Digifesto

the end of narrative in social science

‘Narrative’ is a term you hear a lot in the humanities, the humanities-oriented social sciences, and in journalism. There’s loads of scholarship dedicated to narrative. There’s many academic “disciplines” whose bread and butter is the telling of a good story, backed up by something like a scientific method.

Contrast this with engineering schools and professions, where the narrative is icing on the cake if anything at all. The proof of some knowledge claim is in its formal logic or operational efficacy.

In the interdisciplinary world of research around science, technology, and society, the priority of narrative is one of the major points of contention. This is similar to the tension I found I encountered in earlier work on data journalism. There are narrative and mechanistic modes of explanation. The mechanists are currently gaining in wealth and power. Narrativists struggle to maintain their social position in such a context.

A struggle I’ve had while working on my dissertation is trying to figure out how to narrate to narrativists a research process that is fundamentally formal and mechanistic. My work is “computational social science” in that it is computer science applied to the social. But in order to graduate from my department I have to write lots of words about how this ties in to a universe of academic literature that is largely by narrativists. I’ve been grounding my work in Pierre Bourdieu because I think he (correctly) identifies mathematics as the logical heart of science. He goes so far as to argue that mathematics should be at the heart of an ideal social science or sociology. My gloss on this after struggling with this material both theoretically and in practice is that narratively driven social sciences will always be politically or at least perspectivally inflected in ways that threaten the objectivity of the results. Narrativists will try to deny the objectivity of mathematical explanation, but for the most part that’s because they don’t understand the mathematical ambition. Most mathematicians will not go out of their way to correct the narrativists, so this perception of the field persists.

So I was interested to discover in the work of Miller McPherson, the sociologist who I’ve identified as the bridge between traditional sociology and computational sociology (his work gets picked up, for example, in the generative modeling of Kim and Leskovec, which is about as representative of the new industrial social science paradigm as you can get), an admonition about the consequences of his formally modeled social network formation process (the Blau space, which is very interesting). His warning is that the sociology his work encourages loses narrative and with it individual agency.

IMG_20160506_160149

(McPherson, 2004, “A Blau space primer: prolegomenon to an ecology of affiliation”)

It’s ironic that the whole idea of a Blau space, which is that the social network of society is sampled from an underlying multidimensional space of demographic dimensions, predicts the quantitative/qualitative divide in academic methods as not just a methodological difference but a difference in social groups. The formation of ‘disciplines’ is endogenous to the greater social process and there isn’t much individual agency in this choice. This lack of agency is apparent, perhaps, to the mathematicians and a constant source of bewilderment and annoyance, perhaps, to the narrativists who will insist on the efficacy of a narratively driven ‘politics’–however much this may run counter to the brute fact of the industrial machine–because it is the position that rationalizes and is accessible from their subject position in Blau space.

“Subject position in Blau space” is basically the same idea, in more words, as the Bourdieusian habitus. So, nicely, we have a convergence between French sociological grand theory and American computational social science. As the Bourdieusian theory provides us with a serviceable philosophy of science grounded in sociological reality of science, we can breathe easily and accept the correctness of technocratic hegemony.

By “we” here I mean…ah, here’s the rub. There’s certainly a class of people who will resist this hegemony. They can be located easily in Blau space. I’ve spent years of my life now trying to engage with them, persuading them of the ideas that rule the world. But this turns out to be largely impossible. It’s demanding they cross too much distance, removes them from their local bases of institutional support and recognition, etc. The “disciplines” are what’s left in the receding tide before the next oceanic wave of the unified scientific field. Unified by a shared computational logic, that is.

What is at stake, really, is logic.

a constitution written in source code

Suppose we put aside any apocalyptic fears of instrumentality run amok, make peace between The Two Cultures of science and the humanities, and suffer gracefully the provocations of the critical without it getting us down.

We are left with some bare facts:

  • The size and therefore the complexity of society is increasing all the time.
  • Managing that complexity requires information technology, and specifically technology for computation and its various interfaces.
  • The information processing already being performed by computers in the regulation and control of society dwarfs anything any individual can accomplish.
  • While we maintain the myth of human expertise and human leadership, these are competitive only when assisted to a great degree by a thinking machine.
  • Political decisions, in particular, either are already or should be made with the assistance of data processing tools commensurate with the scale of the decisions being made.

This is a description of the present. To extrapolate into the future, there is only a thin consensus of anthropocentrism between us and the conclusion that we do not so much govern machines as much as they govern us.

This should not shock us. The infrastructure that provides us so much guidance and potential in our daily lives–railroads, electrical wires, wifi hotspots, satellites–is all of human design and made in service of human interests. While these design processes were never entirely democratic, we have made it thus far with whatever injustices have occurred.

We no longer have the pretense that making governing decisions is the special domain of the human mind. Concerns about the possibly discriminatory power of algorithms concede this point. So public concern now scrutinizes the private companies whose software systems make so many decisions for us in ways that are obscure or unpredictable. The profit motive, it is suspected, will not serve customers of these services well in the long run.

So far policy-makers have taken a passive stance towards the problem of algorithmic control by reacting to violations of human dignity with a call for human regulation.

What is needed is a more active stance.

Suppose we were to start again in founding a new city. Or a new nation. Unlike the founders of every city ever founded, we have the option to write its founding Constitution in source code. It would be logically precise and executable without expensive bureaucratic apparatus. It would be scalable in ways that can be mathematically confirmed. It could be forked, experimented with, by diverse societies across the globe. Its procedure for amendment would be written into itself, securing democracy by protocol design.

programming and philosophy of science

Philosophy of science is a branch of philosophy largely devoted to the demarcation problem: what is science?

I’ve written elsewhere about why and how in the social sciences, demarcation is highly politicized and often under attack. This is becoming pertinent now especially as computational methods become dominant across many fields and challenge the bases of disciplinary distinction. Today, a lot of energy (at UC Berkeley at least) goes into maintaining the disciplinary social sciences even when this creates social fields that are less scientific than they could be in order to maintain atavistic disciplinary traits.

Other energy (also at UC Berkeley, and elsewhere) goes into using computer programs to explore data about the social world in an undisciplinary way. This isn’t to say that specific theoretical lenses don’t inform these studies. Rather, the lenses are used provisionally and not in an exclusive way. This lack of disciplinary attachment is an important aspect of data science as applied to the social world.

One reason why disciplinary lenses are not very useful for the practicing data scientist is that, much like natural scientists, data scientists are more often than not engaged in technical inquiry whose purpose is prediction and control. This is very different from, for example, engaging an academic community in a conversation in a language they understand or that pays appropriate homage to a particular scholarly canon–the sort of thing one needs to do to be successful in an academic context. For much academic work, especially in the social sciences, the process of research publication, citation, and promotion is inherently political.

These politics are more often than not not an essential function to scientific inquiry itself; rather they have to do with the allocation of what Bourdieu calls temporal capital: grant funding, access, appointments, etc. within the academic field. Scientific capital, that symbolic capital awarded to scientists based on their contributions to trans-historical knowledge, is awarded more based on the success of an idea than by, for example, brown-nosing ones superiors. However, since temporal capital in the academy is organized by disciplines as a function of university bureaucratic organization, academic researchers are required to contort themselves to disciplinary requirements in the presentation of their work.

Contrast this with the work of analysing social data using computers. The tools used by computational social scientists tend to be products of the exact sciences (mathematics, statistics, computer science) with no further disciplinary baggage. The intellectual work of scientifically devising and testing theories against the data happens in a language most academic communities would not recognize as a language at all, and certainly not their language. While this work depends on the work of thousands of others who have built vast libraries of functional code, these ubiquitous contributors are not included in an social science discipline’s scholarly canon. They are uncited, taken for granted.

However, when those libraries are made openly available (and they often are), they participate in a larger open source ecosystem of tools whose merits are judged by their practical value. Returning to our theme of the demarcation problem, the question is: is this science?

I would answer: emphatically yes. Programming is science because, as Peter Naur has argued, programming is theory building (hat tip the inimitable Spiros Eliopoulos for the reference). The more deeply we look into the demarcation problem, the more clearly software engineering practice comes into focus as an extension of a scientific method of hypothesis generation and testing. Software is an articulation of ideas, and the combined works of software engineers are a cumulative science that has extended far beyond the bounds of the university.

Jung and Bourdieu as an improvement upon Freud and Habermas

I have written in this blog and in published work about Habermas and his Frankfurt School precursor, Horkheimer. Based on this writing, a thorough reader (of whom I expect there to be approximately zero) might conclude that I am committed to a Habermasian view.

I’d like to log a change of belief based on recent readings of Pierre Bourdieu and Carl Jung.

Why Bourdieu and Jung? Because Frankfurt School social theory was based on a Freudian view of psychology. This Freudian origin manifests itself in the social theory in ways that I’ll try to outline below. However, in my own therapeutic experience as well as many more informal encounters with Jungian theory, I find the latter to be much more compelling. As I’ve begun reading Jung’s Man and His Symbols, I see now where Jung explicitly departed from Freud, enriching his theory. These departures are far more consistent with a Bourdieusian view of society. (I’ve noted the potential synergy here).

Let me try to be clearer about what this change in perspective entails:

For Freud, man has an irrational nature and a rational ego. The purpose of therapy is the maintenance of rational control. Horkheimer’s critique of modern society invoked Freud in his discussion of the revolt of nature: society rationalizes itself and the individuals within it; the ‘nature’ of the individuals that is excluded (repressed, really) by this rationalization manifests itself in ugly ways. Habermas, who is less pessimistic about society, still sees morality in terms of social norms grounded in rational consensus. “Rational consensus” as a concept angers or worries postmodern and poststructural critics who see this principle as a basis for social ethics as exclusionary.

For Jung, the therapeutic relationship absolutely must not be about the imposition of the therapist’s views on the patient; psychological progress must come from within the individual patient. He documents an encounter between himself and Freud where he discovers this; he is very convincing. The Jungian unconscious is a collective stock of symbols, as an alternative to a Freudian subconscious of nature repressed by ego. The Jungian ego, therefore, is a much more flexible subject; at times it seems that Jung is nostalgic for a more irrational, perhaps primitive, consciousness. But more importantly, Jung explicitly rejects the idea of a society’s sanity being about its adherence to shared rational norms. Instead, he opts for a more Durkheimian view of social variety:

Can we make any sort of objective judgment about the final result [of therapy]? Only if we make a comparison between our conclusions and the standards that are generally valid in the social milieu to which the individuals below. Even then, we must take into account the mental equilibrium (or “sanity”) of the individual concerned. For the result cannot be a completely collective leveling out of the individual to adjust him to the “norms” of his society. This would amount to a most unnatural condition. A sane and normal society is one in which people habitually disagree, because general agreement is relatively rare outside the sphere of instinctive human qualities.

A diverse society of habitual disagreement accords much better with the Bourdieusian view of a society variously inflected as habitus than it does with a Habermasian view of one governed by rational norms.

There’s a subtlety that I’ve missed again and again which I’d like to put my finger on now.

The problem with the early Habermasian view is that ethics are determined through rational consensus. So, individuals participate in a public sphere and agree, as individuals, on norms that govern their individual behavior.

Later Habermas (say, volume two of Theory of Communicative Action) begins to acknowledge the information overhead of of this approach and discusses the rise of bureaucracy and its technicization. In lieu of a bona fide consensus of the lifeworld, one gets a rational coalescence of norms into law.

Effectively, this means that while the general population can be irrational in various ways (relative to the perspective of the law), what’s important is that lawmakers create law through a rational process that is inclusive of diverse perspectives.

We see a similar view in Bourdieu’s view of science: it is a specific habitus whose legitimacy is due to the trans-historical robustness of its mathematized formulations.

The conclusion is this: scientists and lawmakers have to approach rationality in specific trans-personal and trans-historical ways. In fact, the rationality of science or of law are only achieved systemically, through the generalized process of science or lawmaking, not through the finite perspectives of their participants, however individually rational they may be. But the general population need not be rational like this for society to be ‘sane’. Rather, individual habitus or partial perspective can vary across a society that is nevertheless coordinated by rational principle.

There is bound to be friction at the boundary between the institutions of science and law and the more diverse publics that surround and intersect them. Donna Haraway’s ‘privilege of partial perspectives‘ is a good example of the symptoms of this friction. A population that is excluded from science–not represented well within science–may react against it by reasserting it’s ‘partial perspective’ as a viable alternative. This is a kind of refusal, in the sense perhaps originated by Marcuse and more recently resurfaced in Michael Dumas’ work on antiblackness. Refusal is, perhaps sadly, delusional and seems to recur as a failed and failing project; but it is sociologically robust precisely because in late modernism the hegemonic rationality allows for Dukheimian social differentiation. The latter is actually the triumph of liberalism over, for example, racist facism; Fred Turner’s The Democratic Surround is a nice historical work documenting how this order of scientifically managed diversity was a deliberate United States statebuilding project in World War II.

If a top-down rationalizing control creates as a symptom pathological refusal–another manifestation perhaps of the ‘revolt of nature’–a Jungian view of rationality as psychic integration perhaps provides a more palatable alternative. Jungian development is accomplished through personalized, situated education. However, through this education, the individual flourishes through a transcendence of their more limited, narrow sense of self. Jungian therapy/education transcends even gender, as the male and female are encouraged to recognize the feminine “anima” and masculine “animus” aspects of their psyches, respectively. Fully developed individuals–who one would expect to occupy, over the course of their development, a somewhat shared habitus–seem to therefore get along better with each other, agreeing to disagree as the recognize how their differences are based on arbitrary social differentiation. Nothing about this agreeing-to-disagree on matters of, for example, taste precludes an agreement on serious trans-personal matters such as science or law. There need not be any resentment towards this God’s Eye View, since it is recognized by each educated individual as manifest in their own role in the social order.

Societal conditions may fall short of this ideal. However, the purpose of social theory is to provide a realizable social telos. Grounding it in a psychological theory that admits the possibility of realized psychological health is a good step forward.

Habitus Shadow

In Bourdieu’s sociological theory, habitus refers to the dispositions of taste and action that individuals acquired as a practical consequence of their place in society. Society provides a social field (a technical term for Bourdieu) of structured incentives and roles. Individuals adapt to roles rationally, but in doing so culturally differentiate themselves. This process is dialectical, hence neither strictly determined by the field nor by individual rational agency, but a co-creation of each. One’s posture, one’s preference for a certain kind of music, one’s disposition to engage in sports, one’s disposition to engage in intellectual debate, are all potentially elements of a habitus.

In Jungian psychoanalytic theory, the shadow is the aspect of personality that is unconscious and not integrated with the ego–what one consciously believes oneself to be. Often it is the instinctive or irrational part of one’s psychology. An undeveloped psyche is likely to see his or her own shadow aspect in others and judge them harshly for it; this is a form of psychological projection motivated by repression for the sake of maintaining the ego. Encounters with the shadow are difficult. Often they are experienced as the awareness or suspicion of some new information that threatens ones very sense of self. But these encounters are, for Jung, an essential part of individuation, as they are how the personality can develop a more complete consciousness of itself.

Perhaps you can see where this is going.

I propose a theoretical construct in: habitus shadow.

When an individual, situated within a social fiend, develops a habitus, they may do so with an incomplete consciousness of the reasons for their preferences and dispositions for action. An ego, a conscious rationalization, will develop; it will be reinforced by others who share its habitus. The dispositions of a habitus will include the collectively constructed ego of its members, which is itself a psychological disposition.

We would then expect that a habitus has a characteristic shadow: truths about the sociological conditions of a habitus which are not part of the conscious self-indentity or ego of that habitus.

This is another way to talk about what I have discussed elsewhere as an ideological immune reaction. If an idea or understanding is so challenging or destructive to the ego of a habitus that it calls into question the rationality of it’s very existence, then the role will be able to maintain itself only through a kind of repression/projection/exclusion. Alternatively, if the habitus can assimilate its shadow, one could see that as a form of social self-transcendence or progress.

Responsible participation in complex sociotechnical organizations circa 1977 cc @Aelkus @dj_mosfett

Many extant controversies around technology were documented in 1977 by Langdon Winner in Autonomous Technology: Technics-out-of-Control as a Theme in Political Thought. I would go so far as to say most extant controversies, but I don’t think he does anything having to do with gender, for example.

Consider this discussion of moral education of engineers:

“The problems for moral agency created by the complexity of technical systems cast new light on contemporary calls for more ethically aware scientists and engineers. According to a very common and laudable view, part of the education of persons learning advanced scientific skills ought to be a full comprehension of the social implications of their work. Enlightened professionals should have a solid grasp of ethics relevant to their activities. But, one can ask, what good will it do to nourish this moral sensibility and then place the individual in an organizational situation that mocks the very idea of responsible conduct? To pretend that the whole matter can be settled in the quiet reflections of one’s soul while disregarding the context in which the most powerful opportunities for action are made available is a fundamental misunderstanding of the quality genuine responsibility must have.”

A few thoughts.

First, this reminds me of a conversation @Aelkus @dj_mosfett and I had the other day. The question was: who should take moral responsibility for the failures of sociotechnical organizations (conceived of as corporations running a web service technology, for example).

Second, I’ve been convinced again lately (reminded?) of the importance of context. I’ve been looking into Chaiklin and Lave’s Understanding Practice again, which is largely about how it’s important to take context into account when studying any social system that involves learning. More recently than that I’ve been looking into Nissenbaum’s contextual integrity theory. According to her theory, which is now widely used in the design and legal privacy literature, norms of information flow are justified by the purpose of the context in which they are situated. So, for example, in an ethnographic context those norms of information flow most critical for maintaining trusted relationships with one’s subjects are most important.

But in a corporate context, where the purpose of ones context is to maximize shareholder value, wouldn’t the norms of information flow favor those who keep the moral failures of their organization shrouded in the complexity of their machinery be perfectly justified in their actions?

I’m not seriously advocating for this view, of course. I’m just asking it rhetorically, as it seems like it’s a potential weakness in contextual integrity theory that it does not endorse the actions of, for example, corporate whistleblowers. Or is it? Are corporate whistleblowers the same as national whistleblowers? Of Wikileaks?

One way around this would be to consider contexts to be nested or overlapping, with ethics contextualize to those “spaces.” So, a corporate whistleblower would be doing something bad for the company, but good for society, assuming that there wasn’t some larger social cost to the loss of confidence in that company. (It occurs to me that in this sort of situation, perhaps threatening internally to blow the whistle unless the problem is solved would be the responsible strategy. As they say,

Making progress with the horns is permissible
Only for the purpose of punishing one’s own city.

)

Anyway, it’s a cool topic to think about, what an information theoretic account of responsibility would look like. That’s tied to autonomy. I bet it’s doable.

Bourdieu and Horkheimer; towards an economy of control

It occurred to me as I looked over my earliest notes on Horkheimer (almost a year ago!) that Bourdieu’s concept of science as being a social field that formalizes and automates knowledge is Horkheimer’s idea of hell.

The danger Horkheimer (and so many others) saw in capitalist, instrumentalized, scientific society was that it would alienate and overwhelm the individual.

It is possible that society would alienate the individual anyway, though. For example, in the household of antiquity, were slaves unalienated? The privilege of autonomy is one that has always been rare but disproportionately articulated as normal, even a right. In a sense Western Democracies and Republics exist to guarantee autonomy to their citizens. In late modern democracies, autonomy is variable depending on role in society, which is tied to (economic, social, symbolic, etc.) capital.

So maybe the horror of Horkheimer, alienated by scientific advance, is the horror of one whose capital was being devalued by science. His scholarship, his erudition, were isolated and deemed irrelevant by the formal reasoners who had come to power.

As I write this, I am painfully aware that I have spent a lot of time in graduate school reading books and writing about them when I could have been practicing programming and learning more mathematics. My aspirations are to be a scientist, and I am well aware that that requires one to mathematically formalize ones findings–or, equivalently, to program them into a computer. (It goes without saying that computer programming is formalism, is automation, and so its central role in contemporary science or ‘data science’ is almost given to it by definition. It could not have been otherwise.)

Somehow I have been provoked into investing myself in a weaker form of capital, the benefit of which is the understanding that I write here, now.

Theoretically, the point of doing all this work is to be able to identify a societal value and formalize it so that it can be capture in a technical design. Perhaps autonomy is this value. Another might call it freedom. So once again I am reminded of Simone de Beauvoir’s philosophy of science, which has been correct all along.

But perhaps de Beauvoir was naive about the political implications of technology. Science discloses possibilities, the opportunities are distributed unequally because science is socially situated. Inequality leads to more alienation, not less, for all but the scientists. Meanwhile autonomy is not universally valued–some would prefer the comforts of society, of family structure. If free from society, they would choose to reenter it. Much of ones preferences must come from habitus, no?

I am indeed reaching the limits of my ability to consider the problem discursively. The field is too multidimensional, too dynamic. The proper next step is computer simulation.

Heisenberg on technology as an out-of-control biological process

In Langdon Winner’s Autonomous Technology: Technics-out-of-Control as a Theme in Political Thought, (1977) there is this quote from Werner Heisenberg’s Physics and Philosophy (1958):

“The enormous success of this combination of natural and technical science led to a strong preponderance of those nations or states or communities in which this kind of activity flourished, and as a natural consequence this activity had to be taken up even by those nations which by tradition would not have been inclined toward natural and technical sciences. The modern means of communication and of traffic finally completed this process of expansion of technical civilization. Undoubtedly the process has fundamentally changed the conditions of life on earth; and whether one approves of it or not, whether one calls it progress or danger, one must realize that it has gone far beyond any control through human forces. One may rather consider it as a biological process on the largest scale whereby structures active in the human organism encroach on larger parts of matter and transform it into a state suited for the increasing human population.”

Mathematics and materiality in Latour and Bourdieu’s sociology of science

Our next reading for I School Classics is Pierre Bourdieu’s Science of Science and Reflexivity (2004). In it, rock star sociologist Bourdieu does a sociology of science, but from a perspective of a sociologist who considers himself a scientist. This is a bit of an upset because so much of sociology of science has been dominated by sociologists who draw more from the humanities traditions and whose work undermines the realism the scientific fact. This realism is something Bourdieu aims to preserve while at the same time providing a realistic sociology of science.

Bourdieu’s treatment of other sociologists of science is for the most part respectful. He appears to have difficulty showing respect for Bruno Latour, who he delicately dismisses as having become significant via his rhetorical tactics while making little in the way of a substantive contribution to our understanding of the scientific process.

By saying facts are artificial in the sense of manufactured, Latour and Woolgar intimate that they are fictious, not objective, not authentic. The success of this argument results from the ‘radicality effect’, as Yves Gingras (2000) has put it, generated by the slippage suggested and encouraged by skillful use of ambiguous concepts. The strategy of moving to the limit is one of the privileged devices in pursuit of this effect … but it can lead to positions that are untenable, unsustainable, because they are simply absurd. From this comes a typical strategy, that of advancing a very radical position (of the type: scientific fact is a construction or — slippage — a fabrication, and therefore an artefact, a fiction) before beating a retreat, in the face of criticism, back to banalities, that is, to the more ordinary face of ambiguous notions like ‘construction’, etc.

In the contemporary blogosphere this critique has resurfaced through Nicholas Shackel under the name “Motte and Bailey Doctrine” [1, 2], after the Motte and Bailey castle.

A Motte and Bailey castle is a medieval system of defence in which a stone tower on a mound (the Motte) is surrounded by an area of pleasantly habitable land (the Bailey), which in turn is encompassed by some sort of a barrier, such as a ditch. Being dark and dank, the Motte is not a habitation of choice. The only reason for its existence is the desirability of the Bailey, which the combination of the Motte and ditch makes relatively easy to retain despite attack by marauders. When only lightly pressed, the ditch makes small numbers of attackers easy to defeat as they struggle across it: when heavily pressed the ditch is not defensible, and so neither is the Bailey. Rather, one retreats to the insalubrious but defensible, perhaps impregnable, Motte. Eventually the marauders give up, when one is well placed to reoccupy desirable land.

In the metaphor, the Bailey here is the radical antirealist scientific position wherein facts are fiction, the Motte is the banal recognition that science is a social process. Schackel writes that “Diagnosis of a philosophical doctrine as being a Motte and Bailey Doctrine is invariably fatal.” While this might be true in the world of philosophical scrutiny, this is unfortunately not sociologically correct. Academic traditions die hard, even long after the luminaries who started them have changed their minds.

Latour has repudiated his own radical position in “Why Has Critique Run out of Steam? From Matters of Fact to Matter of Concern” (2004), his “Tarde’s idea of quantification” (2010) offers an insightful look into the potential of quantified sociology when we have rich qualitative data sets that show us the inner connectivity of the societies. Late Latour is bullish about the role of quantification in sociology, though he believes it may require a different use of statistics than has been used traditionally in the natural sciences. Recently developed algorithmic methods for understanding network data prove this point in practice. Late Latour has more or less come around to “Big Data” scientific consensus on the matter.

This doesn’t stop Latour from being used rather differently. Consider boyd and Crawford’s “Critical Questions for Big Data: Provocations for a Cultural, Technological, and Scholarly Phenomenon” (2012), and its use of this very paper of Latour:

‘Numbers, numbers, numbers,’ writes Latour (2010). ‘Sociology has been obsessed by the goal of becoming a quantitative science.’ Sociology has never reached this goal, in Latour’s view, because of where it draws the line between what is and is not quantifiable knowledge in the social domain.

Big Data offers the humanistic disciplines a new way to claim the status of quantitative science and objective method. It makes many more social spaces quantifiable. In reality, working with Big Data is still subjective, and what it quantifies does not necessarily have a closer claim on objective truth – particularly when considering messages from social media sites. But there remains a mistaken belief that qualitative researchers are in the business of interpreting stories and quantitative researchers are in the business of producing facts. In this way, Big Data risks reinscribing established divisions in the long running debates about scientific method and the legitimacy of social science and humanistic inquiry.

While Latour (2010) is arguing for a richly quantified sociology and has moved away from his anti-realist position about scientific results, boyd and Crawford fall back into the same confusing trap set by earlier Latour of denying scientific fact because it is based on interpretation. boyd and Crawford have indeed composed their “provocations” effectively, deploying ambiguous language that can be interpreted as a broad claim that quantitative and humanistic qualitative methods are equivalent in their level of subjectivity, but defended as the banality that there are elements of interpretation in Big Data practice.

Bourdieu’s sociology of science provides a way out of this quagmire by using his concept of the field to illuminate the scientific process. Fields are a way of understanding social structure: they define social positions or roles in terms of their power relations as they create and appropriate different forms of capital (economic, social, etc.) His main insight which he positions above Latour’s is that while a sociological investigation of lab conditions will reveal myriad interpretations, controversies, and farces that may convince the Latourian that the scientists produce fictions, an understanding of the global field of science, with its capital and incentives, will show how it produces realistic, factual results. So Bourdeiu might have answered boyd and Crawford by saying that the differences in legitimacy between quantitative science and qualitative humanism have more to do with the power relations that govern them in their totality than in the local particulars of the social interactions of which they are composed.

In conversation with a colleague who admitted to feeling disciplinary pressure to cite Latour despite his theoretical uselessness to her, I was asked whether Bourdieu has a comparable theory of materiality to Latour’s. This is a great question, since it’s Latour’s materialism that makes him so popular in Science and Technology Studies. The best representation I’ve seen of Bourdieu’s materiality so far is this passage:

“The ‘art’ of the scientist is indeed separated from the ‘art’ of the artist by two major differences: on the one hand, the importance of formalized knowledge which is mastered in the practical state, owing in particular to formalization and formularization, and on the other hand the role of the instruments, which, as Bachelard put it, are formalized knowledge turned into things. In other words, the twenty-year-old mathematician can have twenty centuries of mathematics in his mind because formalization makes it possible to acquire accumulated products of non-automatic inventions, in the form of logical automatisms that have become practical automatisms.

The same is true as regards instruments: to perform a ‘manipulation’, one uses instruments that are themselves scientific conceptions condensed and objectivated in equipment functioning as a system of constraints, and the practical mastery that Polanyi refers to is made possible by an incorporation of the constraints of the instrument so perfect that one is corporeally bound up with it, one responds to its expectations; it is the instrument that leads. One has to have incorporated much theory and many practical routines to be able to fulfil the demands of the cyclotron.”

I want to go so far as to say that in these two paragraphs we have the entire crux of the debate about scientific (and especially data scientific) method and its relationship to qualitative humanism (which Bourdieu would perhaps consider an ‘art’.) For here we see that what distinguishes the sciences is not merely that they quantify their object (Bourdieu does not use the term ‘quantification’ here at all), but rather because it revolves around cumulative mathematical formalism which guides both practice and instrument design. The scientific field aims towards this formalization because that creates knowledge as a capital that can be transferred efficiently to new scientists, enabling new discoveries. In many ways this is a familiar story from economics: labor condenses into capital, which provides new opportunities for labor.

The simple and realistic view that formal, technical knowledge is a kind of capital explains many of the phenomena we see today around data science in industry and education. It also explains the pervasiveness of the humanistic critique of science as merely another kind of humanism: because it is an advertising campaign to devalue technical capital and promote alternative forms of capital associated with the humanities as an alternative. The Bailey of desirable land is intellectual authority in an increasingly technocratic society; the Motte is banal observation of social activity.

This is not to say that the cultural capital of the humanities is not valuable in its own right. However, it does raise questions about the role of habitus in determining taste for the knowledge as art, a topic discussed in depth in Bourdieu’s Distinction. My own view is that while there is a strong temptation towards an intellectual factionalism, especially in light of the unequal distribution of capital (of various kinds) in society, this is ultimately a pernicious trend. I would prefer a united field.