What kind of “harmony” does big-data style social physics design for?
I’m working on the final chapter of my book The Sonic Episteme; it focuses on the ways pop science accounts of “social physics” and the subfield of physics called string theory appeal to concepts of acoustic resonance. Here’s the beginnings of my analysis of Alex Pentland’s book Social Physics.
The concluding chapter of MIT data scientist Alex Pentland’s book Social Physics is titled “Design for Harmony: How Social Physics Can Help Us Design A Human-Centric Society” (SP 193). According to this title, “harmony” is the outcome of a society organized by social physics. So, even though he never explicitly defines what he means by “harmony,” we can infer this from the principles of social physics. In Plato’s Republic, a “harmonious” society was one whose parts were arranged to reflect the series of hierarchical ratios articulated in the theory of the divided line. In Pentland’s work, a harmonious society is one whose parts are arranged in accordance with the principles of social physics, namely, the ratios and probabilities that result from “having a mathematical, predictive science of society” (SP 191). Using statistics to model society, Pentland’s social physics ontologizes acoustic resonance and is thus is a constituent of the sonic episteme.
Social physics uses big data to reimagine society as patterned flows across populations–i.e., as acoustically resonant. As Pentland puts it, “social physics is based on statistical regularities that span the population” (SP 189). As Mader argues, these biopolitical applications of statistics to describe the behaviors of populations is an ontological sleight of hand that substitutes the relations among numbers for the relations among people. Pentlandian social physics tells us that’s what it’s doing: “social physics…describes reliable, mathematical connections” (SP 4) between patterned flows of ideas and behaviors. Being explicit about the mathematical foundations of its ontology lets social physics hide its slight in the repeatedly articulated but never thematized slippage between statistics and acoustics or harmony. Relating the mathematical relations produced by statistics to the relations among sounds, Pentland passes off mathematical abstractions as “a human-scale” phenomenon that we already have an “intuitive understanding of” (SP 189). Like the other constitutents of the sonic episteme, Pentlandian social physics naturalizes post-identity biopolitics behind a dehistoricized concept of acoustic resonance.
So if “harmony” only appears as a chapter title, where is the acoustic resonance in Pentland’s book? Pentland consistently describes the mathematical connections among data sets as large-scale patterns within which individual-level micropatterns nest more or less rationally. Pentland emphasizes that social physics improves on traditional statistical methods because “social phenomena are really made up of millions of small transactions between individuals. There are patterns in those individual transactions that are not just averages…[but] micro-patterns” (SP 10-11). From the perspective of social physics, society isn’t organized by top-down categories (like social identities and demographic categories based on them), but by ‘micropatterns’ or “detailed patterns of idea exchanges” (SP 17) that resonate with one another, building or masking amplitude. Unlike group-level abstractions like we find in classically liberal identity politics or demographic analysis, social physics uses statistics to make a different kind of abstraction. These abstractions are rhythmic flows, like the “rhythm[s] of [people’s] daily habits”(SP 142), “activities [that] have rhythms that are predictable across days and weeks” (SP141), or “the rhythms of a city” (SP 144). As I will show later, “rhythmic” patterns resonate consonantly or dissonantly with both other micropatterns and larger-scale social patterns.
Once individual patterns are observed, they are then compared to others, and to larger-scale patterns, to create a ratio or a rate of flow. For example, Pentland explains that “idea flow” is both a flow that “oscillate[s] between exploration and engagement” (SP 100) and describe “the proportion of users who are likely to adopt a new idea introduced into the network” (SP 83; emphasis mine). In other words, social physics measures oscillating flows and expresses that flow as a proportion comparing individual rates across a group. One of the main tools social physicists use–tracking “idea flow versus performance” in order to “tune the network to improve performance” (SP 94)–is explicitly a frequency ratio, a ratio comparing the frequency or rate at which ideas flow to the rate of return on investment. Similarly, social physics has been applied to study the same sorts of population-wide rates Foucault identifies in his work on biopolitics, such as “poverty, infant mortality, [and] crime rate” (SP 184). Pentland describes the macro-scale pattern that results when you compare individual or small-group micropatterns as a Gaussian-like distribution: in one study, his team found “a large range, from isolated individual traders at one end to traders trapped in an echo chamber at the other end” (SP 33), but they concluded that “the wisdom of the crowd resides in between extremes of isolation and the herd behavior seen when the social network is an echo chamber” (SP 33). In addition to identifying these micro and macro patterns, social physics can be used to normalize those patterns so that they amplify other, even more macrocosmic social patterns, i.e., “social norms.” According to Pentland, “a social norm is a set of compatible strategies that all parties agree produces the best exchange value” (SP 21). Effectively, social physics understands social norms as a type of population-wide cost:benefit calculus, a “consensus” that “account[s] for the cost-benefit trade-offs of everyone in the group” (SP 63) (neoliberal economic theory identifies such cost:benefit calculus as a feature of individual subjects). Though Pentland doesn’t use the language of acoustics, his theory of social physics is grounded in the key components of acoustic resonance as I define it in this book: frequency ratios, oscillatory motion, cost:benefit calculus, proportional flows, and Gaussian distributions. Moreover, because 21st century physicists use acoustics to study the flow of energy (electrical, kinetic, etc.) through fluids, Pentland’s claim that “just as the goal of traditional physics is to understand how the flow of energy translates into changes in motion, social physics seeks to understand how the flow of ideas and information translates into changes in behavior” (SP 5) states that social physics applies the laws of acoustic resonance to society.
That application is evident in Pentland’s various descriptions of how patterned flows interact. Arguing that society is composed of “large integrated patterns of behavior” (44), Pentland re-imagines the kind of rationality involved in the social contract: social physics “goes beyond simple idea flow within a community; it also includes striking a bargain between individuals to adopt behaviors that are synchronized and compatible” (SP 62; emphasis mine). This bargain is not “individually rational” (SP 46) subjects consenting but rational phase relationships among rhythmically patterned flows–that is, harmonic consonance. Like phase patterns, rhythmic patterns of individual behavior and/or idea flow “mesh together like pieces in a puzzle” (SP 62), smaller micropatterns nesting together and with larger-scale social patterns. For example, groups adopt new patterns of behavior when these patterns “ fi[t] into the long-term common sense developed with a community of peers” (SP 239). Pentland emphasizes that group performance is a function of the relative rationality or synchronization among phase patterns among the group’s members as they nest together and amplify the group’s overall signal: “synchronization and uniformity of idea flow within a group is critical” (SP 64) because it determines the group’s success. “The higher the performance of the group, the more people shared a common rhythm, including body movement, speech, and tone of voice. The best performing groups were in sync, literally moving in synchrony with each other” (SP 111). That’s why “star performers promoted synchronized, uniform idea flow within the team” (SP 63). As is the case with sound waves, social physicists think that synchronized phase patterns amplify the overall signal. Pentand calls this phenomenon “social efficiency”: “when one person benefits, the entire society benefits…The exchange of ideas and information between people must reliably provide value not only to the individual but to the whole system” (SP 204). His appeal to Zipf’s law (SP 162) to explain the relationship among the sub-patterns that constitute an individual subject’s overall behavioral pattern sounds a lot like contemporary understandings of how overtones relate to primary tones. Zipf’s law holds that the frequency of phenomena within a bounded set, like “the distributions of the sizes of cities, earthquakes, solar flares, moon craters, wars and people’s personal fortunes all appear to follow power laws.” Power laws describe exponentially proportional relationships, like the relations among overtones. In fact, mathematicians identify Zipf distributions as a type of harmonic series. Given the repeated and varied appeals to rational phase relationships and harmonic consonance, we can infer that this is the “harmony” Pentland refers to in the title of Social Physics’s last chapter.
Big-data style social physics is a constituent of the sonic episteme because it appeals to notions of acoustic resonance–specifically, the idea of (ir)rationally nested rhythmic patterns–to translate statistical abstractions into more concepts laypeople can grasp using intuitive, non-propositional knowledges. Like the other constituents of the sonic episteme, it naturalizes post-identity biopolitics behind this seemingly objective concept of harmony. As Pentland describes it, this style of social physics organizes society by normalizing populations instead of hierarchically ordering demographic groups. I’ll talk about the contrast he draws between social physics and classical demographics a bit later; first I want to explain the role of normalization in social physics. According to Pentland, social physics is valuable because it lets managers normalize the rate of idea flow: “leaders can increase its performance by promoting healthy patterns of interaction within their organization” (SP 105), and social physics is the tool that helps them maintain these “healthy patterns of idea flow” (SP 105). A “healthy” rate of flow keeps the diversity of ideas in proportion to the level of engagement across the population, exhibiting “balanced participation as well as high engagement within the group” (SP 108). Balanced and proportional, healthy flow is moderate. Such moderation is healthy because it contributes to all the social goods neoliberal biopolitics aims to produce, like optimal group productivity or “GDP per square mile…patenting rate, R&D investment rate, [and] crime rate” (SP 166) and an “an inclusive, vigorous culture” (SP 130) full of “dynamic, responsive organisms” (SP 14). Moderation or sophrosyne is one way social physics normalizes society to the large-scale social patterns calibrated by post-identity biopolitics.
Another way it does this is with statistical normalization. Identifying and maintaining a normalized distribution keeps idea flow “in the healthy wisdom of the crowd region” (SP 38). This “region” is the range of the most frequently-practiced individual patterns of behavior, the most common or frequent individual frequencies. Because it thinks frequency ratios and statistical rates are the basic unit of existence, social physics perceives society as a meta-rate or ratio expressing the rate(s) at which individual patterns occur within the group. And the most frequent (normal) individual patterns produce the best idea flow. As Pentland explains, “the set of behavior examples that [people] were immersed in…was the most powerful force in driving idea flow and shaping opinion” (SP 50). Pentland thinks that the more normal a behavior pattern is, the more likely individuals are to bring their own behaviors within the normal range (the unstated implication here is that in this framework individuals who remain irrational or abnormal are so because of choice and/or pathology). From the perspective of social physics, individual behavior isn’t shaped by social identity but by population normalization. In one study, Pentland and his colleagues found that the rate of exposure to an idea (i.e., the normalcy of that idea) was “four times more accura[te]” (SP 52) an index of individual behavior than “age, gender, religion, employment, etc.” (SP 52). Pentland even compares normalized idea flow to a commonly-recognized biopolitical technology, IQ testing (SP 53).
Pentland sells social physics as a technology that produces normalizes populations by enforcing individual moderation and self-mastery (i.e., sophrosyne). Having identified which group-level metapattern is most healthy and consonant with even larger scale social metapatterns (like, say, capitalism), managers decide “which patterns need to be reinforced and which need to be reduced” (SP 108) and then “tune the flows of ideas between people by providing small incentives, or nudges, to individuals” (SP 151). These incentives encourage people to bring their individual patterns into a more rational, amplifying relationship with the overall group pattern. Again using sonic concepts to translate statistical abstractions into more concrete and familiar terms, Pentland uses “tuning” (SP 39) to describe the process of group- and self-normalization. Claiming that “the goal” of social physics “is to have people use these real-time displays to provide them with the sort of social intelligence needed to foster better interaction patterns, thus leading to higher productivity and creative output” (SP 108), Pentland frames tuning as a practice of self-mastery–monitoring one’s patterns and bringing them in ever-more-rational relationship to the group pattern normalized for optimal idea flow. Citizens and employees should “thin[k] about your job as improving idea flow” (SP 119), i.e., managing frequency and rate. As an upgrade on the “science of opinion” that Ranciere wrote about in the 1990s (likely in reference to Bill Clinton’s then-innovative use of polling data in his election campaign), big data-style social physics is still a “realization of the empty virtue Plato called sophrosune [sic]” (Disagreement 106).
Taking frequency ratios as the fundamental unity of existence and managing for statistically normalized populations and moderate, self-mastered individual frequencies, big data-style social physics qualifies as a constituent of the sonic episteme: it appeals to the concept of acoustic resonance in order to naturalize post-identity biopolitics behind the seemingly neutral, even beneficial concept of sound. I’ve shown how Pentlandian social physics is biopolitical, but I haven’t really discussed the post-identity side of the equation, which is where I’ll turn now. But before I talk about Pentland’s explicit claims to overcome both the logic of social identity and past theoretical commitments to representational content and the politics of exception his text produces, I want to focus on the way big data-style social physics reworks concepts of property-in-person so they operate on a biopolitical rather than juridical register.