The Dialectic of Knowledge-in-Production:
Value Creation in Late Capitalism and the Rise of Knowledge-Centered Production


James Curry Department of Social Studies El Colegio de la Frontera Norte

[email protected]


It is now almost taken for granted that we are living in an economic era in which knowledge and information are of crucial importance in the creation of value. Many contemporary discussions of this phenomenon treat the categories of knowledge and information uncritically, usually by merely asserting their increased importance, and assessing the effects. Among the effects touted by some Information Age enthusiasts is the transformation of workers into owners of capital (the knowledge in their heads). This paper utilizes an approach derived from Marx’s method of analysis in Capital, viewing knowledge and information as discrete moments in the process of production and value creation. After an abstract discussion of knowledge as a process, this paper discusses the implications for the overall structure of capitalist economy, concluding that the emancipatory claims for the knowledge economy are overstated.


The importance of knowledge and information in the contemporary economy has been long established (see, for example, Bell, 1976; Block, 1990; Toffler, 1991; Mandel, 1975; Jameson, 1991, etc.). The approach of most of these claims about the “information economy” is uncritical. We are said to live in an “age” of information, where new and exciting opportunities abound and the physical drudgery of the past is replaced by an endlessly dynamic economy whose chief protagonist is the “knowledge worker,” now at the center of history because he or she now owns the key means of production–the knowledge in their head (Drucker, 1993). Others point out that the proliferation of information technology is leading to the creation of entirely new sets of spatial, cultural, and social relations (see, for example, Mitchell, 1995; Stallabrass, 1995; Negroponte, 1995) As a general proposition, the notion that information and knowledge are of central importance both to the functioning of production, as well as to the commodities produced, is essentially uncontroversial. Thus, the purpose of this paper is not to assert that information is important yet another time, but rather to reflect on just what the actual “role” that knowledge and information now play in the late capitalist economy.

Knowledge has been important to production across all historical epochs; in the most basic sense labor is human creativity, mediated by knowledge, and applied to the material world. This has not changed, and it is unlikely that it will change anytime in the foreseeable future. What has changed, apparently, is the quantity, the quality, and the density of knowledge and information, the speed in which it moves and changes, and the amount of it which is embodied in the design, production, and marketing of even the most prosaic products. But even given these sorts of developments, can we really say that we have entered a new era, one in which the entire dynamic of capitalist accumulation has radically changed; an era where, according to Gilder (1989), the economy of mind has triumphed over the economy of matter? Have we, as in Drucker’s formulation, moved to a post-capitalist society? Have knowledge and information superseded physical labor as the pivotal determinant of value in the contemporary capitalist economy? Possibly. But we can’t be sure unless we have a clear conception of just what the “new” economic role of knowledge is.

The analysis presented in this paper will utilize as a methodology the dialectical approach pioneered by Karl Marx in the mid-19th century and perhaps best exemplified in his book Capital. There are two basic reasons for this. The first is that this methodology is best suited to conceptualizing the protean nature of knowledge’s role in capitalist industrial production. Many recent explicit treatments of knowledge in production (for example, Drucker, 1993; Fransman, 1994) usually reify knowledge, emphasizing its importance as, for example, an important input into the production process, but rarely treating knowledge as a process itself, or if so, only in a very limited sense. The advantage of a dialectical approach is that it enables us to grasp a complex subject like knowledge in all its multiple determinations without unreasonably prioritizing any one of those determinations over the other. So instead of using a reified conception of knowledge, taking it as a unified, one-dimensional object which, if properly defined, can then be used to “explain” the rest of the economy, I will attempt to treat knowledge as a process which “runs” alongside, and perhaps even modifies, the various other determinations of capital-as-a-process.

The second reason to utilize a dialectical approach adjunct to Marx‘s work in Capital is that it avoids not only the problem of reification mentioned above, but also the mystification of knowledge that infuses much of the literature. Much of this has ideological roots arising out of the cult of the entrepreneur/manager, as well as the need to comprehend, in the minds of many analysts, the fearsome rise of East Asia (see Nonaka and Takeuchi, 1995). In this reading America’s strengths–its innovative entrepreneurial business culture and its leadership in the science of management–make it perfectly suited to leverage the emerging knowledge economy. Yet the problem also infects more sensible analyses as well. Fransman (1994) in an otherwise cogent analysis of the role of knowledge in the firm commits this type of error. He starts out correctly distinguishing between knowledge and information–something which I will do in this article, but he distinguishes between them on an inadequate basis. According to Fransman “information refers inherently to a closed set of data. However, knowledge is essentially open-ended” (1994:716).

But knowledge is not entirely open-ended. This is the kind of idealist approach Marx spent so much time and energy fighting against over 100 years ago. Recently it has gained currency in the form of deconstruction, which considers all “cultural” production to be part of a vast, open-ended, collection of texts (Ellis, 1989). In order to understand the fallacy of this approach we must understand first that knowledge is a social process. It is perfectly reasonable to think that the knowledge creation process is potentially open-ended. This process can conceivably go on forever. Where it may lead is dependent only on the socially multideterminant process or act of knowledge creation. Thus, in a limited conceptual sense we could say that knowledge is open-ended.

However, if we dig deeper into the problem we can see that there are essentially two types of knowledge to be considered. The first might be termed imaginary or fantastic knowledge. Middle Earth, Hobbits, the reason Frodo went to Mordor, etc., are all part of the knowledge base of human society. Yet this knowledge has no direct relationship to material reality (although there may be, obviously, an allegorical or metaphorical connection). The utility of such knowledge is determined by a set of social and cultural criteria and processes the categorization of which are not relevant to the analysis presented here. In the crudest market sense, such knowledge is valid insofar as it has value; i.e., it can be bought and sold in the marketplace. The second type of knowledge can be termed practical knowledge. Drucker (1993) makes a similar distinction between knowledge as self-enlightenment and, from the Greek, techne, or technology. This type of knowledge has a direct referent to the actually existing material world. It is this type of knowledge that is applied to production and which infuses the technical essence of most commodities.

The point here is not to draw a false distinction between useful, value producing, knowledge and imaginary knowledge which does not produce value. Both of these forms of knowledge produce value. The technology of a movie is an important component of its value. But, as in other similar “artistic” commodities, it is the image content that is the core source of its value. Very few people go to a film because Industrial Light and Magic did the special effects. Nor will a film necessarily be popular because a specific director made it. A particular director’s track record might help get some people into the theaters, but if the film does not appeal to the market then it is likely to be unprofitable. A film generates a very large gross income because its informational content appeals to a large segment of the movie-going public at a particular time.

We can begin to see, then, that knowledge is open-ended only if considered in an ahistorical context. In reality, the production of knowledge is historically and socially determined; the “valorization” of knowledge, or its utility, is dependent on its social context in a given historical period. Knowledge which goes un-utilized, which is locked-up, thrown away, or forgotten, is, for all intents and purposes, not knowledge (in the same way that money that is hoarded, not thrown back into circulation, does not really count as capital). To point out that knowledge accretion is potentially unlimited is nice, but not really saying much, for knowledge dis-accretion is potentially unlimited as well. Contrary to what the deconstructionists might claim, there is a hierarchy of texts. Not all information is as useful or important as other information. Some information moves from particularity and specificity to become generalized and abstract, thus becoming part of the generalized social knowledge (whether the knowledge is that of a particular firm or of an entire culture over historical time is not important here). This sort of information/knowledge could be said to have a relatively long shelf-life. Other information is embodied in the materiality of a commodity and lasts only as long as the useful life, whether materially or socially determined, of that particular commodity. Still other information is useful only as information and has a short or long shelf life depending on what it is (today’s news versus a book on political history for example). The specifications of the Intel 8088 microprocessor (used in the original IBM-standard personal computer), for example, are increasingly irrelevant in the Pentium era. Bill Gates, CEO of software producer Microsoft recently claimed that “there’s not a single line of [computer] code here today [at Microsoft] that will have value, say, in four or five years time” (quoted in Stross, 1986:36; see also, Kenney, 1996).

The point of this article then, is to avoid treating knowledge so much as a thing and instead move towards considering it as a process. By treating knowledge in a Marxian methodological context, I hope to avoid the awestruck tone of much of the current discussion and demonstrate that the “new” knowledge economy has deep roots in the historically extant categories of capitalism. The treatment presented here is somewhat abstract and formalistic and is meant not to serve as a complete theorization of the problem. It is, rather, a more modest attempt at providing some foundations for a more reasonable and realistic take on the changing role of knowledge and information in the global capitalist economy. The first section of this article will develop a conception of knowledge-in-process, distinguishing between three basic moments: knowledge, information, and data. The second section then takes up the question of the role of knowledge in determining value, in particular by extending Marx’s notion of the organic composition of capital to include a knowledge component. The third section of the paper then moves from the abstract treatment of the prior two sections to consider some concrete determinations, specifically the computer and Internet information industries, whose growth and development has sparked much of the recent speculation and discussion about the role of knowledge in the economy.


Knowledge and Information

Knowledge is a general abstraction. For example, knowledge is necessary for making a electronic semiconductor chip. In making a chip, knowledge regarding the behavior of electricity, electric circuits, materials physics, manufacturing processes, etc., i.e., general scientific knowledge, is applied to the production process. Such knowledge exists in a form which is non-specific to the unit or process in question. Often, in an exchange context, it is non-proprietary knowledge. Such knowledge is part of the general pool of knowledge available to anyone willing or able to go out and get it. 1 It is not directly controlled by any specific economic interest. A textbook written about physics by an eminent professor may be copyrightable, but most of the knowledge discussed in the book is non-proprietary. The copyrightable, salable form of such knowledge (i.e., the value added) is contained in the presentation, not in the actual, general knowledge.

Information, relative to knowledge and juxtaposed against it, on the other hand, is a determinant abstraction. The knowledge needed to build a chip is distinct from its specific application. Such knowledge is applied to the production of specific information which is then applied to the production of a specific product, e.g., a Pentium chip. The specifications, designs, etc., of that chip are the result of a process of application of general knowledge to a specific problem to produce proprietary information. If that information becomes stabilized over time it then becomes the knowledge of the firm, and may or may not be proprietary. Information is both the means by which this process occurs, and the end result of the process. 2

Capital can be held to be a general abstraction if we regard the capitalist system as our starting point. Commodities, money, labor power, technology, etc., are determinant abstractions within the context of capital as a general abstraction. They are moments within the overall process of the circulation of capital. Knowledge is a general abstraction outside of the direct nexus of capital. When it is subsumed under or within capital, i.e., put to work by capital, or more concretely, utilized for the purpose of producing a given commodity by a particular firm, knowledge becomes information. Information then “splits off” from knowledge becoming an input to production and that part which “remains” as knowledge, sometimes referred to as “tacit” knowledge (see Fransman, 1994) becomes the impulse and the framework for the application of information to production.

Information has some similarities to money; indeed, in its most basic determination money is merely the notional form of value, value as it appears is its most recognizable, interpretable, and utilizable form. Money is information about value. The use value of money is as a carrier of information about the value possessed or controlled by its possessor. Money serves both as a facilitator of exchange, a universal equivalent which eliminates the difficulties which would occur in attempting to exchange equivalent values in their direct commodity form, and a tool or method for accumulation in its own right (i.e., finance capital). In a similar manner information is a tool or facilitator of production and a source of value in its own right. In the daily unfolding of the production process information circulates widely and routinely, without necessarily directly creating value. This is the basic information associated with the production process–where will the meeting be? when? what’s our supplier’s phone number? what’s the invoice number? the part number? , etc. This information is part of the overall organic or technical composition of capital and thus is part of the value equation, albeit a small one. Some of this information (and the process associated with it) is probably superfluous to the actual production process and thus could be considered a drain on value (wasted or unnecessary effort). Information is also sometimes the goal of production (that all commodities can be considered information nodules will be considered in the next section). Newspapers, computer software, repair and operating manuals, books, magazines, etc., are instances where information itself is the product. In some cases the information is as mentioned above: repackaged knowledge. In the case of news sources (newspapers, TV, Radio, etc.) news is information which has been processed and/or repackaged for easier consumption. This is also the case with news sources on the Internet. Computer and information technologies do not produce a fundamental change. The net replaces newsprint as the distribution medium. Firewalls and other barrier technologies make the information a product which must be purchased.

According to Smith (1993; see also Murray, 1988 and Smith, 1990) in his discussion of Marx’s Hegelian legacy, Marx’s approach in Capital, though quite different from that of Hegel, “employs a framework taken from the theory of the syllogism employed in Hegel’s ‘Logic.’

It too explores the mediations connecting universality, particularity, and individuality. In Marx’s account, “Capital” is the moment of universality. From the inner nature of capital a number of distinct structural tendencies can be derived. In Hegelian terms these form the moment of particularity. And finally there are the acts of individual capitalists, individual wage laborers, and so on, whose acts are structured by those particular tendencies and thus also mediated with the inner structure of capital (Smith, 1993:16).


  • Universal: Capital (value-in-process) Particular: Structural Tendencies (M- MOP-/LP- C’- M’)

    Individual: Individual actors (buyers, sellers, workers, firms, etc.).

As it is, knowledge-in-process manifests a series of determinations very similar to the structural tendencies of capital as a whole. The can be represented as follows:

  • Universal: Knowledge Particular: Information

    Individual: Data

To grasp the idea of these three determinations it is helpful to think of a book. Any book (with the exception of blank books, consists of data. In the case of a book written in English, the data consist of discrete markings (letters) arrayed in rows across the pages.3 Hypothetically these data points could be anything (numbers, random symbols, tiny drawings, etc.) and mean anything. If one stands far enough back from a book it is easy to see that there is data in it, but difficult to see the form the data take, or their meaning. It is only with a predetermined system of discrete letters, which form words, which in turn form sentences (to continue our English example) that the “raw” data contained in the book becomes information. Yet, most books are not merely conveyors of information. Immanent within the information are ideas, concepts, references to real events, etc., or, in other words, meaning. While the system of language (or, more generally, the system of code) which could hypothetically be closed, i.e., created by an individual and understandable only by that individual, at this level of abstraction, corresponding to the level of knowledge as a universal determination, the information contained in the book has no existence outside of its context as a socially determined process, knowledge.

Here we arrive at a formulation that, whatever its faults might be, at least moves beyond reifying or mystifying knowledge. In other words, the theory becomes richer by saying that what everyone simply refers to as knowledge or information is in fact a process and a social relation which at least provisionally, for the purposes of economic understanding, can be treated like capital. It is only by treating it this way that we can legitimately arrive at the proposition, which is at least implicitly contained in the some of the literature on the subject, that knowledge = value. Such a proposition is no doubt controversial, but we can at least begin to see that knowledge is an important component of the value form. But for the moment we need only take into account that knowledge-in-process is an important component of capital taken as a whole. Furthermore, just as an individual commodity embodies the different moments of capital (machine mediated value, direct labor value, etc.), so too can it be said to embody different moments of the knowledge process: knowledge, information, and data. The question of whether knowledge is value, or value is knowledge, will be taken up again in the second section of this paper.

Information and the Commodity

A commodity is a moment in the circulation process of value. A commodity is a container of value. It is the material embodiment of dead, congealed, labor, and, by extension, information. A commodity is produced through the application of. physical activity, knowledge, and information. If, for a moment we forget about capitalist commodities, capital, and value, and concentrate on information we can see that all material products of human activity contain, or embody information and/or knowledge. In this sense there is no difference between a spear and a microchip. The knowledge/information content of a microchip may be far greater than that of a spear, but they are both the products of human mental (and physical) activity applied to the production of a useful object. Some, or even most, or even all, of the human physical activity involved in making a chip may be applied to the chip indirectly through automation, but there is human physical activity involved somewhere in the process. The difference lies in the determinant social context in which the information is produced, i.e., the division of labor, the system of exchange, private property, etc. Information sharing is minimal in the primitive context, it increases vastly in the capitalist context. The spear maker’s information is his personal possession, it is passed on directly to other spear makers. Information and knowledge become abstract in the capital context. Hypothetically, I need not apprentice myself to another chip maker to make a chip–I could get the basic information at a library. The capital, actual skill, markets, etc., is another matter. With the development of writing, information (and language) becomes abstracted (alienated) from its direct dependence on the human brain.

With the advent of abstract information, information itself can then become disembodied use value, and, under capitalism, a capitalist commodity. Books, maps, and other implementations of abstract information are commodities whose use value is their information content. (Whether their content is itself useful is not relevant here–for our purposes the information in a novel is no different than the information in a physics textbook.) The historical change does not occur in the information, but in the process of its dissemination. Books were rare and valuable things when they had to be copied by hand. The printing press made them more common. Then came mass reproduction, the paperback book, etc. There is a continuum form book to computer. In bit form the book can be almost instantly copied from one memory to memory, disk to disk, memory to disk, or disk to memory. It can then be printed easily, though not necessarily cost effectively. Computers and networks are part of a continuous process of improvement of the means of transport and communication over time. In this sense they are nothing more than the latest advance in a long line of advances. Thus the question regards not a change in the nature of information itself, but a possible fundamental change in the economics of information.

There is one possible exception to all that: computer software. It is reasonable to divide computer software into two general categories for the purposes of this discussion. The first is operating software–which should not be confused with operating system software, which is a form of operating software. By operating software I mean the code which consists of the instructions which in effect tell the computer what to do. Examples of this are the ROM BIOS, the operating system, communications protocols, and the various application programs which make computers do useful things such as word processors, spreadsheets, browsers, etc. Once a service performed on-site, now, with the mass production of microcomputers, it is a product boxed and shipped to the consumer (and now, with the Internet, sent directly to the consumer over a wire, as with Netscape). In a sense this information is a part of the machine. It is another component or module (one, however, that can be copied infinitely and frictionlessly). This copyability leads both a new type of business/distribution model as in the case of Netscape and Moore’s business ecosystems and a new set of problems/challenges as regards proprietary control. In the case of Netscape we see a very rapid series of upgrades disseminated over the net as soon as they are ready (or, in the case of beta versions, before they are ready). However, one would assume that as browsers become bigger and more complex–if this will be the case–that the rate of change will slow. Despite the easily changeable nature of bits and bytes it is probably the case that major upgrades of software technology generally occur no more rapidly than major changes in hardware (due both to the difficulty inherent in software engineering and the costs of coordinating marketing and distribution. And operating software in its compiled form is not easily changeable by the consumer.

The second category is data. For example a multimedia program like Encarta consists of an operating component which provides the interface, a graphics engine, a database/search engine, etc. It also contains a large amount of data: pictures, text, sound files, etc., most of which is outsourced from book publishers like Funk and Wagnels, Rand McNally, or whatever. In many cases, multimedia encyclopedias and other informational titles are simple repackaged information from other sources. Indeed, one of the most often made criticisms of many multimedia offerings is that they consist of lousy interfaces wrapped around already existing information (very little new value added). Microsoft’s long term strategy has been to wrap high quality interfaces around high quality information. Many of their acquisitions in the past few years have been in this direction–of both content producing companies and the rights to content (e.g., the photographs of Ansel Adams, the Betteman Archive, etc.).

In the end we are confronted with the easy copyability of the information commodity and the challenge this presents to private property. At one level, this is primarily a technical problem. The Internet is like an unregulated frontier in which the free flow of information is the norm. However, as usage becomes more widespread, and business interests see the potential of the Internet as a distribution medium, more and more content will come with a price attached. The Wall Street Journal, Microsoft Slate, and many others will soon charge for their content; others, (e.g., the San Jose Mercury., Dataquest, and Netscape), already do. There are already even some content providers in Mexico who are charging for access. Capital will colonize Cyberspace, just as it has colonized everywhere else. Should this surprise anyone? Clever people will find ways around this stuff, but those holes will be plugged and the average person will pony up the cash for the content that they want. Piracy is a problem which will continue to be rampant in the Third World, and will be a problem elsewhere. But it is significant to note that both international content and operating software producers, whether Microsoft or News Corp. are, and are likely to continue to be, extremely profitable. As underdeveloped areas become more developed, piracy will diminish. Piracy of Microsoft software in China, for example, helps create a (potentially huge) legitimate market for Microsoft in the long run.

The Knowledge Economy

Knowledge = Value

The value embodied in a knowledge commodity is primarily a function of its informational, or ideational, content. All commodities have what might be termed a knowledge composition. This consists of the technical knowledge embodied in both the design and production of a commodity, and the ideational content of the commodity which is usually associated with advertising. Historically under capitalism the knowledge composition of all commodities has gradually increased. As science and technology advance more sophisticated ways of producing even simple commodities, like soap, are developed, and more sophisticated, technological commodities, like microchips, are brought to market. Additionally, the ideational content of even the simplest commodities has increased in both scale and scope. A hamburger is no longer just a hamburger, but a McDonald’s hamburger, or a Burger King hamburger. The consumer purchases not only the actual food, but the ideational content of the product created through the marketing process which includes advertising. While generally thought to be irrelevant to the value composition of commodities (or parasitic on the process), this aspect of commodities is an important part of the valorization process and hence is part of the value of the product (O’Connor, 1984). The relative increase in the importance and intensity of this ideational or symbolic aspect of commodities has lead some critics to argue that it must be placed at the center of any theory of contemporary capitalism (e.g., Baudrillard, 1988).

The whole question of value is controversial; the main criticism revolving around what is regards as the Marx’s arbitrary choice of labor as its basis. But, according to Mohun (1983) Marx’s “choice” results from the fact that the value form is manifested in exchange, which is measurable in terms of the abstract labor time embodied in a commodity. “As commodities . . . products have simultaneously a natural form and a value form. But the latter only appears when one commodity exchanges for another” (Mohun). Value in Marx emerges as a sort of meta-abstraction in a similar way that capital is an abstraction which is made up of the unity of its many diverse determinations. According to Marx “exchange-value is only an ‘appearance-form,’ an independent mode of manifestation of the value which is contained in the commodity” (quoted in Mohun, 1983:507). But value is not an overarching abstraction like capital, rather it is a specific abstract determination which is necessary to conceptualize the more concrete exchange of equivalents, derivation of surplus, etc. “Value is not a technical relation but a social relation between people which assumes a particular form under capitalism, and hence appears as a property of that form” (Mohun, 1983:507).

Value is thus a slippery concept, but one that, it seems to me, is amenable to a “reconceptualization” or, perhaps more accurately, a shift in one’s emphasis on one or another of its different determinations. Marx’s choice was not mistaken nor arbitrary, nor even historically bound; I cannot walk into a store with my knowledge and “buy” something with it. Only value in its capital form (or more immediately, its money form) makes the world go around. But there is no reason why we cannot identify (and emphasize) knowledge as another determinant of value, one that has become increasingly important as the organic composition of capital has steadily increased on a global basis.

Using the commodity as our point of reference, we can see it as a continuum; the spear and the chip again. All useful objects contain a knowledge component. Even a rock, unchanged by human hands contains a knowledge component when it is picked up and thrown at somebody. (Remember, not all use values are derived from labor–but they find their expression in labor, i.e., purposeful creative activity). The interesting part comes when the economy is sufficiently developed to the point where knowledge attains a certain autonomy in the process (like finance capital). The production of “pure” knowledge has always existed, e.g., art, literature, scientific knowledge, philosophy, etc. A book is a “pure” knowledge product, i.e., a commodity for which information content is its primary use value. Thus we are not talking about a “break” in the postindustrial sense. Book publishers, newspapers, movie companies, etc., produce knowledge/information commodities where the real value of the commodity, the knowledge, or the information content, is relatively autonomous from its material form, i.e., paper, film, magnetic media, etc. The form, e.g., the filmstock (the means of communication), catering, setbuilding, etc., are sources of value in their own right. The various contractors associated with a film create surplus value, measurable in the quantitative sense, by doing this work. But there is no value without circulation; the value-form, for it to have any meaning under capitalism, must be consummated through the act of exchange. As mentioned before, people do not generally go to a film to see the work of a particular carpenter. The actors, writers, directors, and other “creative” workers (the “talent” as they are referred to in the movie industry), who are directly involved in the production of the ideational content of a movie generally make the most money. A major portion of the budget of a film goes to these people. Consumers are not directly buying the services of film technicians. The vast majority of the value of a particular knowledge-content commodity comes from the content, i.e., Speilberg’s or Lucas’ idea.

In technology-based manufacturing the real change comes (i.e., the dialectical shift from quantity to quality–the discontinuity within the process) from a shift in the division of labor where the production of “pure” knowledge is increasingly undertaken autonomously by individual capitals. Put another way, knowledge production is “spun-off” from the overall production process to be undertaken by specialist firms. Prior to the present period, this sort of autonomous knowledge production was undertaken primarily by universities, government bureaus, etc., outside of the direct nexus of capital (surplus value production). Like large-scale fixed capital (infrastructure) this sort of production, because of its slowness, and indeterminantness, shows up as a bottleneck to the circulation process. Hence it is “shifted” to the “periphery” of the circuit of capital. With the rise of information technologies, however, this sort of production, or a variation of it, is increasingly a direct source of measurable value in its own right. Microsoft, Netscape, et al., epitomize this. Privatization of weather services and other data and information services are another determination.

The Knowledge Composition of Capital

Marx developed the concept of the organic composition of capital in part to help to comprehend the implications of technological development in the capitalist economy. In Marx’s conception the organic composition of capital consists of two determinations: its value composition and its technical composition. The value “side . . . is determined by the proportion in which it is divided into constant capital or value of the means of production, and variable capital or value of labor-power” (the ratio of variable to constant capital c/v), and the material (or technical) side “is determined by the relation of the mass of the means of production employed, on the one hand, and the mass of the labor necessary for their employment on the other” (Marx, 1967:612). The notion is conceptualized this way in order to allow for the possibility that the value composition can change for reasons that that are not directly determined by the technical composition. According to Marx the two are directly related; the organic composition of capital is essentially the value composition of capital “in so far as it determined by its technical composition and mirrors [its] changes” (Marx, 1967:612). In other words the value of, say, the inputs required by a particular firm to engage in production are not necessarily directly determined by the technical composition of the other firms which produce them. Any number of factors, all of which fall under the general logic of demand and supply, can effect the value (or more immediately, the price) of those inputs (Harvey, 1982).

For our purposes we need not worry directly about the value side. Rather we want to concentrate on the technical side and think of the organic composition of capital as a general description of the technological development of a given firm, region, industry, state, etc. When we talk about an economic formation with a high organic composition of capital we mean to say a large amount constant capital, that is, machinery, technology, organization, etc., is being employed relative to a smaller amount of labor-power, that is, actual living physical, or mental, labor. This does not imply that historically, nor logically does it follow, that in economies with high organic compositions of capital, technology puts more and more people out of work. As the composition of capital of an economy as a whole increases, certain functions are rendered obsolete and unnecessary, while other entirely new functions are created. Forty years ago there was no chip industry to speak of. Computer functions were performed by huge machines which required large amounts of living labor for their construction, operation, and maintenance. Now, the basic functions of a computer are integrated onto a single silicon chip. These chips are produced by firms which utilize vast accumulations of technological constant capital, including other computers, and relatively little direct living labor. Much of this living labor consists of highly-skilled jobs which didn’t exist until recently.

It is clear that the technical composition of capital, considered in the broadest sense, has a knowledge component. Constant capital in this sense consists of human knowledge deployed in the form of physical material machinery, the general level of the economic organization of society, and the organization of the production process, whether in an individual firm, or among firms involved in a particular production process. An increase in the value composition of capital of a particular society undertaking capitalist production, i.e., an increase in a nation’s wealth, generally, though not necessarily, means an increase in the society’s overall technical composition of capital. A nation whose primary source of economic value is derived from natural resource extraction, for example, an oil producing nation, builds up only minimal technical capital. A nation’s capitalist wealth, i.e., its value-in-process, is a function of its constant capital base, which has as a major component, its knowledge base.

Yet again, we must not go to far with this and reify knowledge by saying that the wealth of nations is equivalent to the knowledge of nations. We are not talking about the organic composition of knowledge here. Rather we are talking about the knowledge composition of capital. This can be expressed in the following way:

Such a formulation expresses the opposition between the knowledge embodied in a machine, or the organization, or what might be termed the technical material field generally, and the knowledge possessed by the worker and applied directly in the production process. Contrary to Guilder (1989), who claims that the increased importance of knowledge in the economy makes the worker who possesses technical knowledge a new form of capitalist–since he controls his own means of production–the fundamental relation between labor and capital is the same as before. The knowledge component of variable capital today may be far more sophisticated than that in the past, but knowledge embodied in the fixed capital that it confronts in the production process is itself vastly more sophisticated. Without the knowledge possessed and controlled by the firm, and embodied in the organization of the firm, the knowledge possessed by most “knowledge workers” is useless.

Yet we cannot ignore the fact that while the composition of capital increases (i.e., the ratio c/v), the component v, conceived here as knowledge increases as well. While a certain number of functions (particularly many service jobs) posses relatively low compositions of capital, more and more others require workers with higher levels of training and education. On a general level the increased composition of capital can be conceived of as an increased density, velocity, and complexity of the social relationships and processes which make up social production under capitalism. This is what we are talking about, in the most basic terms, when we say that we are in an age of information, an information society, or a knowledge economy.

Knowledge and Industry

Knowledge-Based Business

Up until now, we have treated knowledge on a mainly abstract level. The real proof in the pudding comes for all this comes from the existence of a new production model (but not new in a post-capitalist sense). Increasingly we see the both the commodification of knowledge and information that was once outside the direct valorization process. Data of various kinds are more and more marketable commodity. Weather, geologic, and agricultural data, economic statistics, etc., once primarily collected and/or disseminated by government or other not-for-profit entities are all increasingly privatized. Alongside this has been the industrialization of knowledge production. Computer software production epitomizes this change. Software is no longer done in-house or as a direct service, but is “spun-off” as a source of accumulation in its own right. The production process undertaken by a software producer like Microsoft is completely different from the Fordist model; the hourly wage relation and individually segmented division of labor is replaced by salary and stock options and team-oriented production. As computer software becomes both more complex and more of a mass marketed commodity, the knowledge workers who build it cease to be craftspeople who put their individual creativity into their work. Instead, as members of large teams, they are assigned to build small pieces of the whole. The interfirm environment is epitomized by the venture capital model, companies are spun-off to develop new product ideas, and subcontracting and development partnerships become the norm. Building software technology begins to take on an economic logic similar to book publishing:

From an economic perspective the software industry is quite unique. Never in the history of the industrialized world has a single industry offered more powerful economies of scale. The majority of the product costs are fixed (R&D). Therefore, per unit costs fall quickly and dramatically as unit volumes increase. In this world of potentially unlimited scale, market share is critical. If a company can achieve greater than 50% market share, and is willing to lower price, it can become somewhat invincible (Gurley, 1996).

According to Gurley, such concepts are entirely missed by most software executives, but are “stunningly obvious” to Bill Gates and a few others (Netscape and Intuit, for example). A company like Microsoft creates a system in which producing knowledge (building software) can be done rapidly and on a large scale. In a situation which is similar to many corporations that have come before, with a decent product (not necessarily the best) and aggressive pricing they soon have the dominant market share, and the fixed costs of software R&D (or, simply, software production, for Microsoft’s production is R&D and its R&D is its production) for its competitors become too large to bear, and so, in Gurley’s words, its competitors’ boards better start looking for a new business plan.

Large-scale development and dissemination of sophisticated information technology has constantly transformed parts of the economy outside the direct nexus of information technology production. One of the results of the “information revolution” is not only an explosion of new products, but whole new business model types, modes of organization, etc. One case in point is retail. The business models of “Superstores” like Wal-Mart and “category killers” like Home Depot, and Borders are predicated on the existence of sophisticated enabling information technology. None of these operations, with their huge and varied inventories, would be profitable without the development and dissemination of powerful and economical computer-based inventory tracking and ordering systems. The borders bookstore chain is a case in point. The Borders chain, started by two brothers in Michigan, one a mathematician and the other an English professor, is part of a new phenomenon in the bookselling business. Several years ago the general consensus seemed to be that comprehensive, independent bookstores were on the way out–soon to be replaced by limited-selection,discounting centers like Crown Books. What has instead happened, with the application of enabling information technology, is the rise of large inventory chains and other category killer-type bookselling business models. Essentially what Borders did was leverage a unique inventory-control system that the brothers developed. They tried to sell it in a consultancy-type manner but got little interest so they used it to start their own chain. “The IBM-mainframe system streamlined the ordering and restocking process, and allowed borders to determine instantaneously which titles and subjects were hot, which were seasonal, and so on” (Goodrich, 1995). The typical Borders store has 125,000 titles, and the company plans to be operating 110 stores by the end of the year (1995). Recently this model has been brought forward even more by Amazon.Com, an Internet based bookseller located in Seattle. Amazon.Com has what might be called a virtual inventory of over a million individual titles. A customer need never know weather a book they ordered was actually in stock at Amazon.Com’s own warehouse, nor do they need to special order it if it isn’t; if not in stock the book is automatically ordered and shipped as soon as it is received.

This transformation has created new systems of production and distribution organization which require a more knowledgeable worker in some parts of the system. But the fact that knowledge is also increasingly embodied in the machine should not be overlooked. At a superstore similar to Borders that I have visited, the personnel seem knowledgeable and interested in books. But the real success of the company is predicated on the knowledge embodied in the inventory system, and in the knowledge workers that develop and maintain it. The in-store workers seem to be essentially hosts to the system, performing the necessary routine tasks associated with maintaining the actual physical inventory and conducting transactions with customers. Ask them for help which requires them to move outside of their company’s system, for example, to make a special order of a book that is not stocked by their company’s own central warehouse, and they have problems (they are not permitted to make such orders).

The Knowledge Production Process

The development of the means of communication have increased the rapidity in which information and knowledge is disseminated and exchanged, which in turn, has reduced the “shelf life” of information. Weekly magazines like Time and Newsweek loose usefulness as news sources in the face of daily newspapers, which are more up to date. Daily newspapers loose their effectiveness as news sources in the face of radio and TV, which also are more up to date. The newsweeklies must then find a niche: “analysis,” lifestyles, context, reviews, special graphics, etc.

As the information content of commodities increases the necessity for the rapid exchange of information increases, which in turn leads to the development of more rapid means of communication, which in turn enables the further increase in the information content of commodities (and services). The increase in the information content of commodities forms an identity with the increase of the role of information in the production process. In an analogy with capital, an increase in the general rate of turnover implies that if capital is withdrawn it will devalue more rapidly (or by standing still, produce nothing faster than in a situation with a slow rate of turnover). The same thing happens with information. If Intel develops the P6 (Pentium Pro) but decides to withhold that information from the production process for a year, that might be okay, given Intel’s domination of their market. But since the velocity of information in their industry is so high, they are taking a tremendous risk, especially in the face of their main competition in this segment (no pun intended): RISC processor manufacturers who are pursuing their own information-intensive strategy.

The real problems come from the exigencies of the competitive market and the need permanent innovation, which is the same thing as the need to constantly generate new knowledge and information. Microsoft is a company which has developed a very fast moving, learning-by-doing, perpetually-innovating organization for the production of software (see Cusumano and Selby, 1995; Manes and Andrews, 1993; Moody, 1995; Wallace and Erickson, 1992). Like Intel and other technology leaders they are always working ahead of themselves. As soon as they finish one version of an application or system they are already working on the next. Some code is reused, but the competitive advantages come from increased functionality, i.e., new innovations. The software does not wear out like a metal machine–however it becomes obsolete as competitors introduce products or versions with increased functionality, efficiency, etc. also, as machines change. Thus while WordPerfect 5.1 and MS-DOS 6 will potentially exist forever in some form, and some people will use them for many years to come, they will increasingly loose ground to products with more functionality, and increasingly become incompatible with the products which are constantly being upgraded and reintroduced. A hermit with an old 8088 and DOS and WordPerfect will have no problem typing away in his cave–but if he wants to interact with the rest of the world, and/or if he wants increased functionality with nicer features, he will eventually have to upgrade. Thus it would seem that the permanence of bits and bytes is not that significant–just as the permanence of silicon is not that important (is it an accounting problem? like “slippage” in inventories?). Also, its interesting to consider the context in which people copy software–will the millions of new home computer users have access to the social networks necessary to obtained pilfered software? To stay on top an either industry you must innovate, innovate, innovate.

Another area of discussion concerns the software worker. Microsoft uses an iterative process to “build” software in a process that uses hundreds of people working endless hours developing, testing, building, debugging, planning, etc. Indeed, the process is never really finished. It is the case that perfect, error-free, bug-free code is never released. The decision to “freeze” a project build, to decide that the code is “golden,” ready to be released to the outside world, is reached after considering a number of factors, market pressures, internal planning schedules, etc., not limited to technological ones. The interesting point here is not that bits and bytes are permanent, or that the software product is almost completely dephysicalized, but that the service of programming computers is accomplished on a mass basis through the distribution of the programming code on magnetic media or CD-ROM. That the service is boxed and put on a shelf is only significant in two aspects. First the producer is selling a style, a “look and feel,” which is analogous to the image-content of many products in the modern, late capitalist era. This is increasingly important with the introduction if integrated software suites–and of course with the hype about systems like Windows 95. Microsoft represents which might be the paradigmatic knowledge-based corporation insofar as its production organization is a form of Taylorism applied to knowledge production.

What is different about the new knowledge worker? At Microsoft the product is code. Coders (developers) are not creative, think-workers in the way that we traditionally think about them (i.e., college professors, or writers, or film directors, or musicians). They use a pre-defined set of tools to build software components which are hooked-up to other components to create the whole product. There is a division of labor at Microsoft. The project specs are set at the beginning by the managers. The developers build the code, the testers test it, the developers de-bug and redevelop the code, its re-tested, again and again, around and around. It might be useful to think of the software developer as a machinist of the information age. In the same way that hundreds of workers build lots of little pieces and eventually you have a whole 747, software builders build lots of little pieces and eventually you have Windows NT. Ultimately it is hard to think of these people as knowledge workers–brain workers, information jockeys, or information mechanics, maybe, but not knowledge workers.

The important thing that needs to be considered in relation to the new knowledge worker (or production) is not that knowledge and information are more important in the production-labor-commodity process, but rather, what is the nature of the work of the new knowledge worker? How knowledgeable are they? Are they really engaged in the creative application of their own knowledge to the production of something? Or is the knowledge work they do really analogous to a skilled trade, and the information they possess and work with and produce the alienated raw material of information capitalism? The technical, or knowledge, composition of capital is higher, and there are marginally different social relations (processes) on the “shop floor,” but is the new knowledge worker really all that different from the old skilled machinist? Can we not conceive of the new knowledge work/production as the melding of Taylorism with computers? Or as one observer commenting on theories of the information society puts it:

The whole development of the twentieth-century state and society can therefore be regarded as the application of the principles of scientific management. Information, knowledge, and science–including social science–are self-evidently the central requirements of this process. They provide the means necessary to coordinate and control the increasingly complex operations of the economy and the polity. Thus it can be argued that ‘it was the exponents of Scientific Management, in its broadest sense, who unleashed an Information Revolution’ (Kumar, 1995:33; see also De Landa, 1991).

Such processes also create the complex operations they set out to control. The astounding complexity and the sheer size of an operation like Home Depot is enabled by information technology. The gigantic personal computer operating system, Windows NT, containing over 3,000,000 lines of code, was developed to take full advantage of highly complex 32 bit Intel microprocessors (Zachary, 1994). Certainly it behooves workers to develop information-based skills in order to find gainful employment within capitalism’s current informational manifistation. But is the new information-intensive economic logic, i.e., new information-based business models, radically transforming the core capital/labor nexus, as according to the information age’s most fervent boosters? Or are we really just seeing the intensification and extension of trends which have been well underway, and somewhat well understood, since the Nineteenth Century?


Baudrillard, J. (1988) “For a critique of the political economy of the sign.” in Mark Poster (ed.) Jean Baudrillard, Selected Writings, Palo Alto, CA: Stanford University Press.

Bell, D. (1976) The Coming of Post-Industrial Society : A Venture in Social Forecasting, New York: Basic Books.

Block. F. (1990) Postindustrial Possibilities: a Critique of Economic Discourse. Berkeley: University of California Press.

Cusumano, M. A. and R. W. Selby (1995) Microsoft Secrets: How the World’s Most Powerful Software Company Creates Technology, Shapes Markets, and Manages People. New York: The Free Press.

De Landa, M. (1991) War in the Age of Intelligent Machines. New York: Swerve Editions.

Drucker, P. F. (1993) Post-Capitalist Society. New York: Harper Business.

Ellis, J. M. (1989) Against Deconstruction. Princeton NJ: Princeton University Press.

Fransman, M. (1994) “Information, knowledge, vision and theories of the firm,” Industrial and Corporate Change 3.3: 713-757.

Gilder, G. F. (1989) Microcosm. New York: Touchstone.

Goodrich, C. (1995) “The superstore phenomenon,” The Los Angeles Times Book Review (May 28,):4-9.

Gurley, J. W. (1996) “Stunningly obvious: the secrets of software economics.” Above the Crowd (newsletter). CS First Boston Corporation.

Harvey, D. (1982) The Limits to Capital. Chicago: The University of Chicago Press.

Jameson, F. (1991) Postmodernism, or, the Cultural Logic of Late Capitalism. Durham NC: Duke University Press.

Kenney, M. (1996) “New Knowledge, Old Knowledge, and Obsolescence,” unpublished manuscript, University of California, Davis.

Kumar, K. (1995) From Post-industrial to Post-Modern Society: New Theories of the Contemporary World. Oxford UK: Blackwell.

Mandel, E. (1975) Late Capitalism, London: Verso.

Manes, S. and P. Andrews (1993) Gates: How Microsoft’s Mogul Reinvented an Industry–and Made Himself the Richest Man in America. New York: Touchstone.

Marx, K. (1967) Capital, Vol. I. New York: International Publishers.

Mitchell, W. J. (1995) City of Bits: Space, Place, and the Infobahn. Cambridge MA: MIT Press.

Mohun, M. (1983) “Value” pp. 507-511 in T. Bottomore et al. (eds), A Dictionary of Marxist Thought. Cambridge MA: Harvard University Press.

Moody, F. (1995) I Sing the Body Electric: A Year With Microsoft on the Multimedia Frontier. New York: Viking.

Murray, P. (1988) Marx’s Theory of Scientific Knowledge. Atlantic Highlands, NJ: Humanities Press International.

Negroponte, N. (1995) Being Digital. New York: Vintage Books.

Nonaka, I. and H. Takeuchi (1995) The Knowledge Creating Company: How Japanese Companies Create the Dynamics of Innovation. New York: Oxford University Press.

O’Connor, J. (1984) Accumulation Crisis. New York: Basil Blackwell.

Smith, T. (1990) The Logic of Marx’s Capital, Replies to Hegelian Criticisms. Albany: State University of New York Press.

Smith, T. (1993) Dialectical Social Theory and its Critics: From Hegel to Analytical Marxism and Postmodernism. Albany: State University of New York Press.

Stallabrass, J. (1995) “Empowering Technology: The Exploration of Cyberspace.” New Left Review 211 (May-June).

Stross, R. E. (1996) The Microsoft Way: The Real Story of How the Company Outsmarts the Competition. New York: Addison-Wesley.

Toffler, A. (1991) Third Wave, New York: Bantam Books (reissue).

Wallace, J. and J. Erickson (1992) Hard Drive: Bill Gates and the Making of the Microsoft Empire. New York: Harper Business.

Zachary, G. P. (1994) Showstopper: The Breakneck Race to Create Windows NT and the Next Generation at Microsoft. New York: The Free Press.


1. There are, of course, economic considerations involved in education, and autodidacticism is difficult and rare in practice. The distinction posed here is categorical.

2. It should be noted that it is a mistake to map knowledge/non-proprietary, information/proprietary in strict terms.

3. I will ignore pictures, graphs, tables, and other non-standard forms of representation for purposes of simplicity and brevity.

Copyright 1997 Electronic Journal of Sociology