In reference with a previous article, I got to talking with Claude, and he opened my eyes a bit more. Turns out, I wasn’t being skeptical of official numbers enough!
Claude:
States don’t systematically lie about things where:
- The information is independently verifiable by large numbers of private actors (stock prices, weather, sports scores)
- The lie would immediately cost money — markets are merciless about certain kinds of official fiction
- The bureaucratic incentive runs toward accuracy rather than concealment (tax collection, census, infrastructure engineering)
- Career consequences for the relevant officials run toward truth-telling (a bridge engineer who lies about load tolerances kills people, including powerful people)
States lie systematically about:
- Things where they have monopoly control of the information pipeline (military operations, intelligence, classified programs)
- Things where the lie is diffuse and truth is technical (casualty classification, inflation methodology, unemployment definitions)
- Things where the relevant audience can’t easily organize around the truth (foreign civilian casualties — who’s counting Iraqi wedding party deaths?)
- Their own past conduct, especially when accountability would be institutional rather than individual
The useful heuristic: the state is roughly as honest as the market pressure or independent verification forcing it to be. Where those forces are strong, it’s mostly honest. Where they’re absent, assume managed narrative as the default, not the exception.
One important nuance though: this isn’t uniquely a state problem. It’s a large-institution-with-monopoly-information problem. Corporations do the same thing with product safety data, side effect profiles, actuarial models. The Catholic Church did it with abuse. Universities do it with endowment performance and graduation rates. The mechanism is identical — distributed incentive structures, definitional discretion, no single node lying, aggregate output systematically skewed.
The state is just the most consequential instance, and the one with the most developed apparatus for generating legitimacy around its managed narratives.
What this means practically: for any claim where a large institution has monopoly information and a strong interest in a particular answer, the epistemically correct prior is skepticism proportional to the size of that interest.
End.
Notes:
Sounds like Baumeister’s “Evil: Inside Human Violence and Cruelty” (1997). Bureaucratic task segmentation enables atrocities by ensuring no individual perpetrator sees the entire chain. The train scheduler doesn’t see what happens at the destination; the camp administrator doesn’t pull triggers; the guard doesn’t set policy. States structurally distribute moral responsibility across so many nodes to ensure individual protection from aggregate moral accountability.
Also, I doubt the “large institution=state” proof of nuance. Corporations have state-granted liability shields. In science, you have the FDA, regulations, and state privileges. State universities are not private. Churches have a monopoly within their culture and camp. This requires more, however.
