Across the Great (and Small) Divides
Stephan Fuchs/Douglas A. Marshall

Zusammenfassung: Postmoderne Theorien können den Dualismus von Natur und Gesellschaft nur anthropozentrisch und aristotelisch überwinden. Die konstruktivistische Theorie des Beobachtens ist dieser Aufgabe besser gewachsen. "Person" und "Ding" können dann als Resultat von Attributionen begriffen werden, die mit den Sozialstrukturen des Beobachtens variieren. Dinge und Personen sind demnach keine ontisch separaten Gegebenheiten, sondern durch ein Kontinuum von Übergängen verbunden. Von dieser Warte aus kann die Besonderheit der Moderne darin gesehen werden, daß mehr und mehr Beobachter mehr und mehr Verbindungen zwischen Natur und Gesellschaft herstellen.

Structure and Observation

A central observation of Luhmann’s (1997, 92ff., 134ff., 776ff.) work is that any order must be accomplished, or does not come about at all. This latter possibility is actually now more likely than ever, given that modernity can no longer rely on kinship, the fatherly Prince, the Good Society, shared standards of rationality, or the moral community of all well-meaning people to serve as "skyhooks" of order, to use Dennett’s (1995) metaphor. Neither can it be expected that all arguments will any day now converge on HaberMaster reasons in discourse. Order is improbable and precarious. Under conditions of multiple contingency, there are no guarantees that independent experiencing and communicating "systems" will couple and coordinate their experiences and communications at all. Contingency makes order improbable and unstable, and it makes consensus precarious and revokable at any moment.

If they do emerge, order and consensus are never widespread in the beginning; let alone universal, or even a priori. They can only start as local, temporary, and contested selective couplings. Under certain rare conditions, however, they might extend over time and space to include those who were not present when the initial arrangement was made. These strangers have no memory of, or obligation to, the original settlement. Their commitment cannot be relied upon and taken for granted. They could, and often will, reject a settlement and re-negotiate a different order. A critical problem, then, is when one or the other of these two outcomes will tend to occur.

We start, again, with Luhmann, and observe that all events are local. All events disappear; all events disappear immediately and forever if they are not followed by similar events; the method or machine that produces similar events is the key to producing structure. A "machine" is defined here, in loose connection to Heidegger, as anything that does what it does without reflecting on how this is possible. Machines aim for indifference toward variations to assure repetitions and condensations. Machines can be Human or Nonhuman, conversational or organizational; the important difference is habitualization, not the substance or essence of that which is being habitualized, domesticated, or caged.

Machines of this substrate-neutral kind can generalize local and temporary couplings beyond their local circumstances and points of origin. But this never happens by necessity or automatically. It must be made to happen, or does not come about at all. Seen sociologically and constructivistically, universality is an outcome, not cause, of an order being extended across space and over time(1). When this happens, the micro turns into the macro, a local convention turns into the law of the land, and agency turns into structure which constrains further agency. The "intrinsic force" of the better argument is never forceful enough to make order binding, although those who make order happen will, and must, later invoke just such rational forces. "Rationality" is a reconstructive justification of a frontstage observer, who explains success by truth, and progress by reason. History notwithstanding, order either happens or doesn’t. If it happens, someone or something made it happen, usually against resistance and competition from other efforts (White 1992, 9).

Once order is accomplished, "against all odds", it must be maintained against entropy, or the disappearance of structure and coupling. In the last analysis, social entropy results from the fact that it is always more work to maintain relationships than it is to let them disappear. There are many more ways not to be connected in relationships than to be so connected. Modernity increases the ratio of possible, yet absent, to actual ties and connections dramatically. Much of modern society is "structural holes," or the absence of possible ties within a network (Burt 1992). Compounding the problem, order must not only be stabilized against noise, but also against alternative orders, which may profit from the "normative force of the factual," especially when they have been around for several generations. What is more, some social fields, including science and avantgarde art, even encourage and reward disorder in celebrating innovativeness and creativity. Such fields constantly produce and search for noise and disorder, most of which must then be discarded and demarcated from the few "true" advances and "real" breakthroughs. Modernity may not be more rational, more true, or more objective than previous social formations, but it certainly is more restless, lacking any possibility to constrain its own future in a stable ontology or teleology.

To the extent that it does emerge and endure, a social structure effectively constrains the "underdetermination" and "interpretive flexibility" that are endless and "fundamentally irresolvable" only in abstract philosophical principle, not in actual social and scientific practice. Flexibility often happens after the fact, when more can be seen, and when alternatives can be contrasted with what actually happened. Some postempiricist philosophers, and some "soft" constructivists, have inferred relativism from underdetermination, but this ignores that not just anything goes once a structure is in place that effectively constrains what is possible next. Structure comes with history, and together they set the parameters for the future. The best predictor of a system’s current state is its immediately preceding state, ceteris paribus(2). Revolutionary ruptures remain possible, but become more improbable and rare. In Kuhnian terms, meaning incommensurability and relativism do sometimes occur, but are not the standard cases in any culture. Incommensurability is more frequent between cultures, but its actual degree depends on the extent of interactions between them. When there is some interaction across the borders, "trading zones" emerge in which "pidgins" and "creoles" mitigate incommensurability (Galison 1997, Ch. 9).

Sociologically, relativism indicates a crisis in social solidarity and cohesion, not a "theoretically" unlimited number of "logically possible" alternative worldviews. To be sure, an observer of a structure can perceive more, and different, alternatives or states, especially as a detached "theorist" or remote "critic." He can observe what happened, and contrast the actual outcomes with possible ones, including ideal speech situations and other utopias. At the same time, this observer observes within his own "unmarked state," which consists of his own modes and frameworks of observing. The hows of his observing remain invisible to himself at the time he observes what he observes. There are always fewer alternatives in practice than in theory.

Variable Constructs and Constructions

Since they do not follow philosophical legislation as to what science, or any culture, can and cannot do, various networks and their cultures are not of one piece, as if molded after one essential logic, and demarcated from their environments once and for all by some fixed and stable "demarcation criteria." Consider science. All of it is, of course, "constructed" – in the sense that there would be no science if no one was doing it. But, by itself, this is a rather trivial point. The reason this point has attracted so much angry publicity and polemic is that "social construction" has been confused with referential inadequacy. That this is a non sequitur can be seen in the fact that cars, refrigerators, and children are constructed as well, but no one makes a big epistemological deal out of that, and starts doubting their reality. Constructions are real as well. In fact, some constructions appear unconstructed to those who live in and by them. They turn into the stuff that "lifeworlds" are made of, including unproblematic background assumptions, institutions, and paradigms. In a nutshell: While all networks and cultures must be constructed, only some of them are constructivist. This explains why skepticism and realism are distributed differentially across the intellectual field and over time (Collins 1998).

Constructivism has generated much angry controversy and hostile polemic (Fuchs 1996). At bottom, the current science wars in academia rest on failures to distinguish between levels of observation, or between the what and the how of observing (Luhmann 1992). In observing how some science constructs, the constructivist observer cannot contribute to that science, only to his own, since his observations are not to be fed into the network that is the science he observes, but into the network surrounding that second level observer. This network can be, for example, philosophy, sociology, or history of science. This means that constructivism has nothing to say about the "truth" of the science it observes. However, insofar as the networks into which such observations are to be fed are part of science as well, they raise their own truth claims – constructivism understands itself as an accurate account of some science, superior to, say, traditional epistemology.

As a result, first- and second-order observers are united by the "code" of truth and its systematic elaborations, such as objectivity, empirical adequacy, logical non-contradiction, and the like (Fuchs 1997). From here, we can go one step further, and suggest that what matters for second-order observing is not the truth of the first level, and not even social construction per se, but how social and cultural constructions vary and covary with other variables, and how this affects the outcomes of work, including philosophical rationalizations of practice. Constructions vary, and they co-vary with other variables, including the social structures in which such constructions take place.

As a result, we expect to observe a wide empirical variety of networks and their cultures or self-descriptions, differing in the way they construct and re-construct their own realities. For example, at the height of their imperial success and explanatory confidence, some well-established and mature networks produce solid realism and normal science, while more fragmented, insecure, and loosely coupled networks behave more skeptically and relativistically.

Overcoming essentialism and allowing for variation avoids the "New Wars of Truth" between science and its constructivist observers by suspending epistemological arguments about the "correctness" of various philosophical rationalizations of practice. For science, of course, science is about reality, not society; this is the "unmarked state" of any observing which cannot observe itself at the same time (Luhmann 1997, 49). Any science, including sociology of science, explains itself as the outcome of its chief epistemic virtues and merits. The role of philosophy is to rationalize such self-descriptions into a coherent frontstage account that can convince outsiders and novices.

In addition, a structural constructivism that is sensitive to variations, including variations in itself, can observe various philosophies as the ideologies of cultural workers located in different positions in time and social structure. For example, instead of asking: is relativism or realism the "correct" account of science?, structural constructivism turns relativism and realism into "dependent variables." Then, one can observe when, and for whom, a certain science appears as universal and objective truth, and when science is seen as a local social construct. Allowing for variation, we could investigate when and why one or the other outcome occurs, and which social forces either extend or restrict the extension and stabilization of order. Once variation is introduced, the attention shifts from "science as such" to various empirical sciences doing very many different kinds of work, their "subcultures," local settings, and historical changes. This does not mean that there is no unity to science at all, only that this unity must be observed at a very abstract level. At the same time, the unity of science exists only for an observer of science within science, and so is not itself unitarian, but a part of science, and therefore "partial."

We need to overcome essentialism, get rid of the remaining Aristotelianism and agency humanism in social science, turn natural kinds, such as "rationality," into dependent variables, and replace substance by relation everywhere, including in persons (Fuchs 1999). One benefit from allowing for variation is to see various epistemologies as the "ideologies" or "rational myths" of intellectual workers, located in different positions and networks in time and social space (Fuchs 1993a). Another benefit concerns the unhelpful and hopelessly outdated Science/Humanities division and opposition. The traditional explanations of cultural and epistemological diversity, centered around methodological and ontological divisions between the Two Cultures and the Double Hermeneutic, no longer suffice. While the Humanities/Science division may survive in academic organizations, its philosophical elaborations are becoming less convincing all the time.

For example, important variations in scientific cultures cut across the Nature/Society divide (Schneider 1993). With Schneider, we would expect to observe "soft" enchanted physics (e.g., 16th century alchemy) and "hard" realist literature (state socialist art). New research networks are generally weaker than older ones, whether they deal with Nature or Society. Some soft fields have harder catnets inside of them, often crystallizing around some solid equipment with black boxes. Some research frontiers in hard science behave rather softly, exploring new areas of uncertainty. While the traditional "object" – or, in more modern terms, the "referential ecologies" of certain specialties – may make a difference in all this, it is impossible to tell just which difference they make before the science we want to explain informs us about its object. As observers, we have no independent access to some referential ecology prior to specialist investigations, for this would mean we were specialists, not (specialist!) observers of specialists. Therefore, sociology cannot explain science as a result of that science’s referential niche in the world.

In sum, constructivism implies that we cannot support or advance ontological explanations for variations in intellectual cultures because we cannot participate in, only observe, these cultures. Sociological explanations of a science must differ from the explanations that science offers for itself, if only to avoid redundancy. The sciences’ explanations of themselves as corresponding to some reality and reason are not a premise for constructivist observing. Sociological constructivism must bracket truth claims. This indifference was already one of the basic principles of the Strong Program. Sociology of science is sociology, not the science it observes. It cannot contribute anything to the science it observes. It cannot settle a science’s controversies, only its own. Sociological constructivism advcances its own truth claims on a second level, but cannot resolve truth matters occuring on the level of the science it observes.

Postism: Words and Worlds(3)

One of the deepest divisions in academic culture separates the Two Cultures of Science and Humanism. In cultural and science studies, especially their "postist" wings, there has been much excitement lately about breaking down modernist barriers, collapsing metaphysical distinctions, inverting and subverting rationalist hierarchies, narrowing the gaps between opposite ontological poles, and deconstructing Western Enlightenment metaphysics. With great drama and fanfare, the "end of the modern age" is being announced, usually without a good sense of what will or should take its place, or what exactly the "postist" changes are that make the present or imminent future so radically different from the classical modern age.

In the larger culture, "postism" is a construct of cultural workers who specialize in the manipulation of symbols, texts, and the commodification of signs. Therefore, postmodernism is concentrated in the tertiary sector, especially in the reflexive and avantgarde branches of capitalist aesthetics, such as fashion, architecture, cinema, or high art and literature. The new professions specializing in images seem to be particularly receptive to postism because their work consists of creative and reflexive manipulations of symbols, which gradually seem to lose their connection with what they used to represent, and become a "virtual" reality in its own right. This loose play of free-floating signs leads to a semiotic skepticism where signs seem only to point at each other, never to an underlying reality. All that is solid melts into air.

In fact, however, this postmodernism is rather modern. Contingency and antifoundationalism already are major themes in modern sociological theory. If the classics "converge" on anything, it is not, via Parsons, "general values," but the growing suspicion that modernity has irrevocably lost its metaphysical anchors and bearings. This suspicion is strongest in Weber’s Nietzschean moments, in Durkheim, and the later Mannheim. Sociology is precisely the result of the realization that there is nothing transcendental, and that everything that exists is empirical. Postism has a tendency to infer arbitrariness from contingency. But even if there are no true and transcendental universals, some institutions are still stronger than others, still reach further than others, and endure longer than others. One might say that transcendence is a rare and emergent property of immanence. There are some modern institutions, such as the Liberal Self, which have successfully ruled out alternatives – for the moment, and until further notice. That is, transcendence is itself the improbable outcome of empirical stabilizations and condensations of order. Modernity makes such stabilizations more unlikely by increasing empirical and historical diversity and contingency.

The postist attitude or mentality is that of a remote and detached observer, ironicizing from a distance what appears only natural and valid elsewhere. The ironist will do well in the company of other ironists, but not in a fundamentalist religious sect, not in a Senate subcommittee hearing, and not as a speaker during an official ceremony. Within the academy, the ironist and skeptical observer emerges in intellectual fields with very loose coupling, weak policing, and high practitioner discretion. Such fields are very conversational and textual, soon imagining the whole world as a large text. Deconstructivist irony is less of an option when work must be done fast in intense competition with others who also try to get more grants to do more work earning more grants, and so on. Irony is not an option when one needs to justify one’s budget proposal, respond to outside critics, or teaches students. It is the ironists who are skeptical of physics, not the physicists. Many ironists are very serious about irony.

In academia, postism expresses a vague and generalized skepticism toward foundations and truth, supported by multicultural politics and the ascriptive entitlements of standpoint epistemologies. This skepticism extends from the "crisis of representation" to the "illusions of presentism," from the "end of logocentrism" to the "Death of the Author" (see Rosenau 1992). What runs through these motifs are rather idle doubts about the possibility of objective knowledge per se. Do not authors continue to claim credit for the discovery that authors don’t really exist? This sort of skepticism makes it difficult to get any work done, and so we would expect it to be most widespread in intellectual fields that don’t.

Indeed, postism is most prominent in soft fields with little equipment and machinery, such as literature and literary criticism (Fuchs/Ward 1994). In sociology, postism is almost hegemonic in anti-hegemonic critical theory, gender, and cultural studies, but shrugged off as an annoying disturbance in more solid and research-oriented sectors, such as survey centers, status expectation research, or the network exchange tradition. Correspondingly, sociology has both postmodernism and postmodernity. Postmodernity tracks empirical changes in society with the usual methods, whereas postmodernism is more of a global ideological and political attack on science and representation.

In addition, postism is prominent whenever traditional methods become unworkable or unrewarding, such as in anthropology, where the tribal societies are disappearing together with the classical realist ethnography. In literature, there are now many more critics and epigones than classical authors and texts, resulting in an emphasis on "theory" and "reflexivity." Habermas (1990, 192) says that, by questioning the authorial privileges of classical intentionalist hermeneutics, postism also raises the status of the critic vis-a-vis the classic.

What Is Modern about Modernity?

The most famous postist approach in science studies is actor-network philosophy(4). In the footsteps of poststructuralist semiotics, Bruno Latour and his followers in the very influential "actor-network" network have set up their own favorite target for collapsing and inverting. This is the "Great Divide" between Society and Nature, or human vs. non-human "actants"(5). The Great Divide was erected in the 17th century, in the debates between natural and political philosophers, most prominently Boyle and Hobbes (Shapin and Schaffer 1985). The 17th century established the "constitution of truth" calling in the modern era. This constitution denaturalizes society and desocializes nature, setting up the two separate poles like two branches of Government: "I define a world as modern when the political constitution of truth creates those two separate parliaments, one hidden for things, the other in the open for citizens" (Latour 1993, 15).

According to actor-network "theory,"(6) in each of these branches a regime of representation is set up. Science represents things natural that cannot speak at all; the state represents citizens that cannot all speak at the same time. In these orders of representation, Nature and Society have been separated and purified, but can still be called upon to explain each other: Nature explains Society in naturalist and physicalist theories of knowledge and behavior; Society explains Nature in social constructivist accounts of science. That is, the modernist transcendence of Nature is also immanent, and the immanence of Society is also transcendent: Nature can be controlled and manipulated; Society cannot be changed at will because it is larger than the sum of its parts. Somehow, this dialectic allows "us moderns" to rule the rest of the world.

At the same time, modernity separates itself from premodern times by means of yet another Great Divide. This one consists of a series of retrospective revolutionary breaks with the past. These breaks paper over an essential historical continuity that links all "collectives" whose Nature and Culture still form seamless webs or networks in a unified cosmological order. To compensate for the resulting loss of spirituality, the moderns remove God from the external order of Nature, and internalize Him in their Soul during the Reformation. This is how clever Latour believes "we" are, whoever this "we" may be.

After Latour and the other persings(7) have shown how this modern constitution was drafted and protected, they go on to say that it has never existed, really (1993, 39). In fact, "modernity has nothing to do with the invention of humanism, with the emergence of the sciences, with the secularization of society, or with the mechanization of the world" (34). This is a tall order indeed, announced with the chuzpe typical of this school, disregarding centuries of evidence with a quick gesture. Instead, so the story goes on, human and nonhuman actants have always formed "technoscientific" networks that cut across the Great Divides, with no respect for metaphysical and ontological boundaries. Nature and Society have always been co-constructed through coaltions and alliances that know no borders.

These coalitions and alliances produce "hybrids," "quasi-objects," and "cyborgs" that are never purely social, purely natural, or purely textual and discursive. These entities populate the fabulous "Middle Kingdom," before the invention of both Society and Nature. It is this Kingdom Latour wants to rule. Needless to say, it is not the quasi-objects that have elected him to be their spokesman. As all Kings, Latour is self-appointed.

While modernity seems to have effectively separated the two realms through the work of "purification," this has paradoxically accelerated the process of mutual interaction and diffusion between humans and non-humans. Actually, it is in this acceleration that the power of modernity lies. The moderns deceive themselves; they claim separation but practice integration, and must wait for Latour to explain to them what they have been doing all along. Latour is, in this sense, the last great modernist and destroyer of deception, despite his self-identification as an "a-modernist."

From an a-modern perspective, "we" – whoever that is – "have never been modern." Rather, the modernist dualisms and distinctions are the work of "purifications," once "translations" and "mediations" have crossed the boundaries of the Great Divide and created the hybrids and quasi-objects of the Middle Kingdom. Nature and Society form a seamless web that can be untangled and separated only artificially, and only after the fact of boundary-crossings. Neither side of the Nature-Society pole can be privileged in any explanations. Instead, what must be understood are the processes of "enrolment" and "interessement," in which certain actants manage to establish "centers of translation" from which they claim to be the spokespersons of integrated technoscientific networks. Before that happens, however, things natural contribute as much to the construction of hybrids as persons social, which makes them co-equals in the process of defining and assembling reality.

Against social constructivism, actor-network persings maintain that the social – which they falsely and narrowly reduce to the "interests" prominent in the explanations of science by early Edinburgh Strong Programmers – cannot "explain" the natural since interests and interactions are not prior to technoscientific outcomes. Rather, interests are constructed and re-constructed in the very same process of producing these outcomes. Following scientists around, they are observed to assemble "heterogeneous" networks of support which are made from a variety of things social and natural, without clear distinctions.

Some Problems in the Middle Kingdom

There are some serious flaws in this philosophy, the most important one being that it is still philosophy. In part, this flaw follows from a narrow and distorted misconception of the social and sociology. To begin with, it is inaccurate to reduce sociology to interest explanations. While these were popular in the work of some Strong Programmers in the 70s, "interests" do by no means exhaust the sociological arsenal. Social science can also not be reduced to the intentional actions of persons. Sociology does not equate society with subjects or persons, and it does not say that social structure is not itself constructed (Latour 1993, 54).

In their unsplendid Parisian isolation, Latour and his followers disregard US organization and network science, which have always combined things and people, but without adding any special metaphysics and ontology. This is especially true for technological and contingency theories. In network and organizational accounts, things and objects have always appeared – as raw materials and means of production, as congealed and reified structures, as physical constraints on communication, as technological cages of complexity, as dramatic simplifications of natural processes, or as sacred totems of group solidarity. Actor-network philosophy also pays no attention to the Neodurkheimians, who have long shown that modernity isn’t that modern after all, but remains premodern in such structures as urban tribes and everyday rituals.

Investing non-humans with independent agency in a "symmetrical anthropology" also comes dangerously close to anthropocentrism. The principle of symmetry is paradoxical, because such "investing" must still be done by someone, and cannot be done by the things-in-themselves. As Collins and Yearley (1992a), amongst others, have observed, generalized symmetry in fact restores realism by making the "inner properties" (Latour 1993, 52) of objects a factor in accounts and explanations. Latour wants to reveal and rescue the innocent thing-in-itself, before all representation, delegation, translation, and construction. But the dramatic fanfare with which this antimodernist metaphysics is announced – we are being assured that nothing less than a "Copernican Counter-Revolution" is happening here (Latour 1993, 76) – conceals the rather old-fashioned character of the empirical work produced under its umbrella. This work "takes us directly back to the scientists’ conventional and prosaic accounts of the world from which we escaped in the early 70s" (Collins/Yearley 1992a, 322). The Pasteurization of France, for example, tells a rather traditional individualistic story of a scientist-hero, whose cunning machinations and manipulations almost singlehandedly transform French society.

Latour remains squarely stuck in metaphysics because, in trying to overcome metaphysics, he accepts the very metaphysical way in which the Nature/Culture dualism has traditionally been framed. Instead of really collapsing the poles, "generalized symmetrical anthropology" performs an even grander metaphysical trick: It sets up a mega-pole, prior and even more fundamental than to the other two poles. This mega-pole, the most original and truest reality underneath all secondary constructions, consists of collectives of quasi-objects and quasi-subjects. Instead of saying, ’there is only nature’ or, ’there is only society,’ this amodern metaphysics says, ’there are only networks and hybrids’ – and their center of translation is in Paris, at the Ecole des Mines.

The Ecole des Mines is part of the academy, not the Middle Kingdom. Far from anchoring to a primordial bedrock reality, generalized symmetrical anthropology is still simply a part of "science." That is, it is published in books and articles, talked about at conferences, cited in other books and articles, and taught to students. That science is part of society, not nature, and not the Middle Kingdom. Actor-network research occurs only in and as society, not nature. The symmetrical anthropology must still be communicated to, and rewarded by, other (social) scientists, not scallops (Callon), bubble chambers (Pickering), or the anthrax bacillus (Latour). The fact that scallops and bubble chambers cannot communicate may not be a good reason to draw a deep distinction between Society and Nature, but it does put a damper on attempts at telling a scientific story from their perspectives.

This means that "the social" is still "privileged," because we don’t learn anything about scallops if there isn’t any research about them; we don’t know about the air pump except from reading Shapin and Schaffer and other historians, and we would know nothing about hybrids and the parliament of things were it not for good old Harvard University Press, that crown jewel of modernism, publishing Latour’s books. These basic social realities have not changed at all, despite the symmetrical ontology of the new "parliament of things" having supposedly liberated objects from their encroachment in sociology. Symmetrical anthropology cannot go to a deeper level before and underneath science, nature, and society because there is no such deeper level.

Actor-network persings do not, of course, give objects their own voice. In his famous analysis of scallop domestication, Callon (1986) claims to give the scallops equal ontological capacities and status with humans. But without scientific communications and representations, the scallops simply do not enter the picture. "This means that when the scientist says ’scallops’ we see only scientists saying scallops. We never see scallops scalloping, nor do we see scallops controlling what scientists say about them" (Collins/Yearley 1992b). One cannot tell the story from the side of the scallops, just as one cannot know what it feels like to be a bat. Things are still not speaking for themselves – that would be truly antimodern. In effect, then, Callon must either claim that the scallops speak through him, or through the scallop scientists. In the first case, he turns from a sociologist into a scallop scientist, albeit a bad one, due to lack of training and expertise. In the second case, he repeats what scallop scientists are already saying, adding nothing to their picture of the world. Since Callon does neither of these, he actually proceeds in a quite straightforward sociological way: He brings a descriptive and theoretical apparatus – consisting of the familiar actor-network repertoire such as "translation," "enrolment," and "interessement" – to bear on a scientific episode. This apparatus is neither that of the scallops, nor that of the scallop-scientists, but that of the sociologist. Instead of a "Copernican counter-revolution" and an "amodern political constitution of truth" we get old sociological wine in new rhetorical bottles.

To be sure, the reason for the impossibility of perfect symmetry is not some "essential difference" between things natural and things social, between subjects and objects, action and behavior, or science and hermeneutics. In this regard, our argument differs from that of Collins and Yearley, and other "double hermeneuticians" and interpretivists, who want to maintain the special humanist privileges and distinctions for mysterious spirits such as intentionality, subjective meaning, action, Verstehen, and the rest of the romantic arsenal. We do indeed need to get rid of essentialism and overcome dualism, but the way to do this is to overcome metaphysics altogether, not to replace it by another, antimodern one. While there may be no "grand" metaphysical and ontological differences between things social and things natural, people, including Parisian actor network theorists, have usually very little trouble taking the "intentional stance" (Dennett 1987) toward some systems but not others. It is this sort of variation we need to explain. To do this, metaphysics is not necessary; sociology will do just fine.

Constructivist Observing

From a sociological perspective, things social and things natural are not separated by a grand ontological divide, but by more or less contingent, though never arbitrary, social distinctions. These distinctions distribute intentional and causal effects unequally across the landscape of various populations. Sociology, practiced as structural constructivism, can identify some variables that make a difference in such attributions. Since there are many variables operating at the same time, the effect of each depends on the effects of all the others, so that all following causal arguments obtain only ceteris paribus.

Our basic premise is that "action" and "behavior," "persons" and "things," "Nature" and "Society," "science" and "humanism," and the other dichotomies are indeed not opposite poles of Being, separated by an unbridgeable essentialist gap. Rather, they are social devices of description and explanation that covary with other sociological variables, such as the status of observers, the conditions of observing, and the degree to which an observed system has been rendered predictable through normal science.

All other things being equal – which they never are – intentional interpretations and Verstehen are more likely to occur when observers and observed are socially close, and when the observed are few in number. Then, the observer is more likely to use such "soft" and very time-consuming methods as participant observation and Verstehen. One can verstehen – but not that many people. Therefore, when observer and observed are separated by some large distance, and when there are very many systems to be observed, the observer is more likely to conceive of the observed behaviors and effects as driven by impersonal causal forces, to be measured by quantitative formulas, and explained by general theory. Distance and size are, of course, variables, which means that we are dealing with a continuum here, bracketed by "understanding" and "explanation" as opposite ideal types, and Latour’s "hybrids" or "quasi-objects" somewhere in between.

One extreme pole is the pure understanding of one person: love. The opposite extreme pole is pure explanation of all organisms: genetics and molecular biology.

Allowing for variation makes it possible to explain when systems move across the continuum, when they tend to become more person-like or more thing-like, and when they occupy some intermediate position, or the Middle Kingdom. In addition to distance and size, another factor is time. Over time, some systems tend to get better understood and routine, and so move closer to the mechanistic and deterministic thing-pole. Their behavior gets more predictable and, as a result, "intentionality" and "free will" or "decision" decrease. At the same time, time will be counteracted by social closeness and moral boundaries around groups (Smiley 1992, 12, 114). Within those boundaries, intentionality is a stronger assumption than outside. Whatever is far outside the moral boundaries separating "us" from "them" acquires a more thing-like character, implying that "they" cannot participate as equals in "our" constructions of "their" behaviors. However, this may change over time as well, since boundaries are not static and inflexible.

Such mutual exclusions are characteristic of ideological observing, for example. Ideological observing moves the observed closer to the thing-pole of the continuum. The opponent is caused by social forces without being aware of them. If "they" are stuck in ideology, they are unwilling or unable to see through their self-constructed maze of deception, and need to be explained from the outside. Then, "they" become a target for "our" science and explanation, not equal hermeneutic partners in conversation. The explanations ideological enemies give for themselves are symptoms of deception, and so cannot be "in the truth."

The cases at the borders are ambivalent and ambiguous – these are Simmel’s strangers and Kuhn’s anomalies or Latour’s hybrids. On the one hand, strangers are not well known; their mysteriousness and exoticism call for interpretation, rather than explanation. On the other hand, they are not really part of the group, and so are objects rather than subjects, or some of both. That is, object- and subject-status are ascribed, but ascriptions covary with other variables, such as time and distance, which implies that ascriptions will change over time and with interaction as well.

Consider a more concrete example. One does not normally understand one’s spouse as an impersonal system, driven by causal forces, and not being responsible or accountable for her actions. This does not mean that her actions cannot be explained by science, only that science does not reach into love. What the spouse does may indeed be explainable as the result of chemistry, neuroscience, or social class, but explanations of this kind do not work in close and intimate relations. Here, "individuals" occur, and each is supposed to appreciate and understand the other as "really special," not "just" as a particular configuration and outcome of empirical forces and causes. In intimacy, agency terms are more expected and appropriate; not even the hardest-nosed neuroreductionists could approach their wife and kids as a neural network, algorithm, or artificial intelligence, at least not while and during intimate encounters and interactions. As an intimate relationship breaks up, of course, mutual explanations and attributions may change, moving once again closer to the thing-pole.

Scientific explanations of spouses and other intimates may become more serviceable when making sense of some behavior in the common agency terms becomes increasingly difficult. "Insanity" is one concept that signals a break-up of a moral community, when insiders who used to have special privileges in accounting for their own behaviors turn, to some extent, into outsiders and objects for some sort of "scientific" explanation. When social scientists explain the behavior of large crowds, or of structural systems such as states, they conceive of reality as more object-like and physical. What matters is not the essential properties of different natural kinds, but the social contexts in which different observers attribute different faculties to systems for different pragmatic purposes.

Interpretation and explanation also vary with the amount of perceived uncertainty. When some observer is very uncertain about the erratic behaviors of some rather hard-to-predict system, he is more likely to assume that that system has an internal center where it makes decisions and choices according to unobservable rules, beliefs, and preferences. In the movie Backdraft, the fire inspector, played by de Niro, muses that a fire does not grow because of the physics of flammable liquids, but because it "wants to." Agency is being attributed here to the behavior of fires as a result and expression of uncertainty and unpredictability. Another way of saying this is that "agency" is the expected or observed capacity of a system to surprise its observers. Upon being surprised, the observer might try to get closer to this system by softer and more interpretive methods. He might try to develop a "feeling for the organism," and to understand this system "from the inside" – as if it had agency. "Agency" is a moral capacity that a system receives from an observer who is not, at present, entitled or able to make sense of that system in deterministic terms.

Outside of close relationships, most observers will probably try to construct deterministic explanations first, because these are simpler, faster, and more generalizable across classes of systems. Deterministic explanations economize on explanation costs. They are more accomodating to the "bounded rationality" of all observers, or their limited ability to deal with complexity and novelty. This is especially so for organizational observers, because the organization sets the parameters for how and what its workers are supposed to observe, what they are expected to ignore, and because organizations try to simplify and routinize as much as they can. However, when this proves infeasible or inappropriate for some reason, when exceptions and surprises accumulate, these very same systems may be granted faculties such as "spontaneity," "creativity," and "originality." In this case, the organization and its observers make special amends to the rules and routines, such as special programs for "gifted" students who stand out of the pack, and cannot be processed by the routine methods.

In contrast, the observer will tend to become a "scientist" explaining the behavior of his systems from the outside when that behavior is serviceable under the assumption that it is simple, repetitive, and invariant across time and place. For this, it does not matter whether the system is a person or a thing, since "personhood" and "thingness" are the outcomes, not causes, of observations, attributions, and cultural work. At least, this particular way of assigning causes is the specific contribution of sociological constructivism.

An example for an account located toward the middle of the thing-person continuum is rational choice. Rational choice conceives of actors as "persings," combining the "soft" ambiguity and uncertainty of individual preferences with the "hard" maxim that all actors will optimize. Rational choice-type explanations arose when markets increased the numbers of actors one had to deal with. To assume that everyone is behaving "rationally," regardless of individual differences and idiosyncracies, is a strategy chosen when it is no longer possible or necessary to "empathize" with all of one’s partners in exchange. The algorithmic machine of rational action is a radically simplified construct that can be chosen in circumstances when one observes and interacts with very large numbers of strangers. One cannot possibly know or care what all of them actually think or feel, and so everyone assumes that everyone is behaving rationally.

In contrast, "thicker" descriptions and explanations will be chosen when observer and observed are socially close, or even intimate, ceteris paribus. In such cases, an objectifying attitude would violate the moral expectations and taboos of such associations. One grants the other a "rich inner life" that cannot easily be algorithmically compressed into a standard formula, such as self-interest or stimulus-response. This rich inner life also allows for surprises, which preserve the "magic" of the relationship. This magic is a vital Durkheimian sacred object, which would be violated by a "scientific" attitude. This may be the reason why scientists are not considered perfect spouses. In a variation of Black’s (1976, 41) law of law, we could say:

There is more explanation between strangers; there is more hermeneutics between intimates.

However, some strangers deserve an interpretive ethnography; this happens when there are not very many of them, and when their cultures are very exotic and mysterious. In any case, personhood and thingness are outcomes, not causes, of social processes of attribution. People tend to take the intentional stance toward their own pets, granting them some amount of agency, and taking a more interpretive approach towards making sense of them. Pets acquire the "rich inner life" normally reserved for persons, whereas persons with Alzheimer’s stop being observed as having a rich inner life. Such former persons move closer to becoming physical objects in beds, to be handled much as other physical objects. The important sociological difference is not between things and people, but between the attribution of interpretivism or determinism.

Pets move closer to personhood on the person-thing continuum, especially when they have been around for some time to become an integral part of a close moral community, such as a family. Then, they even acquire "character." "Character" imposes some structure and consistency on behaviors, makes sense of them in terms of a network of "characteristic" dispositions, and fits them into a schema that makes prediction more possible. Over time, "character" may reify and generalize into "stereotype." This happens when explanations of behaviors move back along the continuum, closer to the object-pole. Non-pets, or other people’s pets, are not part of one’s intimate circle of associates, and so are treated more as "strange" physical objects and biological organisms. Such organisms may live in one’s house, such as spiders, or even in one’s body, such as bacteria, but they are not part of a moral community, and so do not acquire the privileges of agency(8). Their behaviors do not express "character," but must be explained by the causal methods of hard science.

The choice of methods, stances, and approaches is indeed not governed by intrinsic differences between things social and things natural. Rather, "social" or "natural" are the consequences of processes of attribution that vary from observer to observer, across time and space. Nothing is natural or social in itself. There is no Ding-an-sich. Rorty’s (1979, 321) great insight is that science and hermeneutics are not coextensive with Nature and Culture, but that science turns into hermeneutics when there is a lot of uncertainty, and when the "normal" methods do not seem to work anymore. This happens, for example, in episodes of "revolutionary" science, when imagination and creativity become more valued than methodical and systematic reasoning. Hermeneutics is also a tribute to the modern Self, and its celebrated capacities to invest the world with meaning (Fuchs 1993b).

Conversely, there are very routine areas of culture, such as large batch manufacturing or elementary public school teaching. In such routine bureaucracies, there is little hermeneutics, but much method, for dealing with many things or thing-like persons that are constructed as roughly similar before they are subjected to the same treatments. As a matter of fact, thing-like persons are routinely perceived as standard cases, holders of ID numbers, and fully describable by the bureaucratic formulas and classifications. This changes when there are fewer and richer students in smaller classes in more elite liberal arts colleges. Such organizations are paid and equipped to perceive more individualism. Parents expect teachers to expect that their students are all special in some way. Due to small size, this is now possible. Larger public educational bureaucracies have no way of dealing with all these individuals; they process large numbers of people through standard sequences of courses and examinations, one cohort after another.

To sum up, we do not need a new metaphysics to overcome essentialism and dualism if we make full use of the sociological arsenal. In fact, a new metaphysics does not solve any problems, but simply displaces them to another level, such as the Middle Kingdom of collectives of quasi-objects. Instead, once we allow for variation, we can observe Nature and Society, Subject and Object, Persons and Things, Interpretation and Explanation, or Hermeneutics and Science as the poles in a continuum of social attribution and construction. Processes of attribution and construction depend themselves on other variables, such as size, time, uncertainty, or moral boundaries. This displaces the metaphysical problematic. The next question, then, is: How are these variables chosen and distinguished, and by whom?

Variable Relations

The identification of "independent" and "dependent" variables is itself based on a decision of some observer, and not to be found in reality itself. We probably learn more about the observer from the way he draws distinctions than about the referents of his distinctions. This implies that distinctions are contingent (but not arbitrary); they can be drawn but do not have to be, and they can be drawn differently by different observers, and for different purposes. Different distinctions can be drawn at different times to separate different aspects of the same object. "Correct" distinctions must be learned; they are not imprinted on the things or properties to which they refer. Distinctions are not carved in stone. There can be as many systems of distinction as there are ways of approaching and dealing with the world. Distinctions are sense-making devices that economize on information costs by highlighting this (but not that). Without distinctions, nothing makes sense, nothing can be observed, and nothing can be learned. Even worse, without distinctions nothing matters (Luhmann 1997, 155).

At the same time, distinctions can turn into obstacles for further learning. This happens when they get reified and turn into natural kinds. Then, they become very inflexible, and violations of the established "order of things" are punished as moral offenses and aberrations. One distinction (!) between science and common sense may be that in science, such reifications are less likely to be successful over long periods of time because competition over discoveries increases self-produced uncertainty. In science, unlike in common sense, violators of established classifications are less likely to be persecuted as moral failures. This is so mostly because science rewards innovations, which always include innovations in classifications and distinctions.

All this is familiar sociological lore since the later Mannheim and Durkheim(9), and part of the interactionist, labeling, and constructivist traditions. If no one draws a distinction, it does not exist. If a distinction is drawn but no one pays attention to it, it does not exist in and for the group to whom it is supposed to matter. If distinctions concern people who draw their own distinctions, controversies and conflicts over classification are likely to ensue. Conflicts over classifications and distinctions are thus often conflicts over social order; "correct" ways of classifying and distinguishing are loaded with moral significance and righteousness. This is true especially for distinctions and classifications that concern the sacred possessions of the group, such as what separates "us" from "them," the clean from the dirty, me from you, humans from nonhumans, or the context of discovery from the context of justification. The more outrage and consternation a violation of cognitive order provokes, the more central and sacred that order is to the group’s form of life (and vice versa).

The most powerful distinctions are institutional labels. These are constructs whose constructedness has become all but invisible to those inside, perhaps because no one can seem to remember how this construction was originally done, and by whom (Douglas 1986). Such distinctions have lost their air of contingency, and appear to express the necessities of being, or the natural order of things. As a result, they are guarded by strict taboos and rigid impossibilities. Bourdieu calls such distinctions the "habitus."

Entrenched institutional labels come with an entire apparatus of what Foucault would call "disciplinary techniques," although he may have been too impressed by their pervasiveness and power. Many distinctions are better described as "permanently failing operations," because their contingency can easily be revealed, especially in conflicts over classifications, and when these classifications are novel and unsure of themselves. Conflicts and controversies shatter certainties; they denaturalize distinctions and classifications by demonstrating that they can, in fact, vary. One reason for this is that conflicts introduce second-order observing: one group uses its own distinctions to distinguish another’s distinctions. The latter then appear to the former not as mirroring the real order of things, but as "mere" constructs, "false" ideologies, bias and prejudice, and self-serving rationalizations. Conflicts and second-order observing turn "cultural" certainties into "ideological" justifications (Berger 1995, 81).

Distinctions, including truth or objectivity, are events that must be made to happen, not independent states of affairs or things-in-themselves. Likewise, variables themselves do not indicate whether they are endogenous or exogenous. That distinction makes sense only within a model of reality, not in reality itself. It is that model, not reality, that makes the distinctions that matter to it. The proof of a distinction is thus not its referential adequacy, its "truth," or its mapping the actual distinctions between natural kinds in the real world. Rather, the proof of a distinction is its ability to do work, the most important sort of which is to generate more useful distinctions for different purposes. Such is the key insight of pragmatism, the only philosophy of science that captures something about science-in-the-making.

In science and innovation, "useful" distinctions are those that let familiar things appear in an unfamiliar light. Useful distinctions do not redundantly copy what is already there, but re-arrange things, move them around, switch their contexts, compare that which usually appears incommensurable, and so on. Matters of truth, of course, re-appear on a third level of observation ("it is true that useful distinctions do cultural work"). This inescapability once more signals that the code of science cannot ever be suspended without suspending science itself. Berger (1995:95) captures this paradox nicely in the work of Bourdieu, who demonstrates the status-driven contingency of even the most "natural" distinctions, only to introduce his distinctions as the ones that go beyond ideology to capture objective truth.

Complexity and Prediction Across the Divide

It is possible to describe evolution as increases in complexity, though not in any teleological sense, and not in the sense of some kind of optimization or increasing superiority. Gould has reminded us time and time again that increases in complexity are not the telos of evolution, and that complexity is not, by itself, always and everywhere an advantage. It all depends – on the texture of the environment and its turbulence, for example. Lawrence and Lorsch (1967), and population ecology, have shown how this correlation shapes organizations and their divisions in various environmental niches. Briefly, when the environment is rather stable, uniform, and changing slowly, simple structures are preferred over complex ones.

Complexity is not an advantage per se. The most important reason for this is that complex systems are more prone to internally triggered breakdowns through multiple interacting and spiralling failures. Complex systems have more internal environments than simple systems, and these internal environments, if decentralized, are difficult to coordinate and control from above, or from one single center. Multiple internal environments produce more internal states and events than fewer internal environments, ceteris paribus. Complex brains with many interacting parts, multiple layers, and decentralized feedback propagation are more prone to mental disorder than simple brains with loose coupling and more serial hierarchical processing from a single center. For the same reason, complex and closely coupled technosystems are prone to internally triggered "normal accidents" (Perrow 1984).

Differences in referential ecologies are not simply coextensive with the old Nature/Culture separation – as if all Nature were somehow simple, linear, unchanging, and predictable, while all Culture were complex, subjective, changing, and chaotic. The emerging science of complexity is rapidly breaking down such old-fashioned dichotomies on all fronts (Coveney/Highfield 1995). The logic of bureaucratic cages can be described as simple and linear, while the behavior of particles may be described as probabilistic and uncertain. Darwinian evolution is still going on, with new mutations and unpredicted surprises, while aggregate suicide rates have remained rather stable over several decades.

The simple/complex distinction is primarily that of an observer, who decides either to reduce or elaborate complexity for purposes of description and explanation. Generally, complexity is reduced for prediction, control, mastery, and teaching; it is elaborated for innovation and discovery. It is observers who decide to reduce or increase the number of variables that are being taken into account. It is observers who decide what margins of error are acceptable, and it is observers who decide whether coarse-grained or fine-grained descriptions are more suitable for their purposes. This has little to do with substance or subject-matter, for idealizations, model assumptions, holding constant, and ceteris paribus clauses belong to any science, whether of Nature or Society.

At the same time, there will be variations in the possibility and accuracy of predictions, depending, for example, on the rate of innovation in a cultural area. It is easier to predict what Stephen King will write next, but much more difficult to predict what Peter Handke will. This is because an important part of the "game" in avantgarde art or literature is to be unpredictable, while a big part of popular culture is minimizing the risks in investing and selling a lot of standard products to many consumers whose average reading habits change more sluggishly. This is an example for the gravitational collapse toward social averages: The larger a market or audience for a product, the more sluggish and coarse-grained its behavior, and the more standard and uniform its commodities. This is not so because the culture industry deliberately manipulated false consciousness, but because averages remain longer than individuals and their scores. Departments gravitate toward intellectual mediocrity; pluralistic politics moves toward the middle class and compromises between various constituencies; forests constrain the height of their trees toward the medium size; regressions regress toward the mean, and the standard deviation in baseball scores decreases over time. Life is just a normal curve – at least in the long run. At the same time, normal curves come with many exceptions and surprises at their tails.

What is more, the best way to predict what King will write next is to read as many of his previous books as possible, not, say, an inspection of his current PET scan. Some all-too-eager cognitive neuroscientists and evolutionary psychologists believe they can "reduce" culture to "the mind" – but to whose? Minds produce romantic poetry, microphysics, and neuroscience. How could all this "follow" from mental Darwin machines? Darwinism is not itself the result of natural evolution. To be sure, brains are the result of natural evolution, but what they might think is not.

Of course, as with all predictions, the difference between predictions consists in the amount of error reduction, which means that predictions are good only relative to worse ones. The maturation of the neurosciences might make it possible to reduce the errors in PET scan based predictions of culture, but at present, they don’t even yield reasonable guesses.

Predictions also differ in how fine-grained and time-sensitive they are; we can never predict King’s exact next words, and our predictions will degrade rapidly once forecasts extend into the distant future. So our best predictions are coarse-grained and short-term: We can be reasonably sure – though by no means certain – that King’s next book will be scary. But when it comes to what he will write two decades from now (long term prediction), or just what configuration of characters and plot twists will make his next book scary (fine-grained prediction), predictions do not degrade all that gracefully.

For exactly the same reasons, it is squarely impossible to predict the future "content" of "science." The constructivists have never defined what "content" really means, how fine-grained explanations or predictions of "content" can be, which exact scientific communications in which areas are predictable, how far into the future such predictions can extend, or how many of the vast number of papers or communications can be explained by the same argument. Consider the following illustration. Elsewhere, we have estimated that the annual output of the biomedical sciences alone is about 300.000 articles (Fuchs/Westervelt 1996). Of course, these are just a small fraction of all scientific communications, and even just a small fraction of all communications in the biomedical field, since published papers do not include more informal communications and contributions. But even if we do disregard informal communications, exactly what "contents" can be predicted and explained here? The contents of all of these articles? Including footnotes? The distribution of the references? How far into the future do our predictions extend before they collapse? Is it possible to predict these predictions as well? Who could do all this?

On the level of second-order observation, there are no simple and complex systems in and of themselves – almost any system can be produced or observed, under some possible description and on some level, as simple or complex. To take Coveney and Highfield’s (1995, 38) example, "boy meets girl, family intervenes, the couple dies," is an extremely simple and algorithmically compressed, yet possible, rendition of a highly complex love story, suited to some purposes (such as this one here), but not to others (such as stage instructions to the actors). To take Gell-Mann’s (1994) example, quarks may be simpler when used as an expedient to account for and "save the phenomena" in beam collisions, yet complex when viewed as actual building blocks of matter whose still smaller building blocks are yet to be discovered. Jaguars may be complex in their natural habitat, but rendered simpler when caged, trained, and exhibited in zoos.

Summing up, the simple/complex distinction cuts across the Great Divide as well, and cannot explain variations in intellectual and scientific cultures and specialties, either. Why is there a perceived difference between social and natural facts at all? "The idea that social facts might be socially constructed with no necessary relationship with a preexisting reality is not difficult to understand, or even to accept;" writes Ronald Giere (1988, 58), probably expressing majority opinion. But the reason is that some social facts are more transparent and obviously contingent than others, not that they are social. The important distinction is between strong facts and weak constructs, not between social and natural facts.

What Separates Modernity from Tradition?

Tradition complements modernity in the sense that each modernity constructs and preserves its own traditions. They are not opposites, but depend on each other. Modernity does not "replace" tradition, but continues it according to its own, not the tradition’s, specifications. This includes "breaks" with the past as well, since these breaks and discontinuities are observed not in the past, but by and within what comes next, that is, after the break, rupture, or revolution.

Keeping this observer-dependence of "tradition" in mind, we can probably say that modernity differs from tradition in size and scale, amongst other things. The impact of size on social structure is as dramatic as it is neglected, organization science notwithstanding. Size makes it increasingly impossible to organize society as one large and extended encounter among copresent bodies and their blood relations. Not even tribal societies are encounters, though understanding them as such would yield fewer misunderstandings than understanding modern societies as tribes and encounters. This does not, of course, mean that modern societies have no tribes, or no encounters, or no communities. The difference is not Gemeinschaft versus Gesellschaft; rather, modern Western societies differ from tribes because, in the former, tribes are only one form of association among others.

At each new evolutionary level, previous modes of social association are not dropped and replaced by "higher" forms in some linear and progressive increase in adaptive capacity. Rather, previous modes and levels survive by nesting within newer forms, such as organizations. The archaic and traditional persists, albeit in modified forms. Modern societies have not replaced tribes with organizations, but they differ from tribes because they have more tribes, they have different and novel ways to link them, and they also have non-tribes, such as bureaucracies. At the same time, bureaucracies may be one site where restructured tribes, such as scientific specialties, congregate and communicate. Tribal societies do not have many alternative ways to link their units and components. When they get too large, they split up; they cannot accomodate internal changes by differentiation into different kinds and patterns of non-tribal social organization.

In one important sense, then, tribal societies do differ dramatically from modern ones: due to size and differentiation, modern societies produce vastly more social events and outcomes, such as personal experiences, observations, communications, and behaviors. They also produce these many more events in much less time because their modes of association are much further removed from copresent bodies. In modern societies, billions of conversations can, will, and do occur at any one point in time, whereas tribal societies need much more time to process far fewer encounters and kinds of encounters between far fewer persons.

Modern societies have more observers, none of them "privileged" in any metaphysical or philosophical sense, and they have more different kinds of observers and observations as well. These include families, tribes, science, governments, the UN, at least one post-post structuralist and feminist deconstructess of Durkheim, hundreds of millions of bodies, and Parisian actor-network theorists who believe "we" – whoever that is – "have never been modern."

All this is especially true for modern science. Whatever distinguishes science from myth: it is not truth, philosophical objectivity, logic, or rationality. But there are differences just the same, despite postmodern and relativist essentialism. One difference is that modern science tolerates and encourages vastly more dissidence and conflict by turning some of it into expected innovation. No culture is static, and all cultures innovate, but modern science expects, rewards, and encourages change to an unprecedented degree. What is more, science rewards innovations "blind," that is, without being able to define and fix in advance what they will be, where they will occur, who will make them, and what consequences they can and will have for the rest of society. It is rather impossible to "finalize" science without destroying it altogether. One may set goals for science, but, as in other and all organizations, these goals are only loosely coupled to actual research and its outcomes. In any organization, there are always multiple and conflicting goals, which are also ambivalent, ambiguous, and change over time. The rhetoric of "goals" is political frontstage talk, and this talk is not itself the science that can either get the job done or not. The only thing that is certain in science is uncertainty and change. Science is home-less and end-less, at least as long as it is not destroyed. In science, there are only temporary solutions. All truths collapse over time. Science must not only reward its own obsolescence; it must also find ways to do so and prevent complete chaos, cacophonia, and disorder.

This is the main problem of science, not instrumental control. The main problem in science is that its exponential growth has made it increasingly impossible to come up with a coherent account of its current state, let alone predict its future states. This is aggravated by the fact that any scientific accounts of science are part of science as well, and so increase the complexity of that which they try to explain. The paradox is that a small segment of science, the sociology of science, tries to explain science, including itself, and in so doing adds even more to its explanatory burden. All accounts and explanations must be simplifications, but the sociology of science carries an especially heavy load. Even scientists probably know less than a fraction of one percent of all that could be known about all of science at any given moment in time. We are all amateurs now, especially the experts, who are non-experts about most other things outside of their expertise. Congress knows next to nothing about science, and depends on scientists pushing their respective sciences. Short of undermining the very conditions of its existence, it is as impossible to "control" and "finalize" science as it is to control the weather