ON THE VALUE-NEUTRALITY OF TECHNOLOGICAL OBJECTS Russell James Woodruff I begin with an description of neutralism, followed by a critique. Then I describe and assess the various non-neutralist positions. I end with some comments asking for help in increasing my understanding of these matters. 1. NEUTRALISM Neutralism regarding technological objects (artifacts) is expressed in two claims, each having a causal and a normative variant. There is the causal claim that the creation and existence of an artifact does not necessitate its use, and there is the claim that artifacts have multiple uses, none of which are dictated by the object. Each of these claims can be made normative by placing the phrase "good or evil" somewhere within them. Thus: "the creation and existence of an artifact does not necessitate its use for good or evil," and "artifacts have multiple uses, both good and evil, none of which is dictated by the artifact." The basic idea is that the creation and existence of an artifact is not sufficient in itself to produce favorable or unfavorable differences in people's lives. Other events are necessary in order for the technology to make any favorable or unfavorable differences, and--this is key--these other events are not conditioned, required or determined in any way by the technology itself. This position has considerable initial plausibility. However, it focuses on two issues concerning artifact use while neglecting a third. This neglect may arise from a failure to notice the ambiguity of the term "use." "Use" can mean the purpose to which an artifact is put (what it is used for), the act of using an artifact (that it is used), and the method of operation of an artifact (how it is used). I will first discuss the claim that artifacts can serve multiple purposes, particularly purposes with contrary values. This seems to be a central pillar of attempts to support artifact neutralism. Mario Bunge describes the neutrality of artifacts thus: Most industrial products are morally neutral, in the sense that they can be used for good or for evil. A knife may be used for cutting loaves of bread or throats; a powerful drug to cure or to kill. (1984, p. 310.) In assessing such claims we should first distinguish the purposes of people from the purposes of things. Both people and things have purposes, but it seems that they have them in different ways. When I have a purpose, what I have is an intentional state reflecting some goal that I want to bring about. When an artifact has a purpose, what it has is a function that it more or less adequately performs. In this regard, consider the knife mentioned by Bunge. In terms of function, it has one purpose: to cut things. In turn humans can use the knife for alternate purposes (goals): to kill people, make sandwiches, carve wooden figurines. Note that at the level of artifact function, the neutralist claim about an artifact having a range of purposes is false. The knife has only one purpose (function); indeed, even among knives there is widespread specialization of function. The knife for cutting bread has quite different features than the knife for carving wooden figurines. If we examine artifacts with respect to human purposes, what we find on the one hand are some artifacts with quite limited ranges of purposes. Washing machines, bulletproof vests, B-1 bombers, and wool topcoats are good examples of single-purpose artifacts. On the other hand, we also find things like hammers, canoes, numerically-controlled machine tools, and electrical power networks: all multi-purpose artifacts. Thus, a universal generalization that all artifacts are flexible with respect to their purposes cannot be maintained. In some cases--bombers, bulletproof vests, and topcoats--there is only one purpose served by using the artifact to perform its function. In such cases, contrary to neutralism, the purposes served are "inherent in the technology." But it must be granted that there are many artifacts--canoes, hammers, machine tools, electric power grids, and so on--that can serve more than one purpose when used to perform their functions. The question of the neutrality of artifacts in terms of their flexibility with respect to purposes must be addressed on a case-by-case basis. In the case of functions, however, artifacts are never neutral. Every artifact has a specific set of functions built into its designed features. Turning now to claims about another sense of "use," that of "the act of using," neutralists say that the fabrication of an artifact does not necessitate any future event, specifically the event of using that artifact to achieve some purpose. Emmanuel Mesthene writes, "There is nothing in the nature or fact of a new tool, of course, that requires its use" (1968, p. 135). "The tools by themselves do nothing," states Joseph Pitt. The idea is that we can always choose not to use an artifact, and so avoid having our actions determined or influenced by it. While it is true that the fabrication of an artifact does not necessitate the use of that artifact, we must remember the real-life context of fabrication. Humans purposefully fabricate artifacts in order to extend their capabilities to manipulate material and information. Ultimately motivated by the fact that humans find their needs insufficiently met by their unmediated interaction with nature, humans are in a position where they must make and use artifacts in order to live. So while it is true that there is no causal necessity whereby the making of a particular artifact "pushes" the using of it, there is a practical necessity that "pulls" us to use artifacts, and thus to make them. Although we may sometimes be free to choose not to use a particular artifact, we are never free to choose to use no artifacts. The idea that artifacts are causally passive may remind us that "ultimately" it is "we" who are in control. But saying that we can decide if we want to interact with particular artifacts says nothing about the structure of those human-artifact interactions that we do undertake, and, since we must undertake some such interactions, this argument for neutrality fails us precisely when we need to understand if and how artifacts determine our interactions with them. This brings me to the third sense of "use," "use" as "method of operation." The point to observe here is that if an artifact is to be used, the using must take a particular form. All artifacts, even simple ones like screwdrivers and canoes, impose methods of operation on all who would use them. Whether you are a trader or a pirate there is, roughly speaking, only one way to operate a canoe. The artifact dictates a method of operation and the person who would successfully use it, for any purpose, must see that her using adheres to the dictated method. The determination by an artifact of the method of its use presents itself to would-be users as a set of physical and mental requirements that users must meet in order to get the artifact to perform its function. I see three forms that these requirements take: strength, attention, and skill. It is easy to see that these requirements are imposed by artifacts when we look at cases where the artifacts we use change. Regarding attentional requirements, Susumu Sato found that, for systems engineers using VDTs in their work, both the fixation area and the velocity of eye movements were about 2.5 times higher than those in other kinds of work without VDTs (p. 197). The visual requirements of the work had changed with the change in artifact. H. Finne (1984) found that, compared to manual design, computer-aided design (CAD) placed increased demands on designers' attention in four ways: (1) drawing technique is less seated in the hand, requiring more thought; (2) line and measure are generated simultaneously on the computer, thus inaccurate drawing equals inaccurate measurement, hence greater attention to accuracy is required; (3) one must attend constantly to messages on screen informing of possible errors, guides to data entry, and so on; and (4) a drastically reduced time lapse between decisions calls for continuous decision-making. With regard to skill requirements, studies of changes in the artifacts used in creating manuscripts, in biscuit making, in newspaper production, and in pulp and papermaking all reveal what may seem obvious: different artifacts require different skills (Buchanan and Boddy, Patrickson, Zuboff). Focusing for a moment on the computer, a study of the change in pulp and paper production when it becomes computerized led Shoshanna Zuboff to claim that the skills required have shifted from being "action-centered" to "intellective." Action-centered skills are based on sentient information derived from physical cues, are developed in physical performance, have meaning only in the context of associated physical activities, and are located in the body of the individual person (p. 61). Intellective skills involve abstraction, explicit inference, and procedural reasoning (p. 76). Interaction between tool and user is now mediated by symbols from which meaning must be constructed, using the symbols to create a model of the material process in the mill. In sum, artifacts impose conditions on their users, and thus are not causally neutral. Moreover, we can see they are not value- neutral by recognizing that using artifacts can in itself have positive or negative value. Consider the experience of using video display terminals (VDTs) in developing a new computer program. Sato found that both the velocity and fixation area of eye movements were 2.5 times higher than those in work without VDTs. Instrumental advantages aside, the act itself of programming with VDTs may be intrinsically less satisfying than programming without VDTs. While creating an additional visual source of information about one's work undeniably has its benefits, there is a downside. The fundamental requirement imposed by a visual source of information is that you have to look at it, and looking is not cost-free. This feature of VDT use may explain the fact, noted and puzzled about by Frank J. Landy and others (1987), that complaints of stress by VDT users have not disappeared despite the resolution of many of the ergonomic problems associated with early VDT use. Finne reported that the main difference between manual design and computer-aided design (CAD) is an increased demand on the designer's attention. He sees this increased attentional demand as generally intrinsically disvaluable: "A high level of attention over a longer period of time will result in an overload of a person's psycho-cognitive resources," showing up as fatigue, mental void or general stress (p. 66). Turning to skill requirements, many empirical studies provide evidence that use of skill is intrinsically satisfying (Cooper, 1973; Helpingstine, et al, 1981; Humphrys and O'Brien, 1986; Lawler and Hall, 1970). Skill variety is a standard category among the task characteristics seen by occupational psychologists as relevant to the intrinsic satisfactoriness of tasks. Since artifacts impose skill requirements, including requirements for a variety of skills, and to the extent that the use of these required skills is directly beneficial or detrimental to the agent, the artifact is a causal condition of something valuable, and is so independently of the purposes to which it is put, independently of the further consequences of its use, and independently of human intentions regarding its use and the consequences of its use. Thus artifacts are not value neutral. 2. ANTI-NEUTRALISM There is no one position or argument attacking neutralism. Rather, I find three distinct views, united only in opposing neutralism. One position challenges neutralism on the flexibility of artifacts with respect to the purposes they serve. Another perspective claims that artifacts are expressions of value-judgments are thus are not value-neutral. The third position is that specific artifacts require, as necessary conditions of their operation, the existence of specific social relations in the organizations within which they are utilized. The position that artifacts are flexible with regard to purposes, a position directly challenging neutralism on its own terms, was presented before. I will briefly restate my conclusion. Once we distinguish the purposes of artifacts (functions) from those of humans (goals), three points seems to follow: (1) that all artifacts are quite restricted in terms of functions, (2) that some artifacts can serve a wide range of human purposes, and (3) that some artifacts are quite inflexible in terms of what human purposes they can serve. Thus this argument can claim a limited victory over neutralism, requiring that we rein in unqualified claims about the flexibility of artifacts with regard to purposes. If I am correct here, the best approach to questions of "neutrality" regarding purposes is case-by-case investigation. However, this position does not provide a basis for a general thesis about the non-neutrality of technological objects. Turning to the second anti-neutralist position, that artifacts are the embodiments of value-judgments, I do not see how it refutes the thesis of value-neutrality. While what the position states is true, a neutralist could happily embrace everything about it. We must not confuse people's values with the value of things. That an artifact is an expression or embodiment of people's values says little about the value of that artifact for those whose values it embodies, or its value for others. There are two problems for those who would read from people's values to the value of things. In the first place, the embodiment might not be perfectly successful, that is, the values informing the creation of an artifact may not be fully realized in the actual performance of the artifact when deployed. Secondly, and more fundamentally, even if the embodiment of a value is successful, its realization may not be valuable to those who hold that value, for two reasons. On the one hand, people may be mistaken in their valuing of things. What a person values may not actually be of value to him or her, even according to that person's own standards. It is common for people to value something only to find that valued thing when attained does not have quite the value for them that they thought it would. On the other hand, even if people are not mistaken in their valuings, and a thing embodying some person's values does have the value "pointed to" by that person's values, this gives us no reason to suppose that the thing does not have further value or disvalue for that person. The point is the obvious one that the value of a thing can, and usually does, transcend the human values governing the selection of that thing, even when those values are perfectly embodied in that thing. The case of numerically controlled machine tools illustrates some of these problems. David Noble describes the numerical control ("NC") of machine tools thus: The specifications for a part--the information contained in the engineering blueprint--are first broken down into a mathematical representation of the part, then into a mathematical description of the desired path of the cutting tool...and finally into hundreds or thousands of discrete instructions, translated for economy into a numerical code, which is read and translated into electrical signals for the machine controls (p. 111). NC was intended by its designers and their clients as not simply a technology for cutting metals but as a "management system" that would place control over the process firmly in the hands of management, a system in which the machine would dominate the process vis-a-vis the machinist. But the reality of NC use proved quite different from what was intended. On the one hand, the motivating value "centralized control over operations" was only partially realized in the implementation of the artifact. Precision machining, with numerical control or not, requires constant control over a set of continually changing variables. Noble cites historian and General Electric machinist Frank Emspak on the subject: Cutting metal to critical tolerances means maintaining constant control of a continually changing set of stubborn, elusive details. Drills run. End mills walk. Machines creep. Seemingly rigid metal castings become elastic when clamped to be cut, and spring back when released so that a flat cut becomes curved, and holes bored precisely on location move somewhere else. Tungsten carbide cutters imperceptibly wear down, making the size of a critical slot half a thousandth too small. Any change in one of many variables can turn the perfect part you're making into a candidate for a modern sculpture garden, in seconds (p. 245). The machining process defied the formalized, prespecified requirements of fully automatic control, preventing a perfect embodiment of the value underlying the creation and deployment of NC machine tools. As long as constant monitoring and adjusting are required, centralized control can be only imperfectly realized. On the other hand, to the extent that this realization was successful, and control removed from the shop floor to the office, the realization of this value was not without considerable disvalue for those whose values were embodied in the machine. Ironically, NC, which was supposed to signal management's "emancipation from human workers," was found, if left alone by operators, to make errors "with greater authority than anything we are accustomed to" (Western Electric's Edward E. Miller, cited by Noble, p. 235). With operators merely watching the machine "run itself," with production resting solely upon the formal methods of computerized techniques, the result was "chaos." This was because those techniques, however sophisticated in themselves, rested upon a "limited understanding of the cutting process,and therefore, a lack of fully satisfactory algorithms" (p. 221). Knowing only that an artifact embodies, even perfectly embodies, some value tells us little about the value of that artifact even for those whose value it embodies. The claim that technological objects are the products of antecedent value judgments is thus inadequate for determining the value of technological objects. The third position on the non-neutrality of technology concerns the relations between artifacts and the political conditions within the social organizations in which those artifacts are deployed. A number of authors assert that in selecting a particular kind of artifact one is, intentionally or not, simultaneously opting for particular relations of governance and control. Lewis Mumford argues that all technologies are either democratic or authoritarian, according to scale (small vs. monumental), locus of control (individual craftsmen vs. king or technological elite), source of energy (individual person or animal vs. mass work army or inanimate energy source), and knowledge base (concrete, tacit vs. abstract, symbolic). Langdon Winner and Hans Radder independently argue that nuclear weapons and power plants are inherently political artifacts in that they require highly centralized, hierarchical control. The Labor Process Theory of Harry Braverman and others, holds that technological artifacts developed within capitalism are designed to increasingly incorporate the skills and knowledge of workers into their material structures, in order to shift control of the material process of production from the shop floor to the office, reducing the autonomy of the workers and consolidating the power of management. From this perspective, one would see the development of machines like NC machine tools as an instance of a technology which takes the control of the operation away from shop floor personnel, thereby deskilling them, and placing the control in the office of the programmer. But the power to shape technologies is not unlimited. It meets constant resistance from both other humans and material nature. We cannot read from the intentions of designers and their patrons to determine the performance of the machine. Regardless of how powerful some people are, their intentions can be thwarted, and machines can transcend the intentions of their makers. Machines that have been intended to deskill can perform quite differently. There is much empirical evidence to support the claim that the connection between artifacts and organizational and cultural context is much looser, with considerably more freeplay than this anti-neutralist position suggests. With regard to NC machine tools, studies indicate that the occurrence of deskilling is not determined by the artifact but by variables such as plant size, batch size, type of product, and "managerial philosophy." P. J. Scott concluded from four case-studies that highly skilled machinists are still being relied upon in systems using automated machinery. In a study to determine how NC affects the distribution of machining skills, Bryn Jones concludes "there is nothing `inherent' in the hardware of NC or its concept that would allow for the deskilling and control and surveillance assumed by both theorists of the labour process and publicists for NC installation" (p. 198). Instead product and labor markets, organizational structures and trade union positions all act to influence the forms of skill deployment. Jones does not deny that managements are often motivated to increase control through deskilling; he simply points out that "management cannot construct, de novo, the conditions under which labor is to function" (p. 199). Jones finds machinists, particularly in small shops producing small batches, doing their own programming on the same machines as those used in larger shops, where programming is in a separate department and machine operators are thus `deskilled.' T. D. Wall and his colleagues investigated how new technology affected the nature of people's work. They first noted how changes in work organization in a confectionery firm, with the technology remaining the same, produced "much higher levels of autonomy," 22 percent higher performance, and higher morale. A different confectionery firm built a new factory where work was done by self-managing groups of ten. When compared to another factory owned by the same firm and using almost identical technology, "there were clear reductions in indirect costs" in the new plant. These considerations led the authors to this hypothesis: on the one hand, technological objects are primary determinants of task performance characteristics (such as cycle time, task identity, attentional demands, and cognitive and manipulative skill requirements). On the other hand, with regard to task management, that is, matters of planning, execution, time and task allocation, it is not the artifacts but "managerial philosophy, practice, procedure and structure which are of salience--it is a question of social choice" (p. 17). To test this hypothesis, Wall's group looked at firms deploying new technology. If their idea is correct, then implementing new technologies would have little effect upon operator autonomy and control (sociopolitical conditions), unless the new technologies were accompanied by changes in managerial practices. They first examined the introduction of CNCs into a hand tool manufacturing firm, and found that while attempts to increase managerial control over production processes through deskilling and fragmentation of work continue to be an important factor in shaping jobs, there is nothing inherent in the artifacts to necessitate this. In neither case [involving the two CNCs in this plant] was the technology a major determinant of how it was used. If managers wished they could have designed the work in such a way as to give the operators control of programming and editing, and thus made the jobs more interesting (p. 19). Wall's group then turn to a second instance of the introduction of a new set of technological objects: computers, VDTs and printers replacing typewriters in the preparation of texts. The introduction of this set of objects changes the task characteristics. The importance of typing accuracy is reduced, repetitive typing is eliminated, and new skills and knowledge of commands, file structure and storage capacity are required. How these changes affect workers depends on how the technology is implemented. The authors describe three methods of implementation. First, the word-processing technology can be deployed as a direct replacement for the typewriter for all operators. In this case operators generally experience an increase in job interest through the addition of new skills and the elimination of repetitive work. Secondly, the new technology can be introduced to some typists with the remainder working on conventional typewriters. This is accompanied by a division of tasks. Longer work is allocated to word-processor operators, and shorter documents go to conventional typists. For the operators this implementation restricts the range of each individual's work, and shifts responsibility for task allocation to supervisors. They begin to do part of the work for several authors rather than all of the work for one author. Generally the operators involved do not favor such changes because of the decrease in autonomy, task identity, and feedback from authors. The third form of implementation is an extension of the second. A proportion of workers are given VDT terminals that are linked to a more limited number of printers. Here one gets three types of employees: those using conventional typewriters, those using VDTs and keyboards, and those operating keyboards and printers for their own work and that of other VDT operators. This deployment has all the effects of the second, plus it separates the process of word processing from its output, in both space and time. This distancing particularly increases the loss of task identity and feedback. In addition, the VDT printer operators become "employees" of the VDT operators, marking a further loss of autonomy for the former. Wall and his colleagues note that despite their general disvalue for the operators, the second and third forms of implementation are often chosen over the first because it is cheaper to buy a few word processing units and distribute them to a portion of the workforce than to buy units, including printers, for every worker. They conclude that it is the commitment to cost-minimization, rather than the technology involved, which is the primary determinant of the content and control of typing jobs. In this case, more of the same technology could restore the individual worker's range of control and responsibility. From these studies Wall's group concludes: Technology, whether `new' or traditional, has a major impact on task performance but less impact on task management, i.e. on the control of decision-making rights over the day-to-day management of work....Achieving a balance of control over work depends to a large extent on the strategic choices exercised by management. As Buchanan and Boddy state, "The technical factors of a given innovation do not provide sufficient basis for predicting the social, psychological, organisational and economic consequences of its application. Much depends on how the technology is used" (pp. 20-21). Lest this last statement sound like a reaffirmation of the neutralist position, let me restate that, while artifacts do not determine the politics of technological activity (i.e., the relations of governance and control), they do determine the ergonomics of technological action ("task performance characteristics"), and in this way produce favorable and unfavorable differences in people's lives, and hence are not value-neutral or causally neutral. The main point to be drawn from these studies is that the artifacts developed under capitalism are not as tightly linked to a particular form of social organization as some theorists suggest. Artifacts do not demand the hierarchical control of production, or the separation of mental and manual labor. If these elements are present it is due to causes other than the artifacts themselves. Artifacts do not determine the control relations, task distributions, or skill and knowledge distributions among their users. In this respect artifacts are socially and politically neutral. 3. CONCLUDING COMMENTS In trying to understand and assess neutralism and non-neutralism, I have found that both sides tend to talk very abstractly and generally. I think that theorizing about the role and consequences of technology can be greatly benefitted by being integrated with empirical work in history and social science. Too often philosophers and social theorists seem to rely on one or two examples to support and illustrate their pre-conceived positions. I confess that I am probably no different here. This leads me to my final comment, which is really a request for help. Am I correct in claiming that case-by-case analysis is the way to go in addressing the question of the flexibility of artifacts with respect to purposes? Or is there some unifying consideration that would allow a more general or systematic approach? Am I correct in arguing that we can not read from the values of artifact creators to the value of the artifacts they create? Am I correct in arguing that artifacts determine task performance characteristics but not the structure of the organization within which those tasks are performed? BIBLIOGRAPHY Boddy, David and Buchanan, David A., "Information Technology and the Experience of Work," in Liam Bannon, Ursula Barry and Olav Holst, eds. Information Technology: Impact on the Way of Life (Dublin: Tycooly International Publishing, Ltd., 1982), pp. 144-157. Braverman, Harry, Labor and Monopoly Capital (New York: Monthly Review Press, 1974). Bunge, Mario, Treatise on Basic Philosophy, Volume 7 (Boston: D. Reidel, 1984). Cooper, Robert, "Task Characteristics and Intrinsic Motivation," Human Relations, 26, no. 3 (1973), pp. 387-413. Finne, H., "The Designer and His Job in the Face of Intergrated CAD/CAM Systems," in Thomas Martin, ed., Design of Work in Automated Manufacturing Systems (Oxford: Pergamon, 1984), pp. 65-70. Helpingstine, S. R., Head, T. C., and Sorenson, R. F., "Job Characteristics, Job Satisfaction, Motivation, and Satisfaction with Growth: A Study of Industrial Engineers," Psychological Reports, 49 (1981), pp. 381-382. Humphrys, Peter and O'Brien, Gordon E., "The Relationship Between Skill Utilization, Professional Orientation, and Job Satisfaction for Pharmacists," Journal of Occupational Psychology, 59 (1986), pp. 315-326. Jones, Bryn, "Destruction or Redistribution of Engineering Skills? The Case of Numerical Control," in Stephen Wood, ed. The Degradation of Work? Skill, Deskilling and the Labour Process (London: Hutchinson, 1982), pp. 179-200. Landy, Frank J., Rastegary, Haleh, and Motowidlo, Stephen, "Human-Computer Interactions in the Workplace: Psychosocial Aspects of VDT Use," in M. Frese, E. Ulich, and W. Dzida, eds. Psychosocial Issues of Human-Computer Interaction in the Workplace (North-Holland: Elsevier Science Publishers, 1987), pp. 3-22. Lawler, Edward E., and Hall, Douglas T., "Relationship of Job Characteristics to Job Involvement, Satisfaction, and Intrinsic Motivation," Journal of Applied Psychology, 54, no. 4 (1970), pp. 259-286. Mesthene, Emmanuel, "How Technology Will Shape the Future," Science, 161 (July 12, 1968), pp. 135-143. Mumford, Lewis, "Authoritarian and Democratic Technics," Technology and Culture, 5, no. 1 (Winter 1964), pp. 1-8. Noble, David, Forces of Production (New York: Alfred A. Knopf, 1984). Patrickson, Margaret, "Adaptation by Employees to New Technology," Journal of Occupational Psychology, 59 (1986), pp. 1-11. Radder, Hans, "Experiment, Technology and the Intrinsic Connection Between Knowledge and Power," unpublished manuscript, n.d. Sato, Susumu, "Quantitative Evaluation of VDT Operations through the Analysis of Spatio-temporal Chracteristics of Eye Movements," in G. Salvendy, et al., eds., Social, Ergonomic and Stress Aspects of Work with Computers (Amsterdam: Elsevier Science Publishers, 1987), pp. 197-202. Scott, P. J., "Automated Machining Systems and the Role of Engineering Craft Skills," in H. J. Bullinger, ed., Proceedings of the Second International Conference on Human Factors in Manufacturing and Fourth IAO Conference, June 11- 13, 1985, Stuttgart, West Germany, pp. 213-221. Wall, T. D., Burnes, B., Clegg, C. W., and Kemp, N. J., "New Technology, Old Jobs," Work and People, 10, no. 2 (1984), pp. 15-21. Winner, Langdon, Autonomous Technology (Boston: MIT Press, 1977). Winner, Langdon, "Do Artifacts Have Politics?" in Donald A. Mackenzie and Judy Wajcman, eds., The Social Shaping of Technology (Philadelphia: Open University Press, 1985), pp. 26-38. Zuboff, Shoshanna, In the Age of the Smart Machine (New York: Basic Books, 1988).