Speaking about the forthcoming presidential election and the prospects of Donald Trump, an Iowa GOP official said that “anything can happen” (“The Inevitable Nominee? Trump’s Election Momentum Builds Even as Legal Problems Mount,” Wall Street Journal, August 27, 2023).

In light of what we know about the economics of politics, this reflection is more than just an intuition from a minor party official. The idea that anything can happen in political space is supported by science—logic, mathematics, economics, and some historical and empirical confirmations of political chaos. In his influential 1982 book Liberalism Against Populism, political scientist William Riker summarized what was already known. I am quoting a whole paragraph (from p. 187), but the penultimate sentence gives the concise result:

In a remarkable discovery, [the late MIT economist Richard D.] McKelvey has shown that, for a wide class of differentiable utility functions, once “transitivity breaks down, it completely breaks down, engulfing the whole space in a single cycle set. The slightest deviation from a Condorcet point (for example, a slight movement of one voter’s ideal point) brings about this possibility.” Hence not only is a Condorcet winner unlikely, but also when one does not exist, anything can happen. There is no “small” set of probable outcomes.

(I have explained elsewhere the Condorcet Paradox and related theories as well as some of their implications for a realistic conception of democracy and elections; see my Independent Review article “The Impossibility of Populism” and also my Regulation review of Riker.)

A breakdown of the rule of law, a civil war, the Argentinization of America, or something banal like the election of Joe Biden or Mike Pence—any of that can happen. All of this follows from rational-choice analysis; it assumes that individual voters are rational. (Dropping this assumption, we could envision the possibility that if Caligula’s horse were running under the QAnon banner, the horse might win.)

These results must be qualified. As long as certain institutions—courts, constitutions or fundamental laws, political parties, government assemblies, bureaucracies, decentralized power centers, a free press—frame and constrain voting and its consequences, ordinary times will display more standard and predictable results, which public-choice analysis has explained. When the institutional structure of a society is under heavy strain, like in America today, political processes do resemble a roulette wheel. And we can’t seek refuge in “ordered anarchy” (a Buchananian expression), because this is not the situation now and it is very unlikely to be after the political wheel is spun.