[Stinchcombe, Arthur L. (1985), Can the philosophy of science help science?, Contemporary Sociology 14 (2):164-166]

[ Books | Articles | Reviews | Index | The main menu ]

Can the Philosophy of Science Help Science?

[start of page 164]

Explaining Technical Change: A Case Study in the Philosophy of Science, by JON ELSTER. Cambridge & Oslo: Cambridge University Press and Universitetsforlaget, 1983. 273 pp. $39.50 cloth. $11.95 paper

Northwestern University

This is an introductory treatment of the philosophy of science for social scientists, illustrated by a philosophical critique of theories of technical change written by economists (including Marx as an economist). The author is one of the very few philosophers of science who has a command of several social science disciplines, and his model of what science is like is based on considerable understanding of biology and statistics as well as the command of physics and differential equations that is traditional among philosophers of science. But he still thinks like a philosopher, not like a scientist, in the following sense: He is interested in logical methods of evaluating and criticizing theories, and not much interested in empirical validation of theories. In particular, he does not pay much attention to the question of empirical strategy, of picking the best theory to work on, given the present state of knowledge, and the most strategic body of data to elaborate, modify, or reject that best- bet theory.

For example, Elster discusses why functionalism is a good strategy in biology by presenting the traditional Mendel-Darwin mutation-and-selection model, which says that functional strictures will lead to greater survival and replication of the genes that produce, them. But that model is in substantial empirical trouble, which has led to such revisions as those proposed by Holland (1975, especially chapters 4 and 6). Some of the empirical difficulties are that evolution goes much too fast to be explained by mutation and selection; that sexual reproduction is poorly, explained by it, yet apparently a great species advantage (and sexual reproduction is a structure of the genes, not a particular gene); that populations of a species show high variance on characteristics that seem to explain species dominance in their niches (such as language ability among humans); that niches dominated by one or two species also often support a dozen or two other species each occupying less than one percent of the biomass supported by the niche.

Mutation and selection traditionally conceived are very - good for solving the logical problem of why functional explanations might be logically adequate (and hence for explaining observed functional structures), but they have substantial empirical disadvantages for understanding the numbers of species, the within-species variance, and the rate of change of species that are found in ecological niches. These empirical troubles do not seem to discourage Elster, because in spite of them Mendel and Darwin serve to underpin his epistemological advice to biologists. The true theory, which I am not competent to give a bet on; may also underpin that epistemology, but may not be as different from social selection as Elster imagines.

What concerns me is that he imports his practice of refusing to worry about such major and fundamental empirical difficulties into much of his analysis of social theories. For the converse is that Elster is willing to sacrifice theories that have "only" the virtue of explaining some facts, because one cannot give (or he has not yet found) a perfectly adequate logical account of their virtues. I believe - this fundamentally mars his analysis of functional theories. The central difficulty with rejecting functional theory on the grounds Elster does is that there are a lot of patterns of facts in the world that can be explained by functional theories, in the sense that they are derivations from those theories. Unless one comes up with an alternative that explains the fact that a lot of structures, for instance, tend to occur in the environments in which the tensions that they relieve are highest, functional theories are a good empirical bet as a strategy of explanation. With a bit of ingenuity (including

[end of page 164, start of page 165]

empirical ingenuity) one can often patch up the various logical difficulties that, Elster finds with them.

For example, consider Elster's reformulation of the notion that nonfunctional structures tend to get into troubles that cause them to cast about for alternative structures, and that they stop casting about only when they hit on a functional one. Elster discusses this in connection with an example of Coser's (61-64 ; 148-49) . The idea is that too-rigid bureaucracies will get into trouble from time to time that will lead them to cast about for alternative forms, some of which will embody conflict and improve adaptability, while flexible bureaucracies will not get into troubles that lead them to cast about until they find rigidity. Elster recasts this as a Markov process in which nonfunctional (rigid) structures have a positive probability of making the transition to functional (flexible conflict-ridden) structures, but flexible structures have a zero probability of transition to rigidity.

Elster argues, "In the social case [as illustrated by the bureaucracy example], however, there are no reasons for believing that the speed of the process of adaptation much, if at all, exceeds that of the change of the criteria of adaptation. On the contrary, the very example considered above brings out very well that social changes do not have the slow and incremental character of biological evolution" (64). That is, transitions to a more flexible form do not go fast enough, so by the time the new structure becomes common, the environment will have changed so much that it is no longer functional (e.g., the new IRS regulations may not like flexible structures that live from day to day, without stable categories in their accounting systems). So the criteria of selection, of what is functional, change faster than the Markov evolutionary process for structures can keep up with, unless that evolutionary process is really intentional, rational problem-solving.

It seems to me that if one postulates (reasonably) that the transition rate toward flexibility increases with the rate of social change, because rigid organizations have more troubles in rapidly changing environments, this " logical" difficulty disappears. The evidence from, say, Chandler's treatment (1962: 52-113) of the change in structure of the Du Pont company in the face of market difficulties that were costing them a lot of money shows that the transition can come pretty fast when the world starts changing. That is, Elster's logical difficulty comes simply from the postulate (which he does not explicitly introduce) that the transition rate of rigid structures to flexible ones is a (small) constant. But if losing money causes faster floundering, the Du Pont company may be able to adapt successively to markets requiring different functional structures. The reason I want that logical difficulty in Coser's theory as restated by Elster to disappear is that I want to explain, for example, the tendency that Chandler demonstrates for changes in the number of product markets a firm participates in to result in changes in the administrative structure of firms in the direction of higher flexibility. To put it contentiously, the reason Elster wants to hold on to the logical difficulty, rather than inventing the obvious and easy solution for it, is that he is not worried about the facts for which Coser (and Chandler) invented the functional theory. Elster improves it in explicitness, but by adding the assumption of constant transition rates he creates logical troubles that would not worry Coser, Chandler, or me.

The first half of the book outlines the epistemology of three broad approaches to sociological explanation: causal (Elster is in favor), functional (against outside biology), and intentional (in favor). Then the second half of the book "applies" this epistemology to the field of explaining the rate of technical change and its direction (by which he means whether it saves labor or capital, or both). But epistemology is a narrow approach to those explanations, because it does not take much interest in whether they are empirically supported and empirically fruitful or not. Instead, it asks how far we can assess these explanations by attending only to the logic of the matter. My assessment is that the cores of these theories are empirical commitments, not logical exercises, and so there are rarely decisive logical things to be said. The theory of technical change in which Elster finds most logical ambiguity is perhaps Schumpeter's, though it is not clear whether this is because Elster thinks Schumpeter was confused or because Elster doesn't like Schumpeter's purple prose. But this seems to me to be the theory of technical change that has the most going for it empirically. With some improvements in logic and empirical definition from the work by Robert Nelson and Sidney Winter described by Elster, and by Scherer (1980: 407-438), it is leading to some of the most exciting empirical and theoretical work on innovation.

So I think the critique of theories of technical change carried out here is mostly beside the point, because the basic questions are em-

[end of page 165, start of page 166]

pirical. But it is very well done, and one gets a very good exposition of the core of various important theories of technical change that one cannot, as far as I know, get elsewhere. In particular, chapter 7 ( 158-184) is by far the best exposition of Marx's theory of technology and technical change I know, and a good analysis of why most of us have falsely attributed to, Marx a "fixed coefficients" notion of technology. Elster's philosophical talent leads him to go for the logical core of the explanations he expounds and so makes him a fine source for a broad overview of the field.

I should say that I have been making the general argument about the relative value of the philosophy of science as compared with factual research for a great many years, often in connection with Elster's work. The brunt of

this review is very predictable from its author, especially in conjunction with the author being reviewed. It may not say much about the distinctive features of the work reviewed. People who like clear thought always like Elster's work, and I would much rather disagree with him than with most people, because I can tell what he has said.

Other Literature Cited Chandler, Alfred D. 1962. Strategy and Structure. Cambridge, MA: MIT Press. Holland, John. 1975. Adaptation in Natural and Artificial Systems. Ann Arbor: University of Michigan Press. Scherer, Frederic. 1980. Industrial Market Structure and Economic Performance. Boston: Houghton Mifflin.

[end of page 166]

[Stinchcombe, Arthur L. (1985), Can the philosophy of science help science?, Contemporary Sociology 14 (2):164-166]

[ Books | Articles | Reviews | Index | The main menu ]

1 1