by Daniel Bell
Forecasting is possible where there are regularities and recurrences of phenomena (these are rare), or where there are persisting trends whose direction, if not exact trajectory, can be plotted with statistical time-series or be formulated as historical tendencies. Necessarily, therefore, one deals with probabilities and an array of possible projections. But the limitations of forecasting are also evident. The further one reaches ahead in time with a set of forecasts, the greater the margin for error, since the fan of the projections widens. More important, at crucial points these trends become subject to choice (and increasingly in modern societies these are conscious interventions by men with power) and the decision (to accelerate, swerve or deflect the trends) is a policy intervention which may create a turning point in the history of a country or an institution.
To put it a different way: forecasting is possible only where one can assume a high degree of rationality on the part of men who influence events—the recognition of costs and constraints, the common acceptance or definition of the rules of the game, the agreement to follow the rules, the willingness to be consistent. Thus, even when there is conflict, the situation can be mediated through bargaining and trade-offs, if one knows each party’s schedule of acceptable costs and order of priorities. But in many human situations—and particularly in politics—privileges and prejudices are at stake, and the degree of rationality or consistency is low.
What use, then, do forecasts have? Though they cannot predict results, they can specify the constraints, or limits, within which policy decisions can be effective. Given men’s desire to control their history, this is a distinct gain in social self-consciousness.
Forecasting has many different modes. Social forecasting differs from other modes in its scope and techniques. The most important distinction is that the sociological variables are usually the independent, or exogenous, variables which affect the behavior of the other variables. And, while having the largest sweep—and potentially the greatest power, compared to the other modes of forecasting—they are the least precise.
A quick review of these various modes of forecasting will illustrate what is at stake.
Technological forecasting deals with rates of change or permutations and combinations of factors within classes of events. Just as one cannot predict events, one cannot predict specific inventions. One can, however, forecast the necessary next steps of a succession of changes within a closed system. Thus it is possible to project speed trend curves—an important factor in transportation—from jets to supersonic speeds; one can take computer memories, extrapolate the next level of capacities, and fit these within “envelope curves.” 3 Such projections can be made because technology has finite parameters which are defined by physical constraints. Thus, the maximum speed on earth is 16,000 miles an hour; higher speeds take one into orbit. Or, in computer speeds, one is limited by the character of the transmitting units: originally vacuum tubes, then transistors, now integrated circuits. Theoretically, one can specify the kinds of materials (new tensile strengths or lightness) or the processes (e.g. miniaturization) necessary to achieve the next level of speed or capacity that is sought. Then one looks for these materials or processes. But this, however, is a matter of economics—the costs of the search, the assessment of possible gains, the amounts already invested in existing technologies, the scope of markets for new products, and the like. And this is outside the technological system.
Demographic forecasting—population statistics are the foundation of most economic and social analysis—is a curious mixture of indeterminacy and a modified closed system. The number of children born in any particular time period is subject to changes in values, the fluctuations of economic conditions, and the like. But once a given number are born we can predict from actuarial tables, with a high degree of probability, the numbers that will survive, and the rate of the cohort’s diminution over time. From this one can estimate such social needs as education, health, and so on. But the original decisions are indeterminate and sociological.
There are three kinds of economic forecasting. The first is the simple market survey, based on income data, age distributions, household formations, and anticipated needs, which firms use for anticipating consumer demand, the amount of inventory to carry, and the acceptance of new products. The second, and most standardized, is the creation of time series of macrovariables—e.g. wholesale and consumer price indexes, industrial output, agricultural productivity, rate of unemployment, and a hundred other items—that serve as indicators of business activity and which are combined to make forecasts about the state of the economy. The third, and most sophisticated, is the econometric model which, by defining the actual interaction of the crucial variables in the system, seeks to simulate the reality of the economic system as a whole.
Again, there are limitations. Market surveys are subject to the usual hazards of discrepancy between individuals’ attitudes and behavior, and this widens when there is a high level of discretionary income, since an individual can postpone purchases or be “indifferent” (in the technical economic sense of the word) to alternative items (a second car, a long vacation, a swimming pool) depending on price. Trend extrapolation is subject to a “system break” which comes exogenously. Thus, from 1910 to 1940, the index of agricultural productivity went from a base of 100 to about 125; if it had continued at the same rate for the next twenty years as it had for the previous thirty, it would have reached about 140 in 1960. Yet in 1960 the actual mark was about 400. During the 1940s a system break was created by huge wartime demand, loss of manpower, and a revolution in farm technology because of new fertilizers. The output per man-hour almost quadrupled in the forties and fifties, and at the same time there was a huge reduction in the number of farms and a large migration to the cities after the war.4 The econometric model has the virtue of being a closed system, but its finite parameters are established by the analyst, rather than by physical laws. Its difficulties, of course, lie in the correct identification of the relevant variables and of the exact order in which they interact so that the actual economic flows can be simulated. The Brookings model which makes quarterly forecasts, completed in 1965, contains over 300 equations and endogenous variables and more than 100 exogenous variables—at which point the authors conclude that, “having examined the composite equation system, it should be clear to the reader that the task of building a large-scale quarterly econometric model of the U.S. economy has only begun.” 5
Political forecasting is the most indeterminate of all. In some societies, certain structural features have a high degree of stability. Thus I can predict (with a fair degree of certainty) that in 1976, 1980, and 1984 there will be presidential elections in the United States, or that within every five years there will be at least one parliamentary election in Great Britain—no mean feat, since one cannot make such predictions about many countries. (Can one equally assess the political stability of Italy, let alone African or Latin-American countries?) Through public-opinion polling, I can make some fair forecasts about political events in the stable democracies. But the most important political issues involve conflict situations in which the chief players must make uncertain or risky calculations about each other.6 A game theorist might order the array of choices, but only specific information about motives can permit one to say which choice will be made. And the degree to which crucial political decisions are carried out often depends on qualities of leadership and strength of will; such aspects of personality are not easy to calculate, especially in crisis situations.
There are also three kinds of social forecasting: the extrapolation of social trends, the identification of historical “keys” that turn new levers of social change, and projected changes in major social frameworks.
The most common, especially for short-run forecasting, is the projection of social indicators: crime rates, the number who will be educated, health and mortality data, migration, and the like. But such data have serious drawbacks. First, it is difficult to aggregate many of the indi
cators in a meaningful way. What does it mean, for example, to say that “crime” is increasing? The “crime rate” used by the FBI jumbles together the number of murders, rapes, assaults, burglaries, car thefts, etc., but these figures are not weighted and have no common metric. One can take a pound of potatoes and a pound of automobile and convert them into a common metric of dollar costs, and one can weight different kinds of purchases in a consumer price index. But how does one construct a common crime rate, or a general health index, or extent-of-education rate? The second difficulty is that even when there are fairly unambiguous rates, the time periods for the data are very short, and we do not know how meaningful some of the changes are. (For example, the decline in marriage age which began in the mid-fifties seemed to stop by 1970 and even to reverse itself. As for rates of divorce, are the numbers increasing, or have the rates simply levelled off?) Third, we do not know exactly what relates to what and how. In a general way, we know that residential segregation of races and classes widens the disparity of educational competences; that the amount and character of education affects occupational choices and social mobility in the society; that there are correlations between heavy migration and the rates of crime. Bur we do not have a “model” of the society analogous to our model of the economy, and so we lack precision in relating rates of social change to each other.7
Changes in values and the rise of new social processes herald major societal changes whose drift can be charted roughly along historical time. Tocqueville’s Democracy in America, published in 1835, is a powerful work that still seems fresh today because he identified one of the major “irresistible” forces transforming society: the drive for equality. In a different vein, Max Weber identified the process of bureaucratization as transforming the organization and administrative structures of society, but he also saw this change, which has revolutionized the work life and social relations of most persons in society, as part of a more pervasive process of the rationalization of all life in modern society.8
In the last hundred and fifty years, the social tensions of Western society have been framed by these contradictory impulses towards equality and bureaucracy, as these have worked themselves out in the politics and social structure of industrial society. Looking ahead to the next decades, one sees that the desire for greater participation in the decision-making of organizations that control individual lives (schools, hospitals, business firms) and the increasing technical requirements of knowledge (professionalization, meritocracy) form the axes of social conflict in the future.
But the identification of historical “keys” is quite tricky. It is the fashion these days to read into many social tendencies or new social movements portents that may not be there, or which are quickly erased (since the pace of change in intellectual fashion is often greater than in other fields). There are, therefore, few sure guides as to which new ideas, values or processes are genuine turning points in social history. Failing this—or at least guarding against such overestimation—we turn to changes in social frameworks.
Social frameworks are the structures of the major institutions that order the lives of individuals in a society: the distribution of persons by occupation, the education of the young, the regulation of political conflict, and the like. The change from a rural to an urban society, from an agrarian to an industrial economy, from a federalized to a centralized political state, are major changes in social frameworks. Because such frameworks are structural, invariably they are crescive and difficult to reverse. For this reason they can be more readily identified. But such changes in frameworks are on a large scale, and they do not allow us to specify the exact details of a future set of social arrangements. When such changes are under way, they allow us not to predict the future but to identify an “agenda of questions” that will confront the society and have to be solved. It is this agenda that itself can be forecast.
The idea of the post-industrial society, which is the subject of this book, is a social forecast about a change in the social framework of Western society.
A Methodological Excursus
Social frameworks are not “reflections” of a social reality but conceptual schemata. History is a flux of events and society a web of many different kinds of relations which are known not simply by observation. If we accept the distinction between matters of fact and matters of relation, then knowledge, as a combination of the two, depends on the correct sequence between factual order and logical order. For experience, the factual order is primary; for meaning, the logical order. Mind knows nature by finding some language in which to express an underlying pattern. Knowledge, thus, is a function of the categories we use to establish relationships just as perception, as in art, is a function of the conventions we have accepted in order to see things “correctly.” As Einstein once put it: “It is the theory which decides what we can observe.” 9
Nomen est numen, to name is to know, is an ancient maxim. In the contemporary philosophy of science, nomen are not merely names but concepts, or prisms. A conceptual schema selects particular attributes from a complex reality and groups these under a common rubric in order to discern similarities and differences. As a logical ordering device, a conceptual schema is not true or false but either useful or not.
A conceptual schema—in the way I use the device—rests on an axial principle and has an axial structure. My purpose is to restore some of the informing power of older modes of social analysis.
To set a problem, as Dewey observed, is the most effective way of influencing subsequent thought. Marx set the problem of how to define society by positing the idea of a substructure based on economic relations, and a superstructure that was determined by it. Subsequent writers inverted the relationship by insisting on the primacy of ideological or cultural or political factors or, as the convention became accepted, emphasized the interaction of all factors and denied that any one of them was primary. Thus the attack on monocausal theory ended by denying a general theory of social causation or even the effort to search for primacies. As one sociologist puts it: “Contemporary systems theory sees society as a total unaggregated system whose dynamic nature stems from the interplay of its component subsystems with each other as well as with the external environment.” 10 One posits a set of subsystems—the educational, the occupational, the political, the religious, the socialization—which influence each other, yet there are no clues as to which is the most important, or why. Everything is dissolved into interacting forces.
The idea of axial principles and structures is an effort to specify not causation (this can only be done in a theory of empirical relationships) but centrality. Looking for the answer to the question how a society hangs together, it seeks to specify, within a conceptual schema, the organizing frame around which the other institutions are draped, or the energizing principle that is a primary logic for all the others.
Many of the masters of social science used the idea of axial principles or axial structures implicitly in their formulations. Thus for Tocqueville, in his Ancien Régime, the entire interpretative structure of his work—the emphasis on the continuity of French society before and after the Revolution—rests on an axial structure, the centralization of administration in the hands of the state. In Democracy in America, equality is the axial principle that explains the spread of democratic feeling in American society. For Max Weber, the process of rationalization is an axial principle for understanding the transformation of the Western world from a traditional to a modern society: rational accounting, rational technology, the rationalistic economic ethic and the rationalization of the conduct of life.11 For Marx, the production of commodities is the axial principle of capitalism, as the business firm is its axial structure; and for Raymond Aron machine technology is the axial principle of industrial society and the factory its axial structure.
Conceptual prisms are logical orders imposed by the analyst on the factual order. But since the factual order is so multifarious and complex, many different logical orders—each with its own axial
principle—can be imposed on the same time or social frame, depending on the questions one has in mind. A shortcoming of eighteenth- and nineteenth-century social thought (and of physics) was its naive scientism. Reality existed “out there,” and the only problem was to get a true reflection of it, undistorted by bias, habit, received prejudices, and the like. (In Francis Bacon’s classic formulation, distortions were caused by the idols of the tribe, the idols of the cave, the idols of the marketplace, and the idols of the theater.) Or the mapping of the social world was conceived as being like a Mercator projection in which the map is drawn as a “plan,” no different than an architect’s blueprint, in which the viewpoint is at infinity; i.e. one is not over a particular point on the map, but over all points simultaneously. But even the invariable placing of North at the top of the map is a geographical convention (of relatively modern date), and one can learn much more about economic (and strategic) geography by looking at perspective maps, which are drawn from the standpoint of an observer at a finite point. Seeing Europe “from the East,” i.e. in a map drawn from the standpoint of an observer in the Pacific looking east, gives a more compelling view of the land mass of Russia than any conventional map.12
Conceptual prisms and axial structures are valuable because they allow one to adopt a multi-perspective standpoint in trying to understand social change, but they do not forgo the value of perceiving the “primary logic” of key institutions or axial principles within a particular scheme. Thus the terms feudalism, capitalism, and socialism are a sequence of conceptual schemes, in the Marxist framework, along the axis of property relations. The terms pre-industrial, industrial, and post-industrial societies are conceptual sequences along the axis of production and the kinds of knowledge that are used. Depending on the axis, we can highlight similarities or differences. Thus along the axis of property there is a contradictory relation between the United States and the Soviet Union, in that one is a capitalist and the other a (statist) socialist society. Along the axis of production and technology, both the Soviet Union and the United States are industrial societies and thus somewhat congruent. In this respect, when one looks at the Soviet Union and the United States one need not depend exclusively on a principle of convergence or of inherent conflict, but one can specify the rotating axes along which the distinctions are made. In this fashion, one can avoid a single-minded determinism, such as an economic determinism or a technological determinism, in explaining social change, yet single out a primary logic within a given conceptual scheme. One forgoes causality but emphasizes significance (or in Dilthey’s sense, meaning). One is able, also, to create a principle of “complementarity” in social explanation.13
