Notes on: Latour B. On actor – network theory. A few clarifications (1996) Soziale Welt 47, H. 4: 369 – 81. HTTPS: //www.JSTOR.org/stable/40878163

 Dave Harris

The theory has been misunderstood. It has been given a technical meaning in the sense of a train or subway network, related but distant elements with compulsory circulation between the nodes through paths that give some nodes strategic importance. Computer networks are the best examples of a final and stabilised state of an actor network. But networks here may be local, with 'no compulsory paths no strategically positioned nodes'(369). It is a network of power.

Nor are the networks really social networks but instead are about 'the social relations of individual human actors' rather than 'institutions or organisations states or nations', which are less realistic and too large, all-encompassing. However nonhuman and non-individual entities also included. Social theory is to be rebuilt out of networks, as 'an ontology or a metaphysics' as much as a sociology. Social networks are included but they are not privileged. The network concerned actually comes from Diderot and was intended to avoid a divide between matter and spirit. It avoids the notion of essences and replaces that with filaments or rhizomes in D&G. There are, however many dimensions, as many as there are connections, giving a 'fibrous, threadlike, wiry, stringy, ropy, capillary character to modern societies' (370) rather than a matter of levels or layers or systems. It's been developed in studies of science and technology and the facts produced by sciences and artefacts are essential to understand what holds society together.

In some ways it is 'a simple material resistance argument'' stressing the 'dissemination, heterogeneity and plaiting of weak ties' as the source of social strength, what Foucault refers to as micro-powers, it proceeds by taking local contingencies, unconnected localities which then turn into sometimes 'provisionally commensurable connections'. There are affinities with notions of disorder or chaos philosophy and 'many practical links with ethnomethodology'. Universality and order are exceptions. Local contingencies or clusters are best seen as archipelagoes on a sea. There is no attempt to fill in what is between these local pockets of order — 'literally there is nothing but networks'. This makes ANT reductionist and relativist, but it will end in 'an irreductionist and relationist ontology'

The properties of nets are added to the notion of actor that does some work. The first property of networks is that it avoids the tyranny of distance or proximity, so that elements can be close when connected even though they may be infinitely remote or distant, as in telephone connections, or adjacent animals cut off by a pipeline, or neighbours separated by cultural differences. Geography is prevalent and stops us defining all associations in terms of networks. Geographers are to blame in defining proximity and distance as special, and or ignoring proximity or distance not defined, say on maps, there is a 'tyranny of geographers in defining space' (371). The notion of scale is also changed, especially the macro micro distinction, and is replaced by 'a metaphor of connections' — networks are simply longer or more intensely connected rather than bigger we can thus avoid notions of order that go from top to bottom and the difficulties of tracing links between the scales. In ANT there is no 'a priority order relation' and so it avoids questions of top and bottom and transformations, and indeed can follow changes in scales — 'the scale, that is, the type, number and topography of connections, is left to the actors themselves.' We thus regain more manoeuvre between the ingredients of social life, vertical space, hierarchy, layers, wholeness and we can see how these features are achieved. We can think of global and local entities, and can see how elements become strategic [that is important to wider connections]

The third property that is avoided is the notion of inside/outside 'a network is a boundary without inside and outside' (372). As in Deleuze [?], a network creates the background and the foreground. There is no need to fill in the space between the connections — networks have no negativity to be understood, no shadow.

These properties have made the study of society and nature difficult, and ANT does not have to qualify matters as either social or natural. There are things that are macro social or outside nature but these are best seen as the 'effects of distance, proximity, hierarchies, connectedness, out sidedness and surfaces' and we have to study these through a great deal of work. The topological notion of network itself is insufficient and we need to add a foreign notion — 'that of actor', moving away from mathematical properties to introduce dynamic ones.

'An actor – network is an entity that does the tracing and the inscribing'. It's not a piece of inert matter in the hands of human planners, and it should not be understood as involving human intentional individual actors, especially not people, usually males who want to grab power through a network of allies, doing networking. In ANT, an actor 'is a semiotic definition — an actant-- that is something that acts or to which activity is granted by others. It implies no special motivation of human individual actors, nor of humans in general. An actant can literally be anything provided it is granted to be the source of an action' (373) this is difficult to understand because 'anthropocentrism and socio-centrism are so strong in social sciences'. Actually, ANT is completely indifferent to models of human competence, 'the human, the self and the social actor of traditional social theory is not on its agenda'.

Instead it studies the attribution of 'human, unhuman, nonhuman, in human characteristics' and how they are distributed among entities, connected together, circulated and distributed. These must be grasped in an integrated practice of study. Semiotics is a necessary step following from 'when you bracket out the question of reference and that of the social conditions of production is — that is nature "out there" and society "up there". What is left is a first approximation — 'meaning production, or discourse, or text' as in the linguistic turn, where discourses became mediators, rich and complex. ANT draws from this work what is useful 'to understand the construction of entities — that's 'every entity, including the self, society, nature, every relation, every action, can be understood as a "choice" or a "selection" of finer and finer and branch moments going from abstract structure — actant's — to concrete ones — actors'. We retrace this generative path. [Better than Hjemslev on universal 'expression'?]

'...the granting of humanity to an individual actor or of collectivity..., or anonymity, ... zoomorphic appearance..., amorphousness...materiality' involves the 'same semiotic price', although the effects are be different. The work of attributing and imputing is the same. We have to think of all these actors as 'flows, as circulating objects undergoing trials', made stable, continuous as a result of 'other actions and other trials'. Various texts and discourses retain their ability to define their context, authors and readers, 'even their own demarcation and metalanguage'. The "text itself' remains as central to analysis, just as the slogans about nothing outside the text argued, but the ontological consequences are avoided, 'by extending the semiotic turn to this famous nature and this famous context it has bracketed out in the first place' (374). [cf Kirby's attempt to do this with Derrida]

We saw the problems when the slogans were applied to scientific and technical discourse, and scientific texts. Scholars did not treat fictions or myths seriously anyway and so their distance and scepticism was easy to achieve, but the real test was hard scientific text. ANT semiotics extended to include 'a completely empty frame that enabled to follow any assemblage of heterogeneous entities — including now the "natural" entities of science and the "material" entities of technology'. But as a method to describe generative paths 'of any narration, analysing 'what the recording device should be that would allow entities to be described in all their details… The burden of theory [is] on the recording, not on the specific shape that is recorded'. Actors may be human or unhuman, pliable, cross scales, ordered or not and this 'does not qualify any real observed actor' but specifies the 'necessary condition for the observation and the recording of actors to be possible', not 'constantly predicting how an actor should behave and which associations are allowed a priori'. It is not a theory of action but 'qualifies what the observer should suppose' if they are to record the shapes they seek, and 'any shape is possible provided it is obsessively coded', rigidly or heterogeneously.

We are talking of 'infra language' rather than 'metalanguage' (375). ANT is not a descriptive vocabulary but is concerned with 'the possibility of describing irredeductions', [ie weproceed until we achieve irreducibility?] against all a priori reductions. Nor is it merely empiricist, since theoretical commitments are required in order to define 'such an irreducible space', and so is polemic. Indeed ANT is often accused of being dogmatic or simply descriptive, or blandly claiming that actors are really infinitely pliable or free.

It should be accepted that bracketing out social context and reference does not solve the problem of meaning, despite the claims of the enthusiasts, nor does sketching the shape of associations offer any actual explanation. We need a third strand, which is particularly devious, although it makes AMT look really commonsensical.

Once we consider nonhuman entities and nature, semiotics looks weak because it appears that we can do without words like discourse or meaning altogether, since semiotics classically studies texts rather than things. They did 'scientistically' believe in things in addition to meaning, or even context, but saw a semiotic of things as 'easy, one simply has to drop the meaning bit from semiotics'. If semiotics now becomes path building or order making or the creation of directions, you can include objects as well as language and unified practices that were once seen as different. One consequences to elevate things to the dignity of texts or give texts the ontological status of things. Both are elevations not reductions, and absolute distinctions are lost 'between representation and things'. '… This is exactly what ANT wishes to redistribute through what I have called a counter Copernican revolution'

[So, stap me! The bugger is arguing that human exceptionalism can be replaced not for ethical or political reasons, but for analytic ones!]

This also leads us to a better understanding of explanation, and therefore to a methodology that deals with description. Actor networks connect and this will 'provide an explanation of themselves, the only one there is 'for ANT'. An explanation is 'the attachment of a set of practices that control or interfere in one another' [cf Deleuze on the Logic of sense] and no explanation is stronger in providing connections among unrelated elements or showing how one element holds many others'. This is a property of networks, not something distinct, and explanations grow as networks grow. Networks attract explanatory resources as they grow. Explanations do not get added from outside 'one simply extend the network further' (376) and this extends frames of references, frames, explanations. Each account has to be recalculated, just like in Lorentz transformations. This does not contradict science and its task of providing explanation and causality — science proceeds by extending the network itself, even though its goal is never been achieved. 'Explanation is ex– plicated, that is unfolded'. [but seebelow -- no heretical denunciatins etc]

This is relativist, or rather 'one should prefer the less loaded term relation-ist' and it helps solve the other problems historicity and reflexivity

Pre-relativism, there was a difference between providing an explanation and just documenting the historical circumstances, but now there is no difference. Critics have suggested that this '"escapes socio-historical contingencies"', but this only means that longer term factors have been more important, like networks of power, or networks of world economics, capitalism or zeitgeist. Historians don't like this because they don't like ontology. [and prefer empirical accident, contingency etc]

Historicism is avoided because there is no definition of the things themselves involved. There used to be a clash between historicism and explanation or theory since there was thought to be a distinction between the history of people, of contingencies, of the effects of time on the one hand, and a theory or a science of something timeless or eternal or necessary on the other. ANT says there is the science only of the contingent, that becomes necessary as a local achievement through the growth of a network. The distinction between things, descriptions and explanations is dissolved and this is 'what an explanation or explication is and what has always been the case in the so-called hard sciences' (377) [values smuggled in then].

Reflexivity is not the main goal but it is added since the frames of reference are accessible to the actors. The problems usually arise with the 'epistemological myth' of an outside observer providing a superior explanation for the participants who can merely describe, but now there is no privilege, merely another frame of reference, the observer's, 'on a par with all the other frames of reference'. To move from one frame of reference to the next requires work just as with any other actor, 'to explain, to account, to observe, to prove, to argue, to dominate… To "network"'. In this way reflexivity is not a problem but an opportunity. Privileged metalanguage has to be abandoned [not easy -- and not done in this case, of course] but this was usually only a resource for 'small points limited to very specific local[itieis] — campuses, studios, corporate rooms'. Many frames of reference means many metalanguage strategies, but the only one required is better understood as an 'infra language' the equivalent of a Lorentz transformation, a 'translation' in ANT terms this produces only a 'one-shot account exclusively tailored to the problem at hand, and Callon calls them "disposable explanations' as networks expand.

This is common sense. New accounts do not subtract anything in ANT. The selection principle is not whether there is some fit between account and reality but 'whether or not one travels from a net to another'. No metalanguage helps here. Epistemology does not lead to moral relativism but rather to a 'stronger deontological commitment: either an account leads you to all the other accounts — and is good — or it constantly interrupts the movement, and letting frames of reference distant and foreign [sic] — and it is bad' (378). More mediating point are produced, good, or are reduced, bad. Reductionism or areductionism — 'that's the highest ethical standard for AMT' and more discriminating than searching for epistemological purity, foundations, or moral norms: 'demarcation is in fact an enemy of differentiation'.

So ANT brackets out society and nature to consider only meaning production. Then it breaks with conventional semiotics and grants activity to semiotic actors turning them into 'new ontological hybrids, world making entities'. It builds an empty frame to describe how entities build their worlds and it retains from description only a very few terms 'it's infra language — which are just enough to sail in between frames of reference and grants back to the actors themselves the ability to build precise accounts of one another by the very way they behave; the goal of building an overarching explanation… The search for x applications, that is for the deployment of as many elements as possible accounted for through as many metalanguages as possible'

ANT is therefore 'about a network tracing activity rather than traced networks themselves. 'No net exists independently of the very act of tracing it, and now tracing is done by an actor exterior to the net. A network is not a thing but the recorded movement of a thing'. There is no need to distinguish what moves inside networks, into, for example pieces of information, words, or bodies since nets existed before such distinctions became important — what is important is like the circulating object in semiotic texts, something defined 'by the competencies [it is] endowed with, the trials it undergoes, the performances it is allowed to display, the associations it is made to bear upon, the sanctions it receives, the background in which is is circulating'. These and its persistence are 'the results of the decision taken through the narrative programs and the narrative paths'

Is ANT limited to the world of text and discourse? Is there a gap between text and context? For ANT, this is 'no more than a slight bump along the net' caused by a previous divide between nature and society and discourse (379). Instead there is a continuity, multiple plugs between the circulating objects, the claims outside the text in the social and what the actants themselves really do. The circulating object goes on circulating, things like society text and nature are 'arbitrary cutting point on a continuous tracing of action' themselves part of the obstacles, trials and events leading to attributions of textuality or sociology or natural reality 'not part of what makes the distribution'.

This could be described as a 'generalised narrative path'. All this would be to imply that everything is a text. To call it a force or energy would imply that everything is naturalised, and to call it a social interest extends social action to nature and to texts. ANT originally played with 'generalised symmetry' and used whichever words are connoted in one realm to describe the others to show the continuity of network and the disregard for any gaps, but this could lead to misunderstanding and accusations of Social Constructivism, naturalism, or French belief in the textualisation of everything. The point is that all of the terminologies refer to the filling in of what is between the network, 'and which one is chosen or rejected makes no practical difference, since nets have now "in between" to be filled in'

What circulates should be seen as both co-determined and transformed', what Serres has called a quasi-object. Material objects are poor examples because they are not transformed Quasi objects are moving actants that transform those who do the moving and the moving object. Stability is exceptional. This males ANT different from all other models of communication, which are human- or languaege-centred. or praxis-cemtred. There are also different kinds of morphism -- anthropomorphic but also zoo- and physio-morphic,logo- or techno- ideo-morphic. Again paths with restricted access to humans or mechanisms are exceptional.

So we can destroy 'spheres and domains', reinstate heterogeneity and interobjectivity. Associations are not differentiated though (politics?). This needs to be done,to 'specify the types of trajectories that are obtained by highly different mediations

Still classically in terrain of natural science where network indeed expanded or transformed. Transformatiopns according to accepted mathematical procedures in the case of Lorenz or Einstein. Others debatable eg marxist accounts, even religious ones. Not accepted are attempts to destroy and replace the whole network, condemn it as heresy as harmful or oppressive. This is what happens in soc sci or phil, -- happens with p-humanism!