This blog has a recurring argument: the results that political scientists and mathematicians call “impossibility theorems” are not just mathematical curiosities. They are structural descriptions of how institutions actually fail — descriptions that show up, in recognizable form, in the news every week. The thematic tags below are the vocabulary for those recurring structures. They are not topic labels. They are the names of patterns.
Each tag links to its full archive. The anchor posts listed under each tag are good places to start. The older archive posts have been tagged retrospectively; newer posts are tagged on publication. The blog also has a set of Dispatches from the Underground — a shorter recurring feature that applies the same vocabulary to current events. Posts can carry more than one thematic tag; the connections between tags are often the most interesting part.
Conservation of Impossibility
Formal impossibility results don’t get solved — they migrate. Fix the gap here, and it reappears there. Close this loophole, and you open that one. Arrow’s impossibility theorem doesn’t go away when you change the voting rule; it just shows up somewhere else in the system. This is the tag for cases where the structural problem turns out to be the same problem, wearing different clothes — and for the institutional reformers who discover this the hard way.
Anchor posts: Can a Game Know Its Own Rules? · The IRS Is Here to Help. So Is ICE.
Endogenous Base Rates
When a classifier changes, the population it classifies changes too. People who know they’re being observed don’t behave as they would in the absence of the observer — which means the classifier is partly measuring its own effects. A classification system can be technically accurate while producing systematically wrong outcomes, because the act of classifying has already reshaped the behavior it claims to measure. This is the tag for those loops.
Anchor posts: The IRS Is Here to Help. So Is ICE. · All Statistics Are Local · Your Basket May Vary
Signaling
A signal is informative when it is costly — when you have to give something up to send it. Cheap talk (words without consequences) carries no information in equilibrium, because anyone can say anything. A politician who claims to oppose corruption without paying any cost to say it provides no evidence of the claim. This tag collects cases where costly and cheap signals get confused, where the cost of a signal collapses, or where someone uses the structure of costly signaling to strategic advantage.
Anchor posts: What the Prince Doesn’t Know Can Still Get You Fired · Why the Thing That Might Take Your Job Is So Nice To You
The Setter Model
The most powerful position in a decision is often not the voter, but the person who decides what’s on the ballot. Whoever controls the agenda — what gets considered, when, and in what form — shapes outcomes before anyone casts a vote. The setter doesn’t need to have the best alternative. The setter needs to control which alternatives are available when the choice is made. This applies equally to legislative procedure, to the design of algorithms, to the timing of information releases, and to the structure of markets.
Anchor posts: Honest and Effective (Or, “Montana Has a Type”)
The Penalty Ceiling
Any finite penalty schedule creates a flat region at the top where marginal deterrence fails. Raising the ceiling doesn’t eliminate the zone where penalties stop working — it just moves the zone higher. This is one of the most underappreciated structural problems in rule design. It applies to criminal sentencing, to sports regulation, to platform content moderation, and to any system that tries to constrain behavior through punishment. The problem isn’t the level of the penalty. It’s the finiteness of the ceiling.
Anchor posts: Can a Game Know Its Own Rules?
Rational Addiction
Becker and Murphy showed in 1988 that you can be fully aware you’re forming a dependency and do it anyway — because the current-period benefits are real, the future costs are discounted, and present bias is not a failure of rationality but a feature of intertemporal preferences. The rational addict isn’t fooled. They’re choosing, with open eyes, under constraints that make the choice structurally predictable. This tag is about the specific mechanism by which genuinely good products create structural dependencies — and about what that means for anyone trying to design, regulate, or escape them.
Anchor posts: Why the Thing That Might Take Your Job Is So Nice To You
The Measurement Problem
Every aggregate statistic embeds a choice about what to measure, at what level, and whose experience to weight. These choices are not neutral. Simpson’s paradox is the formal name for what happens when you aggregate across subgroups whose internal trends run in opposite directions: the aggregate can accurately represent no one’s experience. This tag collects cases where the way we measure shapes what we can know — and where the gap between the headline number and the lived reality is not accidental.
Anchor posts: All Statistics Are Local · Your Basket May Vary
Adverse Selection
When you can’t distinguish good types from bad types, you end up with a pool dominated by the bad ones — because the good ones have more to lose from the transaction and are more likely to walk away. This is the logic of insurance markets gone wrong, of political scandals, of any situation where the act of offering or accepting reveals something about the party doing it. Adverse selection is what happens when the willingness to participate is itself informative, in the wrong direction.
Anchor posts: Browse the full archive — this tag draws heavily on the 2013–2015 posts.
Flipping the Conditional
P(A|B) ≠ P(B|A). This is Bayes’ theorem, and violations of it are everywhere in political reasoning. The probability that a suspect is guilty given a positive test is not the same as the probability of a positive test given guilt. The probability that a welfare recipient uses drugs is not the same as the probability that a drug user receives welfare. Base-rate neglect — failing to account for how rare the thing you’re testing for actually is — produces false confidence in results that are, on closer inspection, nearly meaningless. This tag is for those confusions, which turn out to be both very common and very consequential.
Anchor posts: Browse the full archive.
Institutional Patches
Informal norms, workarounds, and tacit conventions often do more structural work than the formal rules they sit alongside. Remove the patch, and the underlying problem doesn’t just reemerge — it often emerges in a worse form, because the patch was encoding institutional memory that is now lost. Chesterton’s fence: don’t remove a fence until you know why it was built. This tag collects cases where someone removed a fence without knowing — or caring — what it was holding up.
Anchor posts: Browse the full archive.
Inside Baseball
A recurring series on the formal mechanics of legislative procedure — cloture, discharge petitions, continuing resolutions, the motion to proceed, and the many other instruments by which congressional rules shape (and are shaped by) strategic behavior. The posts in this series assume that the rules themselves are the interesting object of study, not just the context for it. Inside baseball, it turns out, is often the game.
Anchor posts: Browse the full archive — this series ran primarily from 2013–2014 during the government shutdown and debt ceiling episodes.
The blog also uses a set of navigational tags for browsing by subject area: Elections & Voting, Legislative Politics, Executive Power, Measurement & Data, Political Economy, Foreign Policy & Security, and Academic Life.