Philosophy for Data Scientists
Intro to Post-Structuralist French Philosophy for Data Scientists (Part I)
How the Humanities View Data Science
In this series of posts, I’d like to give a brief introduction to some key French thinkers who have had a disproportionately large impact on the humanities. In particular, I want to focus on humanistic/philosophical work aimed at understanding and interpreting the modern practice of data mining, including data collection and automated profiling. Due to the complexity of this work, however, I will need to approach it in several installments.
Among the philosophes whose ideas we will look at are Michel Foucault, Pierre Bourdieu, Gilles Deleuze, Bruno Latour, Jean-Francois Lyotard, Jacques Derrida, Paul Ricoeur, and Jean Baudrillard. This post will focus on Foucault and Deleuze and provide some necessary background by way of a relatively haphazard introduction to Nietzsche, Hegel, and Marx.
Why Care?
By last count, Michel Foucault, for example, had nearly 400,000 citations. Pierre Bourdieu, in contrast, had a measly 188,000. There are single books by these authors that have had more citations than most scholars. It’s difficult these days to read discussions related to personal data, privacy, and algorithmic discrimination coming out of the humanities that don’t refer to work by these and other French philosophers. For instance, John Cheney-Lippold’s A New Algorithmic Identity: Soft Biopolitics and the Modulation of Control, presents an influential take on automated machine learning from the point of view of a cultural and digital studies professor.
Without a background in the ideas of these philosophers and sociologists (indeed it is difficult to place many of them inside a single discipline), much of what is currently being published in social science, humanities, and marketing journals can be difficult to follow, seem overly abstract, and be full of neologisms and various “postmodern jargon.” I’ll try to unpack these for you the best I can, but part of understanding their work consists in learning their conceptual vocabulary.
Also, if you’re interested in any kind of qualitative social science research, you’ll likely encounter these names, as their ideas serve as philosophical bedrock against the more positivistic/quantitative approaches pioneered by thinkers of the Vienna Circle.
Why write this post? We need to foster interdisciplinary “meetings of the minds” among data and computer scientists, engineers, social scientists and humanists.
Given the ubiquity of data mining in our everyday lives, we must bridge the gap between what the scientist and novelist CP Snow once referred to the as the Two Cultures separating the sciences and humanities.
What can Data Scientists Gain from the Humanities?
This question deserves its own post, so I won’t go too deep into it now. Suffice to say that the humanities are generally concerned with the interpretation of texts, events, behaviors, and symbols, among many other things related to the experience of being human.
If you think interpretation is for wimpy, beret-wearing, Frappucino-sipping keyboard warriors, just think of how many people have been killed because their interpretation of a book or event was considered “wrong.” This is serious stuff with potentially life-or-death consequences for millions of people. (I’m looking at you, Mao and Stalin). Right now in America, fierce cultural debates are raging about what it means to erect statues of and build monuments to those who profited from the ownership of slaves, for instance. John Maynard Keynes describes the power of the humanities so:
“The ideas of economists and political philosophers, both when they are right and when they are wrong are more powerful than is commonly understood. Indeed, the world is ruled by little else. Practical men, who believe themselves to be quite exempt from any intellectual influences, are usually slaves of some defunct economist.”
Keynes is effectively calling out the engineers and data scientists who are blissfully unaware — sometimes proudly so — that they are mere cogs in a social machine likely dreamt up by some long-forgotten philosopher in some dusty old book somewhere. Without engaging with these ideas and considering your role in this system, you’re liable to become a pawn in someone else’s vision of utopia. I explore this point further in another article titled Data Scientists and the Ethics of Power.
Other key functions the humanities play include: explaining, discovering, and justifying the values embodied in technology, articulating better and worse visions of socio-political systems, and exploring the nature of (post) human experience. Put simply, without the moral imagination of the humanities, modern life would likely be efficient, but intolerably sterile and driven solely by the pursuit of profit.
Caveat lector: I’m not a scholar of modern/postmodern philosophy, so please don’t mistake my rather idiosyncratic interpretations of these thinkers’ work as constituting the consensus opinion of experts in this field.
Quick Background: Hegel (1770–1831), Marx (1818–1883), Nietzsche (1844–1900)
Much of the modern French philosophical thinking we’ll explore has been deeply influenced by the German philosophers G.W.F. Hegel, Friedrich Nietzsche, and Karl Marx . Without briefly touching on some key concepts they introduce, it will be difficult to give context to the later work of Foucault, Deleuze, and others. So in what follows, I’ll try to hash out just what you need to know in order to make sense of things later. Keep in mind that the post-structuralists tend to eschew direct argumentation in favor of developing vast, sweeping philosophical visions. They don’t so much as make arguments as present their readers with alternate possibilities for reality. My apologies if I seem to ramble at times.
Hegel: Conceptual Dynamism and Fuzzy Logic
It’s difficult to do justice to Hegel in a few paragraphs, so I’ll just focus on aspects of Hegel that I think are most relevant to data science. To start, Hegel’s ideas suggest conceptual limits on the power of supervised machine learning to account for reality.
Hegel was not a fan of binary, either-or logic, the kind embodied in the crisp set theory of traditional statistics used in machine learning. Things could both be and not be at the same time. For example, you are not, strictly speaking, the same person you were yesterday. Billions of cells which make up your body have died and been replaced by new ones. Yet you appear to be the same person and we refer to you using the same name. Child-Travis and adult-Travis are clearly different, yet still the same person. How can we reconcile this fact? Seen this way, Hegel can be thought of as an intellectual forebear of Lofti Zadeh’s Fuzzy Set Theory, which posits crisp set theory as a special case where set membership functions only take on the values 1/0. Hegel’s speculative logic and “possibility theory” (as opposed to probability theory) share conceptual roots in my view.
Hegel’s understanding of the logic of scientific thinking is immanent in nature. By this he means that concepts contain within themselves their own negation. On the surface, it sounds paradoxical and in violation of the law of the excluded middle (something is either p or not p), which undergirds modern probability theory, but it shares a likeness with many spiritual concepts, such as samsara in Buddhism.
The image above illustrates how concepts evolve over time by integrating these positive and negatively immanent aspects. Synthesis is a creative act resulting from thesis and antithesis “annihilating” one another in an act of creative destruction (thus the samsara reference). We should note that destruction is a necessary step in self-realization. What appears as contradiction can in fact be reconciled in a higher unity if we are willing to go along with it far enough. In mathematics we often find some abstraction to be a mere special case of an even more abstract conception. The dot product is a special case of the more abstract inner product, for instance.
Contradiction is not to be avoided, but to be embraced if we are to understand the nature of spirit (Geist) in its absolute form.
Hegel is perhaps most famous for his claim that there is a logic to history, that it unfolds according to a rational order, of which individual humans — as conscious, rational subjects — also partake. For Hegel, reason and freedom are linked: freedom is precisely the unfolding of reason in this rational order as it strives towards a point of totality or singularity, a point where it comprehends itself as such. What we see as reason is really just the self-expressive movement of reason on this dialectical journey, manifested in human consciousness, human institutions, works of art, and so on. As rational, self-conscious creatures, we can, at best, go along with this dialectical ride, but we cannot escape the “cunning of Reason.” Hegel famously spoke of the “slaughterblock of history,” and saw both good and bad events as necessary realizations of an inexorable march towards self-realization of “absolute spirit.” Reason cannot be tamed.
In my view, a major difference between Continental and Analytic schools of philosophy concerns the essence of logic. Is it intrinsically dynamic and free as Hegel contends? Or can it be subdued and formalized, as Frege, Whitehead, and Russell hoped?
There is a distinctive Zen-like element to Hegel’s Logic of Science. According to Stephen Houlgate’s The Opening of Hegel’s Logic: From Being to Infinity, Hegel’s exploration into the nature of scientific thinking is based on the presupposition-less observation of thought’s own dynamics. In other words, Hegel aimed to uncover the nature of thought by letting thought move on its own. We should not, as Kant did, assume there are certain a priori categories to which thought must be constrained.
In modern form, the transhumanist movement views the human species as a mere step in a larger cosmic unfolding of Absolute Spirit’s self-realization (the so-called Singularity). We should also point out that this idea of the rational unfolding of history is taken up by Marx. Marx’s Communist utopia was the final realization of human history: capitalism was supposed to be just a stop on the way.
Interpretivist social science has been shaped by similar ideas from Husserl’s phenomenology, which aimed to study the objects of consciousness precisely as objects of consciousness, rejecting earlier rationalist claims that substances of mind and body could be cleanly separated. Natural scientists steeped in Enlightenment rationalism may thus initially feel uncomfortable with Hegelian dialectical thinking, which avoids the simple, binary oppositions of self-other, inside-outside, good-bad, and male-female. Indeed, clear boundaries between self and other can and do break down at the biological and molecular levels, as those suffering from auto-immune disease can attest to.
Self-consciousness and the Struggle for Recognition by Others
Hegel is possibly the first philosopher to explicitly grapple with self-consciousness and reflexivity in cognition. For him, recognition by “others” provides grist for self as object. In other words, self-realization — recognizing oneself as a “self” — fundamentally depends on the social recognition of other autonomous objects (persons) who recognize your existence as an individuated person with unique desires and goals. Our personal identities depend on this act of recognition by others. We are not disembodied Cartesian egos: we are interdependent and social creatures embedded in social environments.
Axel Honneth extends Hegel’s ideas to interpret cries of social injustice by marginalized communities as the struggle for recognition. On this view, oppressed groups are fighting for claims of recognition and social legitimacy. Judith Butler, in her book Undoing Gender, explains the struggle for recognition in a way that highlights Hegel’s commitment to a precursor of fuzzy logic:
“To be called a copy, to be called unreal, is one way in which one can be oppressed, but consider that it is more fundamental than that. To be oppressed means that you already exist as a subject of some kind, you are there as the visible and oppressed other for the master subject, as a possible or potential subject, but to be unreal is something else again. To be oppressed you must first become intelligible. To find that you are fundamentally unintelligible (indeed, that the laws of culture and of language find you to be an impossibility) is to find that you have not yet achieved access to the human, to find yourself speaking only and always as if you were human, but with the sense that you are not, to find that your language is hollow, that no recognition is forthcoming because the norms by which recognition takes place are not in your favor.”
Seen from the limits of crisp set theory, transgender and other marginalized groups of people appear as contradictions. We are taught that sexual identities must be either-or, 1 or 0. The mere existence of something like a transgender identity threatens to undermine the most basic divisions of reality, leading some to violence and anger at such metaphysical denial. But Hegel would say that by learning to embrace contradictions, we can actually achieve an understanding of something much greater.
Marx: Alienation and Ideological Superstructures
A follower of Hegel, Marx of course was famous for his Communist Manifesto, but his contributions to social science run deep, even today. Though many of his historical predictions never came to pass, his critique of the capitalist system and conceptual approach is still highly influential. Case in point, see Shoshana Zuboff’s monumental book Surveillance Capitalism.
Borrowing from Hegel, Marx believed there was a rational shape or arc to historical development which would ultimately result at some future point in a Communist utopia, in which the “shackles” of the oppressed proletariat class would be torn off. At the same time, it is often said Marx “turned Hegel on his head” or “inverted” his ideas. For Hegel, the dynamism of thought ultimately accounted for our experience of reality, a philosophical position known as idealism. Hegel famously wrote, “What is rational is real, and what is real is rational.” But in Marx it is the material, economic facts that accounted for reality. So while Hegel was an idealist, Marx was a materialist.
In Marxian thinking, facts about the mode of production of material objects determine the structure of society. As just mentioned above, Marx was a materialist in this sense, as he believed the material conditions(i.e., the (economic relations between capitalists and laborers) of a society determined its non-material structure. Marxians refer to the means of production as the base or substructure upon which society is formed. Everything else, including all social norms, morality, law, and culture are part of the superstructure.
The Capitalist Superstructure
Marx was concerned with how the factory system of his time effectively separated the lone worker from the fruits of his labor in exchange for a wage. This observation was important because Marx believed that humans derived meaning and enjoyment from their productive activities. Humans were essentially creatures of labor, capable of building and creating novel objects to meet their needs. Consequently, the dispossession of the fruits of their labor by the capitalist ruling class was deeply troubling for Marx.
In short, capitalism alienates people from their work and replaces moral questions of value with dollar signs. Further, the overspecialization inherent to organizations in capitalist systems leads to what the French sociologist Durkheim captured in his concept of anomie. Assembling widgets — or soldering iPhone components — for most of your waking moments is certainly not a life Marx saw as conducive to human flourishing.
We can already see some scholars using this image of dispossession to explain how persons are alienated from their personal data in the pursuit of advertising profits by corporations like Google and Facebook. Check out Zuboff’s Surveillance Capitalism if you’re interested in reading more about this kind of thinking and how it applies to the major Behavioral Big Data (BBD) platforms.
Nietzsche: Slave Morality and The Will to Power
Nietzsche would have been a great digital marketer: he knew how to use shock value in order to get the attention of his readers. In his classic On the Genealogy of Morality, Nietzsche lays out a powerful critique of Western Judeo-Christian values. According to Nietzsche’s psychological analysis, Christian morality is based on a reversion of Roman upper-class values. Oppressed early Christians harnessed their “resentment” to glorify their social and political weakness. Consequently, meekness, equality, and other expressions of a lack of power are held up as moral ideals, when in fact they are merely manifestations of what Nietzsche believes to be a kind of “slave morality” of the Lumpenproletariat.
We Westerners have been duped by such slave morality, Nietzsche would say, leading to a kind of moral and spiritual stunting that prevents us from living life in the fullest, most creative, and passionate way. In this sense, there is a clear affinity with Marx’s notion of the ideological superstructure, where bourgeois (capitalist) morality has supplanted our more basic and pre-industrial way of living. Rousseau would agree that we moderns seem to have lost our lust for life.
It’s no secret that Nietzsche was an elitist. He espoused a kind of mythology of the unique, creative, and powerful individual embodied by the Greek heroes in Homer’s Odyssey, for instance. In the grand scheme of history, we forget the masses and remember the great movers and shakers of society. We really only care about developing those Black Swans whose lives leave an indelible mark on future generations, for better or worse. Notice the similarity to the ideas of Ayn Rand and her technochauvinist followers.
Based on his Genealogy of Morality, it’s clear that Nietzsche believed moral values were not based on any kind of deep, unchanging metaphysical truths, but rather on the interests and values of the ruling classes. Might makes right. This idea of moral perspectivism goes back to the era of Socrates, but Nietzsche takes it to its logical extreme. Robert Solomon, in his book Living with Nietzsche, explains that perspectivism is the idea that “all doctrines and opinions are only partial and limited by a particular point of view.” What we know is thus intrinsically limited to our contexts of knowing, perceptual limitations, languages, social upbringing, etc. There’s no God’s eye view, according to Nietzsche. We will see this idea resurrected by Foucault, Derrida, and Latour in later sections.
For now, we can already see how such perspectivism would seem to limit claims of ML objectivity, especially when for-profit companies decide which data to collect. Nietzsche would agree that raw data is an oxymoron. At the same time, however, Nietzsche would likely deplore the efforts of social justice warriors for greater equality and claim that these new data representations of persons — whether biased or not — are expressions of a corporate will to power by Facebook and Google. We shouldn’t hold it back.
Foucault (1926–1984): Disciplinary Power & The Panopticon,
Michel Foucault is perhaps the most important thinker for understanding current issues of social justice in tech. Foucault was really interested in dissecting and illuminating the “invisible” forms of power. He wasn’t so interested in obvious examples of political repression or violence against citizens by the state, but in the way social norms subtly worked to “produce” subjects with easily distinguishable labels. Foucault, like Nietzsche, was worried by the oppression of the unique individual by the conformist masses. In fact, Foucault saw the advance of statistical theory, with it use of theoretical populations, as a case in point of this oppression of the individual and the emergence of what he called biopower.
For Foucault, categorical labels (e.g., “insane,” “man,” “woman,” “black”) allowed states to achieve more or less the same degree of conformity as physical violence might, but with a patina of Enlightenment humanism. In modern society, as opposed to Medieval Europe, deviance from norms was not punished through physical violence, but through labels and the institutions associated with them. For Foucault, schools, prisons, and hospitals were designed to engender docile bodies and mold the masses. We can clearly see that Foucault is launching a critique of the modern way of life. Further, we can see influence of Nietzsche in his rejection of slave morality (glorifying the unique, powerful, and passionate individual) and the support for the exercise of our innate will to power, which serves to distinguish us as unique individuals.
Foucault most famously illustrated the invisible nature of power through his metaphor of the Panopticon prison. In this wheel-and-spoke-like prison design, a central guard could watch each individual cell, but no prisoner could know he was being watched. Without needing to physically brutalize the prisoners, guards could instead achieve conformity merely through the possibility of the gaze. According to Foucault, the power of the gaze arises through its ability to become internalized by an individual. Once internalized, the gaze functions to influence and control the individual in her dispositions to behave in certain ways without any physical intervention. It’s an invisible form of power.
Knowledge and Discourse as Expressions of Power
Ruha Benjamin’s recent book, Race After Technology, draws deeply from the kinds of ideas Foucault explored. Like Marx before him, Foucault was concerned with describing the “ideological superstructure” associated with various socio-historical periods, from the Renaissance until the modern period. Combined with Freud’s theory of the unconscious, Foucault, in his book The Order of Things, called these superstructures epistemes and set out to uncover the unconscious “rules” determining the form of discourse of a particular period. Epistemes can be thought of as the preconditions for the possibility of various discourses, which prescribed various and often unconscious criteria for knowledge. They set the rules for what counted and what didn’t, what had value and what didn’t. Thomas Kuhn’s idea of scientific paradigms is a great example of this kind of thinking. Outside of a given paradigm, certain questions don’t even make sense to ask.
For Foucault, the ability to dictate the epistemes of a period is what really amounts to power: it’s a power to shape the existential narratives of subjects who must operate within the confines of a given discourse. Foucault initially believed discourse was all-prevailing in determining our experience, institutions, and social practices. As Wittgenstein would also argue, we can’t escape from these linguistic and social discourses which shape how we think and behave. From this conclusion we see claims in Science and Technology Studies (STS) such as, there’s no such thing as raw data independent of our linguistic and cognitive apparatus.
Lastly, I should also mention that some critics dislike the way in which Foucault removes agency from persons and places them at the mercy of the current discourses in society. Why resist or protest for change if we are powerless to act outside of the prevailing discourse?
Deleuze (1925–1995): Control, Modulation, and Dividuals
Enter our last philosopher for this post, Gilles Deleuze. Like Foucault and Nietzsche before him, Deleuze is fascinated by power, creative expression, and perspective. But he updates their work in light of the age of the computer and digital revolutions.
Unlike Foucault’s older “disciplinary societies,” Deleuze sees us now living in “societies of control,” which modulate persons through “perpetual training” (e.g., schools) to achieve conformity. The enclosed spaces that Foucault described are now replaced by distributed networks. The mass/individual distinction of societies of discipline no longer holds. Hegelian dialectical thinking appears again as we enter this new technological era, described as an era of the transhuman. Traditional boundaries between what is human and non-human thus become blurred. This is, however, not necessarily a bad thing for Deleuze, as we will see.
In the era of big data, individuals are now dividuals, digitally spread through databases and social networks. Power is exercised by breaking down unitary things into distributed forms. Questions of identity are no longer clear once objects have been cut and reduced into their component parts. While Foucault’s society of discipline relies on mechanical objects like “levers, pulleys, and clocks” to exercise power, societies of control are more about “energy, entropy and computers.”
According to Deleuze, we have entered a new stage of capitalism where services, not products, are the goal. The corporation has replaced the factory. We don’t have persons, but ambiguously “coded figures,” such as “stockholders.” They are faceless and without clear identity. Marketing, not the production of goods, has become the “center or the ‘soul’ of the corporation.” Digital technology splits up individuals and finds new ways of recombining them and extracting value. As marketing theorists Cluley & Brown (2015) explain the shift: “power is exercised by manipulating and extracting value from parts or micro-assemblages.”
Life is Difference: Deterritorialization and Recombination
Deleuze was fascinated with the process of becoming and the related concept of identity. Art and philosophy were special in that they allowed for a ‘deterritoralization’ of ideas into new environments. This deterritorialization process allows us to create new things. Deleuze wants to go beyond the correspondence theory of truth, which drives much of traditional, rationalist scientific thought, to fully detach map from territory.
If you’re a data scientist, you can imagine ‘deterritorialization’ as a change of basis operation (matrix multiplication) for a given vector representation (here an idea or concept). In the new ‘basis representation” (context), we may derive new and previously unnoticed insights regarding the concept’s underlying nature.
If we ditch the correspondence theory, though, some might worry that language will no longer have anything to “hook on to,” and we would just become depressed nihilists. What’s the point if nothing refers back to anything real? Deleuze turns this conclusion on its head, just as Nietzsche did in his Genealogy of Morality.
Instead of giving up on the project of science because language is really just self-referential at its base, Deleuzian thought celebrates the fact there are dividuals that can be freely computed and deleted without any clear semantic connection to real people. By divorcing people from their digital representations, we can play with them in new ways. Just as CTRL-Z gives you the freedom to create things you might otherwise be too afraid to attempt, Deleuze is trying to show how this “detached” aspect of digital reality leads to creativity and new forms of being.
Deleuze believed the goal of philosophy was to create new concepts. For him, life is difference. The goal of life is to think differently, to become different and create differences. We should be happy that we can’t fit our experience into the closed and easily bounded structures of Foucault’s society of discipline. Deleuze is clear: This isn’t a failure, but a reason to celebrate and explore the possibilities for invention and creation. The goal of philosophy and art is to create difference rather than agreement and common sense. We should embrace difference as it fuels the process of becoming.
Detaching the map from the territory gives us room to play, recombine, and create in new ways unconstrained by reality.
That’s it for now. In the next post I’ll look at Derrida, Bourdieu, and Bruno Latour.
