In a recent piece for the New York Times, Tiffany May chronicles a curious use of metrics as a tool for social control. She writes of the Chinese government’s efforts to curb the country’s rising divorce rate. Fearing the rising “source of social instability,” local officials developed a tool to assess the relational potential of couples headed for a split. She writes:
The quizzes—15 questions, scored on a scale of 100 points—were developed as a way to prevent “impulse divorces,” Liu Chunling, an official in Lianyungang, a city in Jiangsu Province, told The Yangtse Late News. Local news outlets reported that the authorities considered a score of 60 points or higher to mean “room for recovery,” and those couples were encouraged to work on their marriages.
While the tool is merely intended as a starting point for such interventions, not the final word, May writes that “at least one couple’s high score resulted in the authorities’ preventing their divorce in another province last year.”
Within China, the role of data is not limited to divorce alone. In Rongcheng, officials are working on an experimental precursor to an expanded National Social Credit Score. Chronicled recently by Simina Mistreanu in Foreign Policy, the system identifies for its 740,000 adult residents a grade between A+++ and D based on everything from their traffic tickets and drunk-driving records, to heroic acts and work to support family. Score well, and you gain access to benefits like competitive loan rates. Score poorly, and you could lose your ability to travel by train and plane, among other things. Such moves to quantify are clear operationalizations of President Xi Jinping’s principle of “once untrustworthy, always restricted.” Haven’t we already seen this episode of Black Mirror?
For those who care about the outcomes in question—whether family stability or good citizenry—such a layer of metrics, supplementing our natural human ecosystems, might be a foundational tool to foster the public good. In being attuned to a quantified version of reality, we might better adjust behaviour and help our social systems self-correct. Such is a view held even by some within the system. From the same Financial Times piece, one entrepreneur stated, “I feel like in the past six months, people’s behavior has gotten better and better. For example, when we drive, now we always stop in front of crosswalks. If you don’t stop, you will lose your points. At first, we just worried about losing points, but now we got used to it.”
But what kind of bathwater comes with the baby? Regardless of whether it is the Chinese government or technology behemoths capturing the data, are we comfortable with the transparency required to develop such abstraction and the social reward and punishment dished out as a result? In one of the more insightful takes on a world of metrics and data, titled “Seeing Like a Market,” sociologists Marion Fourcade and Kieran Healy write:
As digital traces of individual behaviors are aggregated, stored, and analyzed, markets see people through a lens of deserving and undeservingness, and classification situations become moral projects. . . . Because they seem to record only behavior and behavior is seen to flow from conscious choices, scores become ethically meaningful indexes of one’s character. . . . Metrics become moral injunctions. Spend, but in a controlled way. Drive, but not too fast. Eat, but stay healthy. The prosthetic rationality of the Fitbit or the score offers benevolent surveillance, implicitly instructing people to self-monitor and, if necessary, reach higher or turn their lives around.
Metrics become moral injunctions.
Such cases force us to consider whether a metric abstraction clothed over our social world is a useful societal nudge or instead one of the more troubling developments in a world increasingly built around big data, AI, and machine learning.
In his book The Tyranny of Metrics, Jerry Muller outlines a series of problems with a world fixated on the use of metrics for social assessment and control. Muller’s book is an easy read, but his argument is wide-ranging, integrative, and insightful. The core of his book is a series of case studies of metric use in schooling, health care, policing, military, business, and foreign aid; in these stories, Muller lands his biggest punches. He shows how goals are eclipsed by the social movement toward what gets measured. He argues for the unseen productivity costs of measurement and the diminishing returns of a measurement ethos. He even chronicles how metrics discourage risk-taking and innovation. While realizing the value of metrics, Muller is a prophetic skeptic in a world increasingly believing the opposite.
Beyond many of the technical challenge of metrics—that we measure the wrong things or assess the right things poorly—one of Muller’s more interesting points is that transparency can be the enemy of performance. Pushing against the adage that “you can’t manage what you cannot measure,” Muller argues that in many contexts we need ambiguity. He writes:
A thriving polity, like a healthy marriage, relates some matters of the shadows. In international relations, as in interpersonal ones, many practices are functional so long as they remain ambiguous and opaque. Clarity and publicity kill. The ability to negotiate between couples or states often involves coming up with formulas that allow each side to save face or retain self-esteem, and that requires compromising principles, or ambiguity.
Contrary to Alexander Hamilton and Aaron Burr’s claims, effective functioning depends on not being “in the room where it happens.” We are, as Muller states, a world that “overlook[s] the value of secrecy.” Taking Muller’s critique seriously means asking whether transparency—and in particular, on what dimension and presented to whom—should be strategically built into a system. A wholesale buy-in to the vision of empiricism and metric-layered reality is just as problematic as a rejection of the project altogether.
For me, Muller’s book is an uncomfortable read. I am, after all, a social scientist. I have been trained in the belief that we can measure things, and that in doing so, we better assess and control the social systems we stand in. For example, when Muller identifies the problems with the U.S. News and World Report college ranking system, my mind scans for ways to tweak our measurement. When he shows risk aversion among physicians in response to rating systems, I start thinking of nudges to counteract this metric problem.
There are, he reminds us, real consequences to constructing a layer of data on top of the world.
While this tendency to correct the metric rather than question metrics reflects my own bias, I do think some of the problems Muller identifies are of a technical nature. The ways we assess student performance will improve the more we can individually customize versus collectively group our students. Policing will hopefully improve with better data capture and improved interpretation. I remain a social scientist in my belief that we need precision in our tools, and improvements in metrics make an incredible difference in our world.
However, Muller is downright prophetic about a few crucial points. There are, he reminds us, real consequences to constructing a layer of data on top of the world. Unlike our measurements of natural phenomena, we respond to the numbers. Gravity does not grow stronger if the natural phenomenon feels insecure about its 9.8 m/s^2 force. We on the other hand do change behaviour, and such pivots remain unexpected by many of those who do the measuring.
More importantly, I would argue that being measured erodes our sense of moral responsibility for judgment. Insomuch as metrics appear to reduce the uncertainty in complex analysis, they can lead us to believe that we are not making a choice. After all, aren’t we just following along with what the system suggests is optimal? We outsource moral responsibility to the numbers.
We outsource moral responsibility to the numbers.
Muller concludes his book with a helpful checklist of how to approach our use of metrics. In it, we are reminded as readers that it is not metrics per se that are bad, but rather our fixation on them and misguided belief that all issues are solved through measurement. One of his most essential reminders comes in the following injunction that “recognizing the limits of the possible is the beginning of wisdom. Not all problems are solvable, and even fewer are solvable by metrics.”
What does Muller’s book mean to those of us interested in constructing and participating in a world of greater social integrity? Muller is right that many of us are projecting an almost beatific power onto data. Omniscient and omnipotent data, too often we bow before thee. We need to find a better posture toward a world reflected in data.
Given our exaltation of metrics, it is worth using a bit of theology to assess the problems of such a posture. Early in his book God Without Being, Catholic philosopher and theologian Jean Luc Marion teases out the distinction between the idol and the icon in our approaches to understanding God. In our idols, we create a kind of static rendering of the unknown. Idols freeze the transcendent, pretend to encompass the infinite. This posture stands in contrast to viewing our concepts of God as icons. For Marion, icon-thinking receives the invisible (such as God) as precisely invisible: our concepts become windows rather than mirrors.
While Marion’s focus is on our language of God, there is something exceptionally astute here when we apply his thinking to the way we understand our world through data. Even for me, as someone with decidedly empiricist leanings, the integrity of a metric-driven project hangs on our commitment to not calcify and confuse such understandings of our world (the metric) with the world itself. In a mildly destructive form, this idol-like approach to the use of metrics would mean that we let the measurement of an object be equivalent to the object itself (social credit as goodness, student test scores as academic ability). In a more insidious version, we let such metrics evolve into moral categorizations and extricate ourselves from having to morally engage with the world itself.
In a non-deified form, metrics can be helpful. To the extent an expansion of management by metrics comes with an improvement of our measures, I am slightly less worried than Muller about the net results. Viewed appropriately (metric as icon instead of idol), metrics can provide a self-referential reflection and perhaps even move us toward higher awareness and engagement with the world. Consider the most favourable view of the Chinese metric experiment with marriage:
“Through the guidance of the questions, couples can reminisce on the moments of their relationship and reflect on their familial roles and responsibilities,” Mr. Liu, who oversees Lianyungang’s civil marriage registry, told the newspaper.
But Muller is right to encourage us to ask hard questions about whether transparency is worth the cost. At the very least, he is right to suggest it is not always worth the praise we hoist on its shoulders. For our marriage example, it is just as easy to imagine this metric falling into the hands of the wrong people and being used to justify profoundly destructive relationships, or preventing the rightful empowerment of women within society. The metrics idol can be a jealous god.
As humans, we do tend to muddy the water, don’t we? Like all technologies, the wheat comes with the chaff, and we are left without the ability to separate the two easily. Given the ever-expanding scope of metric-driven thinking, we must be conscious of how measurement nudges our attention and evolves a sense of moral obligation. As a guide in this journey, Muller’s book is a must-read. Data that calcifies numbers into moral clarification and germinates moral inaction is profoundly problematic—unjust, even. But data as icon that reflects the world and thus creates insight and more conscious action is a beautiful thing.