Rigging the Information Game
Trump's killing government statistics. Don't expect big business to care
It came as a surprise to me, because I am an idiot, to learn that the root of statistics is “state”—as in a sovereign government, not as in Kentucky. It’s true, just ask Wikipedia: the original German meant something like “description of a state or country”. In gross terms, as Western European states consolidated their war-making and tax-raising (and bond-paying) agencies, rulers needed information about what resources they had. Furthermore, sovereign governments were among the few organizations at the time that could compel and tabulate such massive amounts of data. (Nobody thinks like this, but the tables presented in a nineteenth-century U.S. Census, even the fraudulent ones, represent thousands or more person-hours of labor, like the construction of a house or a whole block of houses.) Such were the demands of governments for processing power that the information age was spurred along by developments like the 1890 U.S. Census, for which IBM developed a tabulating machine using punch cards.
Of course, these days “statistics” has another, and probably more prominent, meaning: the mathematical approach or discipline to understanding and presenting information using tools rooted in probability theory. “Stats”, as a course, is much more likely to teach you about how to use linear regression than to instruct you in calculating the number of sheep in Wales. Yet the root meaning still lumbers along, awakened every now and then when there’s a controversy regarding government numbers. For example, Marquette political scientist Phil Rocco’s book Counting Like a State investigates how federal tensions with state and local governments affected the 2020 U.S. Census, while Morton Jerven’s book Poor Numbers looks at how the mechanics of producing one statistic—GDP—work (or, rather, do not work) in the context of African states.
The hook for talking about statistics today, as you might have guessed, is the inevitable politicization of Bureau of Labor Statistics. Donald Trump, proving himself again to be Richard Nixon without the smarts or the scruples, reacted to an embarrassing release of employment statistics by firing the head of the BLS. I mention the Nixon connection because President Nixon tried to do something similar when he suspected the—and there’s no nice way to put an anti-Semitic conspiracy theory—Jews in the BLS were seeking to undermine him with statistics. (I learned about this as a laddish aide de camp at the federal Nixon Library; the museum’s exhibit describing Nixon’s efforts are described in this Slate article. Poignantly, the links to the Nixon Library’s documentation of that scandal—painstakingly collected— have been removed!) Nixon mostly failed in his attempt to purge the BLS (see the link for more) and the BLS presented accurate information regardless.
Whereas Nixon equivocated and could be distracted in his effort, Trump just fired the BLS head. It’s like a nefarious motivational meme: You can just do things!
Jamelle Bouie, the essential Times columnist for the era, reacted reasonably (that is not a joke!) on Bluesky:
I simply want to argue that there is a logic here. It’s appealing to believe that transparent, rigorous, and publicly available government statistics form a necessary public good. And I agree that they are broadly a public good: when good data is available, it provides a benefit to everyone that is not diminished by others possessing it.
However.
We no longer live in an era in which good data or the ability to process it is the exclusive, or even primary, realm of governments. Private corporations are now quite good at collecting and managing data. (Perhaps you’ve heard of, say, Amazon.) And over the past twenty years (maybe more, maybe less, I’m not an expert here), financial firms in particular have become good at turning to alternative indices to get ahead of—and move faster than—government statistics. For instance: do you want to predict economic growth or retail success? Turn to satellite photos for alternative, and in many ways superior, evidence compared to surveys! Do you want to know what the inflation rate really is in a crumbling economy? Collect the data yourself! After all, the United States (from yesterday) is not the only major or Western Hemisphere economy with unreliable economic data; notoriously, countries like Venezuela have long reported dubious stats when they report at all, and China’s prime minister once dismissed his country’s GDP figures as “unreliable”, preferring alternative data sources.
If you have the resources to get better data faster than the federal government, then, you might simply not care that much that official figures are cooked. In fact, you might welcome this development because it offers you a deeper moat. If public goods vanish, after all, those who can afford private goods1 are better off than those who needed the free ones. Yes, even stats can be a realm of class warfare.
The horrifying thing about the collapse of a high-trust society is that at every turn you can imagine how undermining the general good is rational for someone, at least in the short term. And if you want to be hyper-rational about it, if the society is collapsing, it’s better for you to move fast and get what you can first rather than being left with nothing when everything is ruined. You can be sure some merchants in Rome did well supplying the Vandals.
or club goods, you can put your hand down, nerd
I'm a fellow idiot: I thought "statistic" was derived from "baseball." They say learning should be lifelong and I'm confident that's far too short. I particularly appreciate the nerd who has inspired me to learn what "club goods" are. But I wonder where his instructor learned that IBM existed in 1890.