Hello everybody, just a quick note at first because there are some new people here. Welcome! I hope you like my silly posts. If you are new to me, I am an epidemiologist (by training), a statistician (by profession, reluctantly), and a longtime leftist (by an inborn tendency to masochism, I guess). I am working on this newsletter as part of a long and open-ended collaboration with Nate Holdren developing a theory of social murder in service of a Marxist analysis of COVID-19 and public health more generally. I am also in the proposal phase of writing a critical book about statistics — the idea, at this point, is that it will teach you some statistics as it simultaneously teaches you to hate statistics. Fun! You may send me distance reiki, good vibes, prayers, incantations, and whatever other psychic tokens you’d like in support of this. Whatever.
I mention these two strands of my “work” not only because there are some new readers here, but because the two converged in a project I worked on in my day job recently. The details of the project aren’t important; what is important is that it was a data analysis related to the surgical treatment of carpal tunnel syndrome. You’ve certainly heard of carpal tunnel syndrome — remember Janice’s carpal tunnel from pulling espresso shots in The Sopranos? I was interested to discover, in the course of the project, that there actually is no “gold-standard” diagnosis for carpal tunnel syndrome.
The carpal tunnel is the name given to the little lacuna in the bones and ligaments, located on the palm side of the hand, that the important “median nerve” passes through to connect the hand to the wrist. Carpal tunnel syndrome is one of a number of so-called “repetitive strain injuries” and is most often caused by — you guessed it — repetitive hand motions that can cause the carpal tunnel to narrow, pinching the median nerve, causing a number of symptoms like pain, numbness and tingling, and in many cases resulting in inability to use the affected hand.
A random book I picked up recently from a used bookshop — Allard E. Dembe’s Occupation and Disease: How Social Factors Affect the Conception of Work-Related Disorders — devotes nearly a hundred pages to carpal tunnel and other “cumulative trauma disorders” (another, more vague, term for “repetitive strain injuries”). I have not finished reading this chapter, but two interesting aspects of the book are already apparent. First — I will take this back up shortly — how the politics of scientific evidence come into play in instances of complex causality. Something like the loss of a finger is causally straightforward. Second, repetitive strain injuries have been recognized since, literally, antiquity (not Hippocrates describing the arm paralysis of a worker long engaged in twisting twigs!). Unsurprisingly, repetitive strain injuries of the hand, wrist, and arm exploded with the industrial revolution and corresponding increases in repetitive, mechanized factory work and also the associated clerical work required to do capitalism. “Writers’ cramp”: worse than it sounds.
The more recent history of carpal tunnel in Dembe’s book, which I haven’t had a chance to read closely yet (just bear with me, I’m a hardworking single mom to two neuroic cats) involves the medical search for some kind of definitive pathological lesion, a dispositive physical sign. The search for such a lesion — while it led to a huge wave of psychologizing discourse about how carpal tunnel syndrome is a mass hysteria — has been and continues to be totally fruitless. Physicians mostly diagnose carpal tunnel syndrome “clinically” (that is, just by examining somebody and asking them questions) rather than via a diagnostic test. Some “objective” diagnostic tests do exist, but they can’t be too objective, because they don’t work. So why do we need to “biocertify” carpal tunnel at all? Why, as Ellen Samuels (who, as far as I know, coined the term “biocertification”) might say, do we need to authenticate a person’s experience of carpal tunnel syndrome with a biological or physiological sign?
Well, probably because of the stinginess and cruelty of America’s workers’ compensation system. I read an interesting article from 1990 on the “social recognition of repetitive strain injury” in which the author, through a detailed comparison of the epidemiology of repetitive strain injuries, data collection on workplace hazards, and workers’ compensation practices in Australia and the United States, ultimately argues that it was the workers’ compensation systems that either facilitated (in Australia, where physician diagnosis leading to compensation was fairly easy and straightforward) or hindered (in the United States, where biocertification was administratively burdensome and the process of claiming benefits adversarial) recognition of the scale of the problem.
Nate wrote an excellent Marxist social history of workers’ compensation laws, which I am tempted to digress in order to summarize here; I will not do that and instead urge you to buy the book. I will, however, mention just a few points he raises. With the passage of workers’ compensation legislation in the US, Nate convincingly demonstrates that the “tyranny of the trial” (whereby workers had to rehearse and perform their injuries in a courtroom to have a chance at recompense) was replaced by the “tyranny of the table,” which abstracted injuries from the workers-as-people in order to aggregate them, rank them, and fix dollar amounts to them. (Another piece of late-90s/early-oughts nostalgia, remember Michael Moore’s Sicko? At the beginning of the movie, we meet a man named Rick who lost two fingers to a table saw; uninsured Rick had to choose between reattaching his ring — $12,000 — or his middle — $60,000 — fingers.)
Nate argues that the passage of workers’ compensation laws actually resulted in a “moral thinning” of the concept of injury itself — these laws effectivey voided it of all but direct pecuniary/liability considerations, which includes voiding it of meaning and denying injured workers the dignity and justice that comes with recognition of that meaning, the meaning of the injury for them and how it affected them. There are many rich strands in Nate’s book that I don’t have the time, space, or inclination to draw out further here — like the law as part of the state, as a facilitator of “social murder” (the depoliticized killing and maiming of working-class people built into the structure of capitalism itself and the major problematic Nate and I are trying to work with).
Indeed the system of law around workers’ comp and other occupational health issues deals very awkwardly with social murder. You will recall that around this time last year the California Supreme Court handed down a ruling in Kuciemba v. Victory Woodwork, Inc. holding that employers can’t be held liable for “take-home COVID” (COVID, by the way — classic occupational disease, devastating example of social murder at scale). A fair amount of the court’s reasoning was that, essentially, COVID was too widespread to pin on an employer, that the scale of social murder represented by COVID was too big to regulate. And that’s kind of, in a sick way, really right — viewed through a “biopolitical” lens, the workers’ compensation system is set up to facilitate social murder, not to ameliorate it, and to make sure no single employer can be held too responsible for it.
There are two ways that I see the story of statistics I’ve been developing as relevant here. First, I’ve been reading a lot of French dudes, specifically those of the “convention theory” school of science and technology studies (STS) — some of the only people I have found thus far to examine statistics from a “science and society” perspective. These guys are hugely indebted to Foucault (who I love, even though he basically never calls what he’s describing capitalism even as he’s describing capitalism) and have recounted the history of statistics from an STS perspective focusing on the abstracted epistemological procedures of statistics as well as its boring but huge (and hugely important) administrative operationalization. What do I mean by abstracted epistemological procedures? I mean what the convention theorists talk about as creating, from the conceptual materials and space of possibility afforded by historical context, new objects that enable new forms of knowledge and new forms of power. (Seriously, who is going to buy me the Foucauldian power/knowledge streetwear?) In the case of statistics, it is in “encoding” individual instances as members of a class (aggregating numerous distinct cases of carpal tunnel syndrome into risk groups, for example) and creating “classes of equivalence” between different types of encoded objects to facilitate analysis, comparison, and valuation, like the assignment of monetary value to fingers on a hand.
Second, the idea of causality currently de rigeur in epidemiology (my home discipline, in case you have forgotten by the end of this long ass post) and the corresponding school of quantitative/statistical “causal inference.” There have been many critiques of the causal inference framework already penned. There will, furthermore, be a very long and detailed critique of causal inference methods in the statistics book I mentioned, and possibly in abridged form in this newsletter as well. Causal inference purports to be able to infer causal relationships from observational data, which are notoriously thorny, messy, and beset with tons of systematic biases as compared to neat, clean data from randomized controlled trials. It accomplishes this by dressing up the same old associational statistics — “correlation is not causation” — in a ridiculous costume of absurd, unverifiable assumptions that supposedly enable a researcher to “interpret” a statistical association as a causal relationship in the real world. I’ve got tons of issues with this, as you can probably already tell, but the relevant one here is that the project of causal inference is eerily parallel to the project of establishing liability in cases of complex causation, like occupational repetitive strain injuries, or clusters of cancer following environmental contamination, and so on. Why do we need to “certify” causes mathematically? Because we need to know how to apportion liability, that’s why.
I am ending here because this is already too long and I didn’t intend to do this today — just wanted to publish something (anything!) for the new folks. This is the type of shit I’m into, and I hope you’ll stick around!