Thomas Schelling's documented contribution, the rationality of irrationality, threats working best when the threatener might be unable to back down, precedes the coinage: Nixon's version, per Haldeman's memoirs, was explicit, I want the North Vietnamese to believe I've reached the point where I might do anything to stop the war, the madman theory, operationalized in the 1969 nuclear alert, bombers airborne with weapons, meant to be detected by Soviet intelligence. The documented outcome: Hanoi was not visibly deterred, the war continued, and the doctrine's own ledger remains contested, the 1969 alert being the theory's largest documented test and its least documented success. The game-theoretic core survives in the literature: a reputation for unpredictability extracts concessions, at the documented price of being unable to make credible promises.
The folk economy of the doctrine runs below the nuclear shelf: the erratic boss, the volatile negotiator, the player who flips the board, each collecting the madman premium, deference purchased with apparent instability, and the folk tradition's documented counter: the premium decays, unpredictability being a wasting asset that must be re-proven with escalations, until the madman must either do the mad thing or lose the reputation, the doctrine's own logic pushing toward the act it was designed to avoid. Bloom's interview files the personal version: being slightly unpredictable as competitive advantage in social games, the same premium at kitchen-table stakes.
The claim: the madman theory is the archive's deepest note on the value of illegibility, the player who cannot be modeled cannot be played, the doctrine being the trickster file's state edition, and its documented history being the archive's caution: the strategy's greatest practitioner ended the 1969 alert with nothing, was read correctly by Hanoi as a bluff, and lost the war anyway, the madman premium existing only where the adversary's intelligence is worse than the act. Verdict: the archive's verdict is the record's: feigned madness works on audiences, not on adversaries with archives of their own, and the doctrine's real product is internal, the domestic belief that the leader might do anything being exactly as useful to the leader as the foreign one, which the archive's psyop wing files under dual-audience operations. The bombers flew. Hanoi read the file. The theory survived its failure, as all beautiful theories do. The archive's note: the unreadable player is also unreadable to his own side. Ask the archives. They are the only ones who ever knew what the madman would do next, and they knew it in 1969.
If unpredictability deters, what deters the unpredictable, and how do you negotiate with a strategy that is paid to be unreadable?