Interests and the Shaping of Technology:
an Unresolved Debate Reappears

Table of Contents

1. Abstract
2. Introduction
3. Interest Models in the Sociology of Scientific Knowledge
4. The Actor Network Approach in Sociotechnical Systems
5. The Concept of Interests in Political and Sociological Theory
6. Expedient Methods: Expectation
7. Implications and Requirements
8. Interests in Knowledge Claims in Technological Debates
9. Knowledge Forms and Interests
10. Notes 1 - 60

Abstract

The concept of interests is crucial for analyses of technological development and technological knowledge claims as social products. It must overcome both the limitations of purely subjective notions, and objections to other-ascribed 'real' interests. Critics of interests models in the sociology of scientific knowledge claimed the use of a notion of interests with objective social referents was unsoundly based. Despite problems with the theoretical formulation and its use in certain studies, none of their points are fatal for this approach. Another current in science studies, and now applied to technology - actor-network theory - takes an exclusively subjective notion of interests, open to the same criticisms as that in other pluralist/action frameworks. Theoretical debates in politics and sociology indicate a way of transcending the problems, through a conception of interests as having subjective and objective elements, dialectically related; derived by the analyst from the structural location of actors and the specific circumstances of interaction, though prior derivation is limited in practice; and complex and contradictory, being referred necessarily to potential outcomes and subjective identities. The conception can be made consistent with principles and other concepts widely adopted in technology studies - such as a view of knowledge as socially constructed, with ideological aspects related to systems of domination. It can accommodate a model of agency which retains the complexity and flexibility of actor-network theory without overstating the extent of actors' power to determine outcomes. In deconstructing and contesting forms of knowledge such as risk assessment exercises, it fits usefully with an analysis of social determination at various levels, from a general rationality down to specific methods and claims.

Introduction

I want to consider the concept of interests and its use in the social analysis of technology. My objective is to reinstate the concept and argue its importance, and to elaborate a particular conception of interests which is consistent with the principles and approaches adopted by many colleagues in technology studies.

I am aware the argument here remains at a very abstract level. I certainly do not want to give the impression that this is an esoteric theoretical debate. The concept is a crucial explanatory resource, and one which many of us use anyway, perhaps more loosely than we should. We therefore need to demonstrate that it can be used fruitfully and consistently and that the problems raised, whether by sympathetic analysts, or as objections to its use at all, can be resolved. I claim no originality for the path I suggest; it is there in a body of literature which does not appear to have been brought to bear on the debate in studies of science and technology. My formulation is by no means fully sorted out, and I continue to have misgivings; but I have been sufficiently encouraged to keep going in this direction.

I have found it useful to go back to a debate in the early 1980s, conducted in the pages of Social Studies of Science and elsewhere, on the use of interest models in the sociology of scientific knowledge. It is important for several reasons. Some of the points raised remained unresolved, possibly because of the different angles from which the authors approached the topic. Several of the protagonists are themselves now involved in the social analysis of technology, and are bringing their particular conceptions of interests - though some would not use the term - to this area.

The arguments here are thus a particular aspect and a recasting of the general critique of certain formulations of a 'new sociology of technology' - actor-network theory and what we have come to call 'social constructionists' for want of a better phrase. [1] The debate between us is, I think we are all agreed, not correctly termed one between macro and micro theorists, but between different conceptions of how to connect levels of social analysis or whether it makes sense to make such a distinction anyway. Where we part company is in the concepts and forms of explanation considered appropriate to different levels of analysis.

This discussion is intended to allow me to reinstate, or convince others of the usefulness of, concepts these two schools deny themselves in one way or another, and to connect, where it is so evidently needed, insights and principles from the sociology of scientific knowledge to other forms, particularly knowledge within technical communities and about the desirability and impacts of technologies. I suggest too that the resolution I propose here offers an understanding better suited to political intervention to challenge knowledge claims in these areas and hence particular technological developments and the social forces behind them. I hope it will also prove useful for interests models of scientific knowledge, though I shall not pursue that possibility explicitly here.

Interest Models in the Sociology of Scientific Knowledge

Let me first go back to an earlier debate on the use of 'interest models' in the sociology of scientific knowledge. Protagonists were not aligned in any simple way on two sides of a divide. Barnes and Mackenzie had made central to their studies, and defended in subsequent theoretical discussions, the concepts of cognitive and social interests as means of explaining how actors' choice of procedure and theory were orientated towards particular social objectives. [2] They took a starting assumption which would be common to many, that the content of the knowledge claims themselves is not an adequate explanation of the controversy, nor the method and procedure followed by the successful claims an adequate explanation of the outcome. But as both Woolgar and Yearley argued, [3] the 'interest model' adopted by Barnes, Mackenzie, Shapin and others was not the only possible interpretation of the general tenets of the strong programme

It did inform several convincing pieces of empirical work. [4] In a study of statistical theories of association, for example, Mackenzie argued that the theorising and judgements of his subjects 'have to be understood as embodying different "cognitive interests" ... differing goals in the development of statistical theory resulted in approaches to association that were structured differently.' [5] Cognitive interests he explained as 'those aspects of the actual or potential scientific applications of theories that "feed back" into theoretical developments by structuring scientific construction and judgement of theories.' [6] Mackenzie clearly takes a conception of social interests as having objective elements in terms of benefits to social groups from an outcome - thus not necessarily as expressed in stated preferences - while not claiming these interests are necessary and sufficient to explain it. [7]

The theoretical elaboration of the concepts of cognitive and social interests in Barnes and Mackenzie's work, it has to be said, left ambiguities. It was not pursued far enough to show convincingly that they had taken into account the problems I shall indicate later with the attribution of objective interests, nor did they theorise or demonstrate mechanisms by which objective interests contributed to determining beliefs.

Woolgar attempted to demonstrate, by turning his ethnomethodological analysis onto Barnes and Mackenzie themselves, that their use of the concept was not justified. Interests 'enjoy an unproblematic existence, to be drawn on at will by the investigator.' [8] Barnes and Mackenzie had imputed a causal type explanation, which then got diluted as they allegedly backed off from seeing interests determining knowledge production to the weaker term of 'influencing'.

To some extent I can agree with the latter two limited criticisms; the connections are not altogether convincingly made, something which Barnes and Mackenzie acknowledge. Nor is the use of the two categories of cognitive and social interest clarified sufficiently to establish the separation or indicate the connection. [9]

But the main thrust of Woolgar's attack is that the concept is adopted without reflection, and he relies on arguments about counterfactuals to try to show the necessary circularity of the explanation. I see no objection to insisting on 'a high degree of methodological reflection' and reflexive consistency - that analysts' approaches to explaining others' knowledge claims should be applicable to their own. But I take Barnes and Mackenzie to be entirely justified in their arguments against Woolgar's rejection of this theoretical resource. Their own case against the futility of his position needs no additional help from me; as Barnes said, 'only rabid positivists insist on the requirement (to identify independently cause and effect) and abhor theoretical entities.' [10].

Like the actor-network theorists I go on to consider, Woolgar saw actors' subjectivities as constructed entirely in the course of interaction with other actors and with no external social reference points. Thereafter Woolgar's case dissolved into hyperreflexivity because of his own ontological position: that interests and moreover context only exist in the form and to the extent that they are constructed by actors and by analysts themselves. I see little point in trying to pull him out of an inward spiral that can only lead to complete analytical and political paralysis. [11]

Yearley I believe helped to focus on some of the ambiguities in Barnes and Mackenzie's formulation and use of the interest model. [12] Again I want to show how some of his points may have some justification as criticisms of specific studies but are by no means fatal and insoluble for the project as a whole and for its extension to other areas of knowledge. The first of Yearley's major points concerned the position of interest model advocates towards instrumentality in scientific procedure: whether it should be seen as a transcendent and 'authentic interest in prediction and control', and therefore separate from ideological characteristics of knowledge production; or if particular forms of instrumental rationality are themselves ideological - the view I would hold and which, as Yearley argues, implies 'a variable, interpretative connection between knowledge and interests'. [13] But I do not accept that this latter view is 'rather at odds with' an interests view of knowledge; it merely indicates the need to recognise the multiple interests of any subject, and hence to explain the sources of variability in differing constructions - in other words, a push for greater determinacy in our explanation to take into account all sorts of conditioned forms of thought as well as responses to immediate conditions.

Yearley's second main point was that interest-model theorists usually made the connection with social interests by observing that scientific actors were actively engaged in politics for a particular social group or class; he characterised the claim thus: 'scientific thought was given a particular content because it was a resource employed for political ends.' [14] I shall not attempt to judge here whether these particular interest-model studies did rely solely on this sort of link between knowledge and interests. But I shall argue that such a direct connection is by no means the only or the most important one, nor need we be limited to such evidence, useful as it may be.

It is the differences between interest theorists and Callon and Law [15] which I shall follow more closely. Callon and Law claimed they wanted to duck out of the debate so far, and 'avoid a commentary on the relative merits of these two positions'. [16] But in putting forward their actor-network framework for understanding the social shaping of scientific knowledge, they were introducing a very different concept of interests - or rather denying the usefulness of interests in any other sense than as subjectively defined goals [17] - and thereby placing themselves fundamentally in opposition to an interest model with referents in social structures beyond the immediate interactions.

The Actor Network Approach in Sociotechnical Systems

Callon and Law along with other proponents of the actor-network approach like Latour and Bowker have more recently applied their analysis to technology, or rather sociotechnical systems. [18]

The general features of the actor-network approach are probably well known by now. Actors in general seek to enrol others to a conception of the issue which favours them, and manipulate and transform others' (subjective) interests. In particular, central figures in the development of sociotechnical systems - Law's heterogeneous engineers, Latour's Machiavellian princes - attempt to enrol the necessary actors for their project by translating their conception of the system. The approach follows the process of enrolment and counterenrolment, and the changes, as the actors interact, in the project as conceived and in the interests of each.

Let me make it clear I am not opposing the use of the techniques of the actor-network approach. The merits of the model and the studies based on it are clear. It encourages exploration of the complexity of the set of events surrounding a particular development; this is a welcome antidote to the reductionism of many frameworks, and allows for multiple influences. With a stress on an unfolding and unpredictable process, the approach allows for changes in the character and objectives of actions resulting from interactions. It acknowledges the problematic nature of technological advance, the possibility of redirection at any point, the plasticity of outcomes. It thus not only contributes to a critique of technological determinism, but also effectively counters instrumental notions of technologies as tools, chosen and applied in pursuit of predefined objectives, with unproblematic results - in other words, as some simple expression of the operation of broader social and economic forces. The actor network approach shows the intimate interweaving of technical and social elements, or rather that 'technical' meanings are inextricably social. In stressing in particular the social construction of the knowledge embodied in artefacts and about their functioning and effects, it allows us to push back the boundary between what is accepted as technically determined and what can be seen as socially constructed constraints on choice. The approach emphasises the importance of studying the specific locales of technological development and decision-making. The studies, and their presentation, have been rewarding and rich in insights.

What I have argued against elsewhere, however, [19] and implicitly continue to do here, is their position of agnosticism towards accounts of broader structures and processes of social systems, their denial of categories developed for such levels of analysis, and their refusal to make prior assumptions about institutions in which specific actions are located. I maintain that the insights of the approach can be retained, and the problems they fall into can be avoided. The route is not to construct supposedly a new account from the level of detailed action up. This can only lead to an overstatement of the malleability of existing social arrangements faced by actors, which helps neither an explanation nor the actors we might be trying to help; and to them slipping in by the back door implicit pluralist social models despite their avowed agnosticism. It is rather to find ways of selecting broader level analyses and and making them consistent with accounts of specific events - theorising ways the context constrains and shapes particular outcomes. The approach would certainly recognise that actions are undertaken to change aspects of the context or maintain them against challenge. But it would contend that they in turn are shaped and constrained by the prior existence of social systems - relatively stable and durable assemblages of practices constituting economic, political relations, knowledge and language conventions, and so on.

The conception of interests in the actor-network approach - in declaration and in practice - is entirely subjective: interests are equated with, and analysed through, stated preferences or those revealed directly in action. [20] The notion is entirely consistent with an individualist action framework, a pluralist social model, and other concepts deployed. [21] As such it is subject to the same criticisms of explanatory inadequacy and ideological function as earlier pluralist approaches in conventional political science and sociology.

The Concept of Interests in Political and Sociological Theory

The concept of interests has been the subject of a long debate in political science and sociology literature. Like power, if it is not essentially contested, [22] then certainly different conceptions are entailed in particular approaches, hence political viewpoints, and alignment to (I have to say it) social interests, and there are limits to the progress that can be made in debate over it, between what are largely incommensurable paradigms.

Perhaps the context in which the concept has been most thoroughly debated is in the debates on power which had their origins in Bachrach and Baratz's critique of pluralist accounts of community politics in the USA. [23] It is worth following because the pluralist framework and subjective notion of interests is closest to the actor-network theory and because it takes us through - and beyond - the notions of objective interests and the philosophical problems which have been used as justification for rejecting any reference to social structures.

In the pluralist paradigm, the dominant current in much Western political science, interests are entirely subjectively defined, manifested as preferences as to outcome exhibited through participation in political arenas. Pluralists are actively opposed to ideas of unobservable, unarticulated and especially unrecognised interests. Elite theorists, adhering to a notion of power operating also in covert ways - such as through the withholding of demands because of an anticipated negative response from powerful groups - extended the identification of interests to include those not fully articulated and pursued - latent or suppressed wants and grievances. [24]

Lukes' well-known extension of power to a 'third dimension' acknowledged a prior stage in such processes and the possibility of unrecognised real interests because of systematic manipulation of consciousness. [25] Thus in the absence of observable conflict, there can still be a potential conflict between the interests of those exercising power and the real interests of those dominated.

While indicating the need for an analysis in terms of both structure and agency, Lukes still framed his discussion within the individualist problematic carried over from the theories he was criticising. His analysis remained a dualist one, 'in which morally responsible agents choose their actions under conditions of more or less relative autonomy.' [26] Crucially, he failed to establish a sound way of establishing 'real' interests, taking them as wants as they would be expressed in hypothetical conditions, those of relative autonomy of the actor. Benton has thoroughly exposed the weakness of this view. [27]

McEachern likewise illustrates the inadequacy of Lukes' counterfactual approach, by showing how it breaks down when one considers an interaction between government and capital in which the latter appeared successfully to be imposing a policy on the government. Lukes' procedure of 'imaginary reconstruction in the absence of an exercise of power' gives no guidance to the analyst on how to establish what the government would have done. To fall back on stated policy preferences would be to revert to pluralist procedure. [28]

Many other authors, however, have used difficulties like Lukes' to attack the whole idea of the analyst ascribing 'real' interests, seeing the procedure as necessarily arbitrary or circular, and have pointed to the dangerous implications of political leaders taking action on behalf of, and against, those who do not recognise their own 'real' interests. The problems, and a rejection of the concomitant notion of ideology as 'false consciousness', have allowed many to drop the concept of interests altogether in favour of goals or objectives - subjectively defined. They also leave an uneasy feeling among those who like myself retain the term in an implicitly objective conception, hoping rather loose usage will allow us to avoid being confronted with the problems.

Such objections may be justified for Lukes' formulation, but they by no means invalidate a conception of interests which involves reference to objective social location. Rather, as Stolzman says, [29] they represent a mistaken view of what in involved in defining such interests. Let us start from the recognition that even though individual actors act on their own definition of a problem or other motivation, an actors' subjective interests are influenced by processes beyond the individual, and the consequences of that action are in large part determined by social structures of which they may or may not be aware. We then, as Stolzman insists, have to face the methodological perils.

We can look at some further treatments of the concept by authors who have sought ways of relating outcomes of interactions to broader social features, and from these suggest a way forward. For McEachern, power is the ability of an actor to secure its interests or to have them secured, and

interests are structurally located, that is, socially significant groups have their primary interests as a consequence of the place they occupy in the social structure. [30]

Analysis thus entails the investigation of actual and potential effects of action on a wider set of groups than those immediately involved. It is also necessary to consider a longer time span than the events in focus to allow for the working through of effects; to stop at the closure of debate or the decision output would preclude the understanding of wider interests. It is

important to stress that (interests) are objective, in the sense of being located in the structures of social life, rather than being subjective, that is, existing only in the perception of that life. There is a link between the objective character of interests and the ways in which they are perceived by individual and collective actors, but this does not affect their objective character. [31]

Similarly Westergaard and Resler see interests as determined by social location,

as the possibilities and potential objectives of action which are inherent in economic positions regardless of whether the incumbents of those positions in fact so define their objectives at any given time. [32]

Alford and Benson talk of 'structured interests', 'built into a sector, either positively or negatively, in the sense that the operation of the sector serves or conflicts with their interests.' [33] The structure of a sector is 'an arrangement typically empowering and defending some interests at the expense of others.'

Benton, however, rejects the idea of objective identification of interests independent of the ascriber, [34] claiming it is usually accompanied by counterfactuals and thus lapses back into conflation with wants; or else ethically or historically transcendent viewpoints, or universal theories of human needs, are adopted to anchor it, with the concomitant ontological and epistemological problems. Instead he argues for the identification of alternative objectives through the 'effective action' of subordinate groups in abnormal conditions. A conception of power based on objectives as revealed in the symbolic content of practices, Benton argues, can sustain Lukes' criticisms of pluralist and elitist theories, but does not entail a notion of 'objective interests'. The objections are close to those of pluralist critics of Lukes, and yet, in retaining 'interests' for other purposes, Benton formulates a notion close to that of Giddens and Jessop as considered below.

Giddens, in a brief exploration of the concept, likewise rejects conflation of interests with wants, though sees the two as closely related. [35]

... to say that A has an interest in a given course of action, occurrence or state of affairs, is to say that the course of action, etc. facilitates the possibility of A achieving his or her wants.

Interests and power are closely related, but contingently rather than logically. Interests, in that they concern the modes of realisation of wants in contingent social and material circumstances, can thus be objectively determined. Giddens rejects the notion that interests are structural properties of collectivities as used by McEachern. Rather, actors have interests by virtue of their membership of particular groups.

Jessop produces a similar but more fully developed formulation which I believe indicates a resolution of many of the contentious points in the voluminous debate. The analysis of interests is to

... be undertaken in a relational context concerned with comparative advantage rather than some notion of absolute interests posited in isolation from specific conjunctures. A situation, action, or event, can be said to be in an agent's interest if it secures a greater net increase (or smaller net decrease) in the realisation of that agent's conditions of existence than do any feasible alternatives in a given conjuncture. This implies that an agent's interests must be assessed in relation to the structural constraints and conjunctural opportunities obtaining in a given period. [36]

Actors can face a conflict of interests in relation to different conditions of existence, particularly in that actors are involved in different relational systems and have multiple identities or subjectivities. Thus interests can only be assessed in terms of alternative outcomes in particular situations for specific subjects. The balance for a subject changes with a change of conjunctural opportunities, structural constraints or a redefinition or recombination of identities.

There is thus a dialectical relation between subjective and objective interests.

Objective interests must always be related to a particular subjectivity occupying a particular position in a given conjuncture: a particular subject can nonetheless miscalculate these interests since they are defined in terms of the conditions actually necessary for its reproduction rather than the subject's own views of these conditions. [37]

Benton's conception of interests is similar but ultimately limited to use in political discourse rather than in the analysis of power relations: distinct from wants; ineradicably evaluative; to be assessed by evaluating various evidence and considerations with no privileged place for self-ascription; and closely related to social identities, the formation of which is a contradictory process. Hindess also finds the concept of interests inadequate as a starting point in analysing interactions; it cannot provide a 'general model for explanation of the ways actors are mobilised in struggles.' [38] Rather, analysis should identify the various distinct and possibly contradictory objectives actually pursued, partly a result of specific conditions, possibly changing as a result of interaction, and with no simple relation to what might be ascribed as 'interests'. In both these and in Jessop's formulation is a potential resolution of the problems of attribution of interests and the 'paradox of emancipation' in socialist revolutionary practice: the oppositional use of interests must be rooted in at least some aspect of the social experience of the individual actor. [39]

However, if with Benton and Hindess the necessity of interests in analysing power relations is rejected and instead 'objectives' are considered - even as extended and complicated by the inclusion of other evidence of aspiration besides stated preferences - there is a danger of missing one of the main points of McEachern's study: that it is possible for certain sections of society to have their 'primary interests' secured through the operation of systems of domination as mediated by other actors, without their direct participation in interaction and hence any manifestation of 'objectives'. It seems essential then, to uphold McEachern's derivation of 'primary interests' from social location. They equate with what Jessop terms 'conditions of existence', which are routinely maintained by the reproduction of systems of domination. It goes without saying that a key sectional interest is exactly in maintaining the existing social order.

Expedient Methods: Expectation

The problem remaining with the use of a conception of interests as derived from structural location and specific conjunctures is the practical difficulty of their identification. Derivation of basic conditions of existence is relatively straightforward for certain groups and spheres of activity which fit neatly into the fundamental categories of a mode of production: namely classes, class fractions and individual capitals, in production and exchange. The further one gets from these categories the less extensive a prior derivation can be, and the greater the dangers of circularity and reductionism. There are two reasons: first, there is objectively a considerable degree of independence in activity at different social levels - between economic forces, political action, ideologies; and second because other spheres have seldom been theorised in sufficient detail. [40]

The pragmatic approach adopted by many social researchers uses this a priori analysis of interests at least as a starting point, even if they retain doubts about its philosophical foundation. [41] The analyst is then faced with groups without clearly understood locations, with the complexity of a limited area and specific circumstances, and the multilayered and often conflicting interests of actors.

Here most resort to some of the more expedient ways by which non-decision theorists produce what they termed 'expectation' of behaviour: [42]

- defining human needs normatively, so that the set of unsatisfied needs requires investigation; [43] - comparing similar social collectivities or spheres of activity [44] and taking suggestions from the diversity of circumstances; [45] - comparing the same group or sphere at different times; - looking at the actions of a group in abnormal times, the spread of alternative views during crises, and reactions to changes in conjunctural opportunities; [46] - identifying contradictions between behaviour and some principle already established as accepted by an actor; - identifying contradictions between the claims of a dominant group to fulfil some function, and empirical evidence; - identifying codified rules of exclusion; [47] - looking for what Offe terms 'misunderstandings' or 'overdeterminations', that is, contradictory depictions of certain objectives or demands; - studying the actual or potential effects of a change on a group, in terms of actual or expected losses; - and similarly in the maintenance of an existing arrangement, the absence of any attempt to redress maldistribution of some item of acknowledged value.

The methods all face the obvious problem of justifying the identity of something that does not exist, since only in an observable conflict of objectives is a counterfactual readily providable. [48] They have particular limitations, and particular normative criteria, which they generally fail to establish and to which serious objections can be raised. They are all expedient; each may or may not work, and the whole set may not be exhaustive or adequate. [49] But none of these considerations invalidate their use with caution as heuristic devices.

Implications and Requirements

So we arrive at a provisional conception of interests along these lines: having subjective and objective elements, dialectically related in that neither can be specified without reference to the other; derived from the structural location of actors and the specific circumstances of interaction, though the extent of identification by the analyst is limited by inadequate theorisation so that analysis must fall back on more expedient methods; referred to potential outcomes and subjective identities and therefore complex and contradictory. They are related to, but not the same as wants or objectives; the actual objectives pursued must be reconstructed.

This notion of interests doubtless needs to be refined. It remains a theoretical resource for explanation, but like Barnes I see no reason to apologise for this. But I believe it overcomes the problems which Lukes ran into and which have been seen to justify the rejection of any attribution of objective interests. And above all it can be made consistent with - or put another way entails a particular formulation of - other key concepts in social analysis, particularly power and ideology.

The notion does of course presuppose that we accept that it is valid and necessary to analyse social phenomena in broader categories than those appropriate to individual interactions, and that even at a detailed level we require for an adequate explanation of actions and interactions an understanding of context in which they take place.

It is consistent with a conception of ideology as a system of beliefs necessarily connected to social domination. If we are analysing all knowledge forms as socially constructed, then we cannot maintain an idea of ideology as the antithesis of 'truth' or somehow objective knowledge of social life - except for its expedient use in political discourse. But neither is ideology a superficial phenomenon entirely determined by the economic or political levels of society. Rather, it is itself a site of struggle, over beliefs: to get people to construct their subjective interests according to particular aspects of their social location and exact circumstances, to define their identity differently. [50]

Thus our idea of interests can be consistent with the notion that people actively construct or interpret their subjective interests; it does not depict agents as 'interest-dopes'. [51] I would, however, put less emphasis than, say, Barnes on the active, conscious, deliberate element of this process implied in the phrase 'constructed using resources'. If we accept a notion of practice, as routinely repeated and reciprocated actions with an order which can be analysed as exhibiting a particular rationality, then we should acknowledge mental practices - and the more passive reception and transmission of beliefs.

It allows for multiple, conflicting subjectivities, corresponding to an actor's position at the intersection of various social structures, and hence of ideological struggle over subjective interests - to persuade individuals to identify their interests in an outcome according to particular identities. We can depict those interests as a hierarchy, as short or long term, immediate or more fundamental, and with possible conflicts between them. It does not require, and would warn against, the attribution of a single interest. [52]

Similarly, I would argue it entails a notion of power which, for whichever element we might retain the term itself, acknowledges the systematic maintenance of systems of domination as well as active exercises of power in overt or covert forms when these systems are challenged or need defending.

It avoids the notion that interests are only involved in an interaction, or that interests can only be identified in a particular set of knowledge claims, if and to the extent that those participating stand to gain directly from an outcome.

One of the criticisms levelled by actor network theorists at the prior identification of social interests as influencing specific actions is that it constrains the explanatory process. It supposedly artificially imposes the analyst's preconceptions and rigidly ties the actors, and reduces their motivations, to static and inappropriate large scale categories, and so is unable to accommodate the way interests are shaped by enrolment in the process of interaction. The obvious point about the starting objectives of the actors at the point the analyst chooses to start, and their source, has already been made. There certainly is a danger of reducing the social locations of actors and determinants of actions to simple categories, and as I have pointed out, there is a limit to the extent to which prior interests can be attributed, but this is not a necessary failing of the approach.

The method is retroductive, in that we specify conditions and mechanisms which produce the events we are trying to account for; we then use empirical observation, at whatever level is appropriate to reflect back on the adequacy of our explanation and hence modify our theoretical model at suitable levels. It is an iterative and continuous process, and any findings must be treated as provisional. [53]

There is no need to abandon the attribution of interests to actors on the basis of a prior analysis of social location, and every reason not to. The conception here can accommodate a notion of agency which allows for the development of actors' subjectivities during interaction; but first it does not assume a clean slate at the start, and second it insists that those interests continue to be articulated with reference to (objectively existing) social structures in which actors and interactions are located, and that courses of action are constrained by those structures - including of course, the ideological structures which influence ways of thinking - not just by the purposive action of the limited set of other actors immediately involved.

The conception developed is consistent with an active, knowing agent, but one which can never have full knowledge of the objective conditions in which it operates, nor foresee all the consequences of its actions. It recognises that much action is in the form of structured, routine practices, and much of the knowledge informing action is not the conscious purposive sort which the actor-network theorists focus on exclusively, but tacit unquestioned mental practices, seldom discursively acknowledged. [54]

Interests in Knowledge Claims in Technological Debates

The notion of interests I am trying to develop here is intended to be useful for analysing those interests underlying - though not necessarily causing in a simple way - social action, but it is useful here particularly to explore how it can be deployed in analysis of systems of belief in and about technology as a social entity. We can use it to show how interests are incorporated in knowledge forms and content, in the sense that the acceptance, rejection or modification of a particular view of an issue, the set of meanings attached to a phenomenon, or more specific claims, will advantage certain groups. Here is the direct parallel with the debate over scientific knowledge, and while I have specifically used the idea in a critique of technology assessment and similar policy science exercises, those studying processes of negotiation of scientific knowledge among practitioners may want to find parallels.

I shall say little in this section of the paper about an analysis of the institutional structures and processes which produced a particular form of knowledge - the sort of analysis that must be done alongside and in conjunction with deconstruction of the texts.

For a set of exercises which claimed to quantify and compare the health risks associated with electricity production from alternative generation routes - and project the method as neutral, objective and necessary for rational decisions - I attempted to demonstrate [55] how specific assumptions had been incorporated in the form of the exercise and at a multitude of points at which judgements were required on the part of the assessor. I found it useful to depict these at a number of levels.

At the most general level was a rationality - not in the evaluative sense, but as a general pattern discernible in a set of practices, here intellectual practices, through which they are orientated towards particular objectives - or more often a combination of different rationalities reflecting the origins of both practitioners and techniques. This is the sense in which we can identify an economic rationality, important here in the derivation of many of the favoured techniques in policy sciences from welfare economics, or in the broader sense that, say, commodity or gender relations come to be projected onto depictions of other social relations. Here we might also accommodate Marcuse's notion of scientific rationality, stressing domination over nature through prediction and control - as I have indicated, I would treat that as no more transcendental or universal than any other aspect of scientific knowledge - and hence of technical rationality, the dominance of which is for for many authors the defining characteristic of technological societies, in which the means of controlling nature has translated readily into the social world. [56] The problem with this is, of course, specifying mechanisms by which these rationalities come to be incorporated, not to mention to avoid oversimplification and reduction of multiple influences. It nevertheless strikes me as reasonable to suppose that people theorise, construct images of the natural and social worlds, in terms of the tacit knowledge which guides their mundane daily practices.

At a more specific level, we can identify particular problem definitions, with consequent restrictions on discourse. In these exercises the depiction of the problem as one of 'risk' or 'acceptable risk', rather than, say, blame or responsibility, had seldom been questioned. [57]

Then in the scope and form of the exercises fit assumptions about knowledge: positivist assertions that facts can and should be separated from values, here represented in an insistence on the division between stages of risk 'estimation' by experts and 'evaluation' in political arenas.

The terms of reference of these exercises then provided further ways of precluding particular outcomes - which options were included and which excluded. Predictably virtually all these comparative studies of energy systems assessed supply side options and thus excluded at a stroke the potential reduction of hazards through energy conservation.

Further narrowing came in the form of quantification used. Quantification can of course be analysed as a general practice, as a feature of particular rationalities, and indeed as ideology. [58] Where these sorts of policy science exercise are used in decision-making, quantification may in effect be the reification of social relations, turning the measured results of inequitable patterns of distribution of costs and benefits into normative criteria. [59] But at this level we can simply observe the exclusion of items that could not be quantified in the units chosen and hence would be ignored or downgraded in importance in the 'evaluation' stage.

And then within the terms of the exercise we can find innumerable points at which the assessor had to make decisions about how to proceed with the exercise where no formal guidelines existed: where to make a simplifying assumption, and what that assumption should be; how to interpolate or extrapolate or adapt data from other sources where there were gaps; how to assess reliability and when to reject anomalous data; which model of a mechanism of release, transport or uptake of a dangerous substance to select; and finally how to aggregate and present the results.

Only at the lowest level of detail - the methods of calculation, the selection of data, the specification of margins of uncertainty - did it seem appropriate to acknowledge the possibility of deliberate manipulation of information to achieve a desired result. Indeed, in my experience of negotiating the significance of particular findings in such an exercise, I was not convinced it is possible, necessary or helpful to draw a clear line between unconscious and conscious shaping of results. By this stage the general tendency of the exercises to produce the results desired by certain parties in the energy sector had been established, without the need for attribution of malice and conspiracy which has tempted critics.

Knowledge Forms and Interests

Let me list some of the advantages of this admittedly rough and pragmatically devised scheme of levels and try to demonstrate its consistency with the notion of interests I have been trying to develop.

First it is consistent with and most useful in conjunction with a treatment of knowledge as analysable at a broad level and having structural aspects. The general norms, procedures, assumptions, methods become resources drawn on, deliberately or otherwise, in constructing and justifying such assessment exercises.

Second it helps in understanding the asymmetry of defence and challenge: those opposing established interests not only have to tackle knowledge claims in their own terms - contesting 'facts' as 'mistaken' - but if they are to avoid being confined within terms of reference which will disadvantage them and restricted possible outcomes, they must seek to expose the assumptions behind the exercise, undermine its framework and thereby broaden the scope of debate. In effect they will be challenging deeply rooted notions about the sensible, rational way to approach the issue. They will be disadvantaged to the extent that dominant groups are able to invoke, deliberately if necessary, widely held notions, however vacuous.

The schema points away from a narrow notion of the interests involved in any interaction - the idea that one can only legitimately demonstrate that 'social interests' are involved in a debate if the protagonists themselves somehow benefit, or perhaps those on whose behalf they are presenting a case. Rather the ideological content of their case may reflect and reinforce some aspect of the wider social order. In their construction of knowledge claims in technological debates, experts may in effect be acting on behalf of interests not directly involved. My argument parallels that about bureaucracy; in trying to explain why bureaucracies behave as they do, it is not enough to consider what bureaucrats themselves gain from the process; it cannot be understood without reference to the broader effects for the maintenance of a social order of interposing a bureaucratic treatment of an issue.

The argument has probably been worked out best for technocratic rationality and ideology. The dominance of experts in politics, the depoliticisation of decisions, does not automatically support a classical picture of technocracy in which the experts themselves wield ultimate power. A more convincing argument is that technocracy is a form of management most appropriate to the problems of a particular stage of an economic and social system which is still based on the same fundamental social relations.

So the scheme can help avoid the weak tactic of trying to condemn an expert case simply by demonstrating institutional connections or claiming that the expert has some immediate self-interest in the case put forward. This is clearly not the case where the expert is institutionally detached. The scheme instead points to a correspondence of assumptions which may advantage certain interests.

It does not require features at all levels to be entirely consistent. Coherence of the whole view, the relation of specific claims to general tenets, will have, to differing extents, to be constructed, maintained, and argued by proponents of an exercise. Conversely, though I have indicated the systematically disadvantaged position of oppositional groups, it may be possible to attack contradictions between levels, by say depicting different interpretations of evidence consistent with a stated principle, or demonstrating a conflict between an invoked principle and the actual practice of an assessor, perhaps the selectivity of their avowed 'rationality'.

It allows complexity of determination and fluidity of outcome. There is no necessary direct, reductionist connection of knowledge form and content to external interests; the detail needs to be constructed and argued by the analyst.

Finally it may also help us reflect on our own role in analysis. There is a strong argument, which I shall not elaborate here, that a partisan approach is necessary for analysis as well as to satisfy any political commitment we may have to supporting particular actors - in other words there are limits to a non-interventionist analysis. [60] The implication of my argument about interests and their embodiment in knowledge claims is that an intervention in debate need not be narrowly and instrumentally partisan, constructed to produce the answer required by a particular actor; rather the assumptions in our framework can be consistent with a view of the world which is critical of more fundamental aspects of the existing social order. It allows us if necessary to keep our distance from a particular social or political group, to take a form of detached partisanship, perhaps interpreting the issue in a different way, while effectively engaging in political action against the same conditions.