155.
10
Wassily Leontieff, "National Economic Planning: Methods and Problem s." Chal-
lenge, July-A ugust 1976, pp. 7-8.
Referring to the further consequences of this new capacity, Leontieff writes:
Such systematic inform ation proves to be most useful in assessing structural— in
this particular instance technological— relationships between the input require-
ments on th e o n e h a n d , and the levels of output of various industries on the other. In
the case of households these relationships would be between totál consumers'
outlay and spending on each particular type of goods. Stocks o f equipm ent, build-
ings and inventories. their accum ulation. their maintenance and theír ocoasinnai
The Information Society 209

reduction are described and analyzed in their mutual dependence with the flows of
all kinds of goods and Services throughout the entire system. Detailed. as con-
trasted with aggregative, description and analysis of economic structures and
relationships can indeed provide a suitable framework fór a concrete. instead of a
purely symbolic description of alternative methods of production. and the realistic
delineation of alternatíva paths of technological change (Ibid., p. 8.)
11
Indeed Information is merely the negative measure of uncertainty. so to speak. Let
me say immediately that I am nőt going to propose a quantitative measure. In
particular. the well-known Shannon measure which has been so useful in Com­
munications engineering is nőt in generál appropriate fór economic analysis. fór it
gives no weight to the value of the Information, if beforehand a large manufacturer
regards it as equally likely whether the priceof his product will go up or down. then
learning which is true conveys no more information, in the Shannon sense. than
observing of the toss of a fair coin. (Kenneth J. Arrow. In fo rm a tio n a n d E c o n o m ic
B e h a v io r, ed. Federation of Swedish Industries (Stockholm. Federation of Swedish
Industries. 1973].)
12
As Arrow remarks.
The presum ption that free markets w ill lead to an efficient allocation o f resources is
nőt valid in this case. If nothing else, there are at leasttw o salient characteristics of
Information which prevent it from being fully identified as one of the com m odities
represented in our abstract models of generál equílibrium : (1) it is by definiticn
indivisible in its use, and (2) it is very d iffic u lt to appropriate. (Ibid., p. 11.)
13
Róbert K. Merton, "Singletons and M ultiples in Science." in The Sociology of
Science, the papers of M erton, ed. Norm an W. S torer (Chicago: University of
Chicago Press, 1973. pp. 343-370).
14
The problem is that econom ists have no direct measures of such "in p u ts " and treat
them as "re sid u a ls," nőt accounted fó r by direct increases in the pro du ctivity of
Capital or labor. As Michael Spence writes.
The difflcu lty in measuring inform ation has hampered research concerned w ith the
eflects of inform ation on jeconom ic] grow th. It is common practice to estimate the
e ffe c to f education and knowledge on growth in GNP by first estim ating the impact
of reál factors like the increase in capilal stock, the labor force, and so on. One then
attríbutes the grow th that is nőt explained in these reál factors to increases in
knowledge.
15
There is a huge and growing literature on all these questions. I have drawn largely
on the reports of the Harvard Program on Inform ation Technology and Public Policy
fó r the matéria! in this section.
16
See The Production and Distribution of Knowledge in the United States (Princetcn:
Princeton University Press. 1962). Fór a detailed discussion o f M achlup's types of
knowledge in com parison w ith those of Max Scheler and my own, see Bell, The
Corning of Post-lndustrial Society, pp. 174-177. Since, fó r me. the heart of the
postindustrial society is th e new ways in w h ich knowledge becomes instrum ental
fó r Science and social policy, I have attempted an “ objective d e fin itio n " that would
a llo w a researcher to plot the growth and use o f knowledge.
17
Marc Porát has reformulated the 1967 National Income Accounts to make them
consistent w ith accepted practices. and despite somé adm itted deficiencies, he has
hewed to the standard usages. As Porát points out,
Dániel Bell 210

M achlup’s accounting scheme innovated rather liberalfy on the National Income

Accounts and Practices whereas this study does nőt. . . . His work includes an
adm ixture of "p rim a ry" and "secondary" type activities, whereas this study keeps
them distinct. Third, a variant of finaf demand is used by M achlup as a measure of
knowledge industry size, whereas this study uses prim arily the value added ap-
proach bút reports both sets of figures. . , .
'T h e Inform ation Econom y" (Ph.D. diss., Stanford University, 1976), 1:81-82.
18
Machíup's key data can be presented in tabular form :
D istribution of P roportion of Gross National Product Spent on Knowledge, 1958
Type of Knowledge and Source Amount in M illions
of Expenditures of Dollars Percentage of Totál
Education 60,194 44.1
Research and development 10,090 8.1
C om m unication média 38,369 28.1
Inform ation machines 8,922 6.5
Inform ation Services (incomplete) 17,961 13.2
Totals 136,436 100.0
Expenditures made by:
Government 37.968 27.8
Business 42,198 30.9
Consumers 56,270 41.3
Totals 136,436 100.0

Source: Fritz Machlup, The Production and D istribution o f Knowledge in the United
States (Princeton: Princeton University Press, 1962), pp. 360-361. Arranged in
tabular form by permission.
19
G ilbert Burck, "Knowledge, the Biggest G rowth Industry of Them A ll," Fortune,
November 1964.
20
Jacob Marschak, "E conom ics of Inquiring, C om m unicating, D ecidíng," American
Econom ic Review 58, no. 2 (1968), 1-8.
21
The statistics and tables here, except where noted, are taken from Porát, "The
Inform ation Econom y," vol. 1. The page citations refer to that volume. The figures
on trends in the work force are from a briefing packet that Mr. Porát had prepared
fór presentation at an OECD conference. I am grateful to him fó r makíng these
materials available to me, and fór his correspondence in clarifying somé of my
questíons. His revised work is scheduted to be published by Basic Books.
22
I am indebted fór this technological inform ation to a research paper by Paul
DiMaggio, a graduate student of sociology at Harvard.
23
Georges Anderla, Inform ation in 1985, A Forecasting Study of Inform ation Needs
and Resources (Paris: OECD. 1973), pp. 15-16.
24
Little Science, Big Science (New York: C olum bia University Press), p. 31. Por a
critical discussion of the use of logistic curves and somé questions about Price's
various starting points, see my The Corning o f P ost-lndustrial Society, chap. 2, "The
Measurement of Knowledge and T echnology," pp. 177-185.
The Information Society 211

25
Anderla. Information in 1985, p. 21. The major speciálist journals were: Chemical
Abstracts and Biological Abstracts (which between them accounted fór m ore than
550,000 items. more than halt of the one m illión produced in 1971). Engineering
Index Monthly, Metals Abstracts. Physics Abstracts, Psychological Abstracts. and a
Geology Index Service.
26
The figures are taken írom a paper by Lee Burchinal of the National Science
Foundation. ' National S cientific and Technical Inform ation Systems.’ ' presented to
an international conference in Tunís, Aprít 2 6 .1 9 7 6 .1 am grateful to Dr. Burchinal
fó r the preprint.
27
Cited by Colin Cherry. "T he Spreading W ord o f Science,” Times Uterary Supple-
ment, March 22, 1974, p. 301.
28
Remarks made at the Science Inform ation Policy W orkshop, National Science
Foundation. W ashington, D.C., December 17, 1974.
29
One maíor d ifficu lly is the inadequacy of our statistícs. As Peter H. Schuck remarks,
What is perhaps more disturbing, given the imminence of national econom ic plán-
ning, is the abject poverty of our econom ic statistical base. upon w hich a good
theory must be grounded. In recent years the inadequacy and inaccuracy of a broad
spectrum of economic indíces— includm g the Wholesale price index, the consumer
price index, the unemployment rate. and business inventory levels—have become
quite evident. The Wholesale price index, fór example. reflects only list prices rather
than actual transaction prices (which are often lower) and uses anachronistic
seasonal adjustm ent factors; yet it is consídered a bellw ether statistic in econom ic
forecasting. ("N ational Econom ic Planning: A Slogan w ilh o u t Substance," The
Public Interest, Fali 1976, p. 72.)
30
D. S. L. C ardw ell, TurningPointsin Western Technology (New York: Science History
Publications, 1972).