The Vote Market: What Democracies Actually Trade

For forty years, the best empirical work on policy responsiveness has produced two findings that do not fit together. The first: when the preferences of low- and middle-income Americans diverge from those of the affluent, the relationship between what the less advantaged want and what government does is statistically near zero — a flat line that holds within every policy domain examined, economic, foreign, social welfare, and moral alike (Gilens 2012; Gilens & Page 2014). The second: outcomes that the donor class and prestige media bitterly opposed keep happening anyway — Brexit, the 2016 American election, a string of European populist insurgencies, and, on a forty-year fuse, the overturning of Roe v. Wade. A theory of democracy has to explain both: a system that ignores the median voter on divergence, and a system the elite periodically loses control of.

“Democracy” doesn’t explain this pattern. Neither does “oligarchy.” Both labels misdescribe the mechanism because both treat the question as one of who rules. The better description is a market — with one tradable unit, several distinct costs, and an insurance policy attached. Before formalizing it, watch it run.

The Forty-Year Purchase

In 1979, white evangelical Protestants were a large, electorally diffuse population. In the responsiveness data, diffuse moral preferences are the worst-performing asset in American politics: Gilens’ domain analysis finds the poor’s largest responsiveness penalty precisely on religious and moral issues. Unorganized moral majorities lose, consistently, to everyone.

The founding of the Moral Majority that year, and the broader religious-right institution-building around it, changed the asset class. What was constructed had three specific features: leadership that could credibly warrant delivery of votes, a punishment capability (primary challenges) for politicians who took the votes and defaulted on the payment, and a clear asking price — judicial appointments and abortion policy in exchange for turnout. The arrangement held across four decades, two party realignments, and a sequence of standard-bearers the bloc had every doctrinal reason to reject. It held because the agreement, not the agent, was the asset. Delivery came in June 2022, in Dobbs v. Jackson — an outcome national polling majorities consistently opposed, achieved through the one channel most insulated from electoral repricing: lifetime judicial appointments.

Note what the case is and is not evidence for. It was selected because it satisfies the model’s conditions, so it demonstrates mechanism, not frequency — a process-trace, not a sample. But as process-trace it is close to complete: a diffuse preference becomes tradable only after organization; the payment is made in policy nearly costless to capital (no investor bloc’s returns depended on abortion law); execution runs through blocking and appointment chokepoints rather than legislation; and a national majority is defeated by a warranted minority over forty years. Whatever theory of democracy a reader holds should be able to generate this case. Most can’t. The one developed below predicts it — and makes a forward prediction about what happens to the bloc now, taken up at the end of this section’s return.

The Market and Its Floor

The model: politicians are firms assembling minimum-cost winning coalitions; donors supply capital; voters hold a commodity whose cost of acquisition varies with organization, salience, and information. Thomas Ferguson formalized a version of this in 1995 as the investment theory of party competition — parties as blocs of investors, mass preferences prevailing on issues where no investor bloc has a stake. What follows extends that frame with a floor condition, a cost asymmetry, and a timing anchor, each resting on an independent evidence line.

First, the terms, because “price” in arguments like this one tends to smear across four distinct quantities, and the distinctions matter. The concession price is what a politician pays a bloc for its votes — policy. The coordination cost is what it costs sellers to organize into something that can transact at all. The information cost is what it costs an individual voter to know what their vote is worth. And the entry price is the literal expense of campaigning — roughly sixteen billion dollars for the 2024 federal cycle, per OpenSecrets — which sets the capital barrier for counter-elites. These are related but not identical, and each claim below specifies which one it is about. One more correction to the casual version of the metaphor: the bloc transactions in question are mostly brokerage, not sale. The Moral Majority never owned evangelical votes; its leadership warranted delivery and took its commission in access and standing. That distinction imports the broker’s failure modes — leadership defecting from members’ interests, members defecting from leadership’s warrant — and both will matter later.

The floor condition (a claim about coordination cost). Gilens’ one domain of partial exception to the flat line is social welfare, where policy is somewhat more responsive to the poor and middle class — and his own analysis attributes the exception to coincidence: on healthcare, education, and Social Security, the preferences of the less well-off align with those of powerful organized interest groups. The influence runs through the organizations. Read as market structure rather than footnote, this is the model’s central fact: mass preferences become tradable only when coupled to an organized bloc that can warrant delivery. “Organized” is defined here ex ante, by the three features visible in the evangelical case — warrantable leadership, punishment capability, delivery infrastructure — not ex post by winning, which would make the floor circular. Below the floor, preferences are not cheap; they are non-transactable, because no counterparty can warrant them. This is why diffuse moral majorities perform worst of all in the domain data, and why the same moral preferences, organized, overturned Roe.

The veto asymmetry (a claim about concession price). Branham, Soroka, and Wlezien’s 2017 reanalysis of the Gilens data found the rich and middle agreeing about 90 percent of the time, the rich winning only 53 percent of disagreements — and the inequality that does exist operating mostly through negative power: the rich and middle blocking policies the poor favor. A cost function predicts exactly this concentration, because blocking is structurally cheaper than enacting: a veto requires capturing one chokepoint in a system deliberately built of chokepoints, while enactment requires buying the whole chain. Cheap action gets bought more often than expensive action; the asymmetry lands where the discount is. Two features of the chokepoint system deserve more suspicion than they usually get. First, it is not static: the veto surface has been growing — the routinized filibuster, debt-ceiling brinksmanship, single-judge nationwide injunctions are all twentieth- and twenty-first-century accretions, each adding a purchasable chokepoint that the constitutional design did not contain. Second, every chokepoint comes wrapped in a coordination story — deliberation, stability, minority protection — and the story is sincerely believed by the institutional actors closest to it, who experience the mechanism as the thing that makes governing possible. The positional disagreement is the structure: what reads as coordination from inside the institution prices as extraction from below it, and no one needs to be lying for both readings to be load-bearing.

The timing anchor (a claim about coordination cost, historically priced). A 2023 pooled analysis of Germany, the Netherlands, Norway, and Sweden (Lupu & Pontusson, eds., Unequal Democracies) found policy corresponding far better to affluent than to low- and middle-income preferences in all four countries — and found the one historical exception, left governments’ relative responsiveness to lower incomes in the economic and welfare domain, present before 1998 and gone since. The direction of skew is now uniform across rich democracies; what varied, and varied predictably, was magnitude. Where unions, mass parties, and churches functioned as seller-side cartels, votes were expensive and policy bent toward members. The post-1998 convergence is the cartels’ decay showing up in regressions, on a schedule that tracks the post-Cold War opening of the capital account. Correlation, not demonstrated causation — but a synchronized, cross-national shift with a mechanism attached, which is more than the alternative explanations currently offer. And note the period’s own self-description, because it is the era’s tell: the convergence was narrated at the time as coordination — efficiency, openness, a rising tide, Thatcher’s “there is no alternative.” A distributional shift with an identifiable beneficiary class, presented in the vocabulary of natural necessity, is a move this essay will keep encountering: a maintained arrangement wearing the costume of physics.

Falsification conditions: the model predicts (a) unorganized mass preferences lose at high rates even on low-stakes issues, while blocs meeting the ex ante organization definition win even against affluent opposition; (b) responsiveness rises with electoral proximity and competitiveness, when acquisition costs spike — Gilens documents exactly this around presidential elections and razor-thin congressional control; (c) measures of seller-side organization predict responsiveness gradients across countries and time. A definitive kill: blocs meeting the organization definition failing at rates indistinguishable from unorganized publics, controlling for size. The existing evidence runs the predicted direction on all three; (c) remains consistent rather than conclusive.

The Price-Suppression Technology

A market with chronically low prices on one side needs an explanation, and the standard one — manufactured consent — explains too much. If elite media manufactures mass preferences, every mass outcome is either elite-permitted or elite-fabricated, and no observation can count against the theory. It also fits the record badly: in 2016 the prestige media opposed the winning candidate nearly uniformly and lost, which is hard to square with a unified consent factory.

The defensible version requires no coordination and makes a narrower claim. Start with the behavioral premise the casual version skips: a voter cannot estimate the marginal effect of an outcome without a model of the system the outcome operates in. Event-only information supplies facts without that model. Event-driven journalism — bound to the story unit, dependent on access, optimized for speed and engagement — systematically omits structural context, not by order but by selection: each outlet’s incentive structure reproduces the omission independently, which is why the pattern survives elite fragmentation. Rival factions manufacturing rival narratives still share a single information regime, one that prices in events and prices out structure. The claim, stated at the strength the evidence supports: information asymmetry about policy stakes is a necessary condition for persistently cheap votes, and the current information environment demonstrably lacks the features that would correct it. A voter given discrete facts without a model of how outcomes propagate cannot estimate what their vote purchases or forecloses, and will systematically underprice it.

Two pieces of evidence keep this from being pure functionalism. The causal half — that information environments move voting behavior at all — has experimental-grade support: DellaVigna and Kaplan’s natural experiment on the staggered rollout of Fox News through cable markets found measurable shifts in vote share from channel entry alone. That demonstrates the lever exists; it does not yet demonstrate the mispricing mechanism, and the distinction is held here deliberately. The demand-side half has a documented geography: PRRI’s 2021 survey work on QAnon belief found its strongest predictors clustering where the model says acquisition costs should be lowest — no college, rural residence, lower income, declining group prospects, and a media diet narrowed to one channel-family. The QAnon purchase is the limit case of the transaction: a bloc acquired with pure narrative, no material concession, the story not even required to be true. The scapegoat, the narrative instrument of choice in such acquisitions, prices like subprime credit — almost nothing down, with the costs deferred: foreclosed coalitions, escalation risk, and a narrative that must be continuously serviced to keep performing.

The hypothesis’s tier placement is explicit. Moving it up requires content analysis linking structural omission to measurable stake-mispricing, or further natural experiments on information-environment change — platform migrations and cord-cutting shocks are the available instruments. Falsified by: voters in structurally richer information environments selling their votes no dearer than voters in event-only environments.

The Scale Ceiling

Why don’t the sellers cartelize? The deep constraint is coordination cost as a function of polity size — and there is an inversion of Madison in it that the primary text supports more directly than is usually noticed. Federalist No. 10’s celebrated cure for faction is to extend the sphere: multiply interests across a large republic until no majority faction can cohere. Madison names the factions the machine is built to defeat — the rage for paper money, for abolition of debts, for an equal division of property — which is to say, the debtor-relief politics of organized economic majorities. The extended republic was a vote-price suppression mechanism by design, in the precise sense that its designer enumerated the majority demands it was designed to make incoherent; the modern historiography of the founding’s economic stakes (Beard 1913; Klarman, The Framers’ Coup, 2016) reads the Convention’s architecture the same way.

The claim needs its correct scope, because counterexamples discipline it. Large polities sustain organized blocs at scale — India’s caste-party machines, the mid-century American South as a regional cartel — and the floor condition predicts those blocs transact successfully, which they do. What scale defeats is the economy-wide seller cartel: Denmark organizes six million people in one labor market under one bargaining regime, while an equivalent American cartel would need 330 million people across a continent, against employers arbitraging fifty regulatory regimes and, since the late twentieth century, an open capital account. Germany’s sectoral bargaining at 84 million shows the ceiling is not a hard wall.

But notice what the ceiling is actually made of, because the inventory matters. Population is given; everything else in the list is enforced. Fifty arbitrageable regulatory regimes are a maintained feature of American federalism — federal labor law preempts state variation wherever Congress says it does, and where it doesn’t, that is a standing decision. The open capital account is a policy with a documented off-switch: capital controls were the operating norm of the rich world from 1945 to the early 1970s, dismantled by identifiable legislation and treaty. Even the founders’ contribution was, by construction, construction. A constraint that has to be continuously enforced is not a law of nature; it is a rule with a maintenance budget, and rules with maintenance budgets have repeal paths. The scale ceiling is real, binding, and expensive to move — but “expensive to move” is a price, not a physical constant, and the essay’s own discipline requires saying so. The same correction applies to the organization floor: its height is itself set substantially by law. The Wagner Act of 1935 lowered American coordination costs by statute, and union membership quintupled in the decade after; Taft-Hartley in 1947 raised them again, and the long decline dates from roughly there. Labor law is legislation about the height of the floor — which means the floor is an object of political contest, not a precondition of it. Nothing load-bearing in this essay is a mountain. Every constraint in it is maintained by someone, at some cost, and the system’s deepest rhetorical move — already met above in the era of “no alternative” — is describing maintained arrangements as physics.

The cross-national regression of responsiveness on union density, polity scale, and capital openness remains the test, and the hypothesis is stated as exactly that — a hypothesis, falsified by large polities reaching Nordic-gradient responsiveness at low organization cost.

Repricing Events

The model’s strongest test is the moments that look like its refutation: when the demos seemed to win outright, and when elites visibly lost control. Both turn out to be price events.

The New Deal and Bretton Woods settlement — capital controls, confiscatory top marginal rates, the postwar labor accord — is routinely cited as the demos voting its way to a different capitalism. The architecture says otherwise: drafted by elites (Keynes, Harry Dexter White), sold by an elite (Roosevelt, explicitly, as saving capitalism from itself), implemented through existing institutions. But it was no gift either. Two distinct mechanisms had spiked the price, and they should be kept separate because they operated separately. Domestically, the sellers had cartelized: union membership quintupled from 2.8 million in 1933 to 14 million by 1945 — from under 13 percent density to a 34 percent peak — and the 1934 strike wave in Toledo, Minneapolis, and San Francisco had demonstrated coordination capacity before the Wagner Act ratified it. Externally, a competitor regime was offering the world’s workers an alternative settlement entirely, which put a revolutionary tail on the cost of not buying mass acquiescence. The elite repriced under both pressures. And the reversal is timestamped: when the competitor collapsed and capital recovered its exit option, the price collapsed with it — which is what the post-1998 disappearance of left-government responsiveness in the European data records. The demos’s greatest victory, examined closely, was a renegotiated rate, since renegotiated back.

The ruptures point the same direction, darker. Brexit and 2016 are routinely read as the masses breaking elite control. Look at the capitalization tables: Brexit’s leadership was a hedge-fund and press-baron faction arrayed against the CBI consensus; 2016 was a counter-elite with its own media stack against an incumbent elite with the legacy one. These are hostile takeovers — a rival elite spots mass grievance the incumbent coalition has been ignoring, because ignoring it was until that moment free, and arbitrages it. The demos appears as the undervalued asset, not the acquirer. The model also supplies what a theory of takeovers needs and capture theories lack: an account of incumbent error. The same information regime that suppresses vote prices degrades the incumbents’ own price discovery — an elite that manufactures the narrative environment ends up consuming it, and mispriced grievance accumulates longest exactly where the narrative monoculture is densest. That converts the rupture cases from anomalies into a prediction: takeovers should cluster in polities and periods where elite information feedback is most closed. The civil rights movement, the strongest apparent counterexample to the takeover reading, fits the floor condition instead: a genuinely organized seller bloc whose purchase was closed partly because Cold War state elites needed the product for legitimacy abroad (Dudziak, Cold War Civil Rights, 2000). Mass organization was necessary. It was not sufficient until a buyer appeared.

And the evangelical arc now returns with its prediction. Brokerage has a known endgame: a paid seller has nothing left to withhold, and a broker who has delivered has nothing left to warrant. The model predicts the bloc’s leverage — defined across all four of its channels: candidate selection, agenda-setting, judicial pipeline influence, and turnout bargaining power — should decay materially within two presidential cycles of Dobbs. The early post-delivery data (abortion-referendum losses, suburban realignment) runs the predicted direction. If bloc leverage across those channels is undiminished through 2028 and 2032, the prediction fails and the model loses its dynamic component. That is the essay’s cleanest kill condition, and it has a clock on it.

The Insurance Policy

Which brings the argument to what elections are. Not cheap — sixteen billion dollars a cycle is the entry price, not a rounding error — but low-variance, and the distinction is the point. The implicit premise deserves stating outright: for actors holding durable assets, variance dominates expectation — a bounded, recoverable loss is strictly preferable to a small probability of total loss. Consider the removal problem as a risk profile. Removing an American president requires an election: expensive, bounded, rule-governed; even its worst recent stress test stayed within bounds. Removing the ruler of Russia or China requires a coup, a succession crisis, or a collapse: low routine cost, unbounded tail. Popper’s minimalist defense of democracy — its value is not that it expresses the popular will but that it removes rulers without bloodshed — is this point restated as variance reduction. Democracy converts the catastrophic tail risk of succession into a continuous premium payment.

One scope discipline: this is a claim about maintenance, not origin. Elections were not invented as elite insurance, and the functionalist version of the claim would be unfalsifiable. The claim is selectionist: among the mechanisms a fractured elite can live with, bounded rotation is the one whose losers survive to compete again, which is why elite factions keep paying for it — and why the system’s most dangerous moments are precisely those in which a faction concludes it cannot afford the next premium. The policyholder is not the demos. The faction that loses an election keeps its assets, its networks, and its next chance; under the alternative mechanisms, losing factions historically keep none of the three.

What Kind of Regime Is This?

A reader is entitled to ask whether all this describes democracy or documents its absence, and the question deserves a direct answer rather than an elliptical one. The answer is that the dichotomy is the residue of the idealized baseline this essay opened by discarding. Every durable regime runs on a legitimacy formula under which the population ratifies elite governance, plus a withdrawal clause for when divergence grows intolerable; the dynastic versions priced the clause in blood. What distinguishes the regimes called democracies is not that the demos rules — by default, on the evidence above, it rarely does — but that the withdrawal clause is cheap to invoke and the ratification is genuinely repriced at intervals. The pre-1998 European gradient and the Dobbs arc both show the system is contingent, not fixed: where sellers organize, the price schedule moves, sometimes a long way. That is a smaller thing than popular sovereignty and a larger thing than a sham. The honest name for it is a constrained exchange system with uneven tradability — and the live political question within it is never the binary “do the people rule,” but always: what does the current price schedule look like, and who can afford to change it?

Alternative Explanations Considered

The coincidence reading. Branham, Soroka, and Wlezien’s own interpretation of their reanalysis is deflationary: rich and middle agree 90 percent of the time, the rich win disagreements barely more than a coin flip, so perhaps there is no structured inequality to explain. This is the simplest alternative and gets the strongest engagement. Three documented patterns resist it. The negative-power finding is itself structured: the asymmetry concentrates in blocking, exactly where a cost function predicts and where coincidence alone predicts nothing in particular. The cross-national 1998 break is a synchronized, dose-like shift that a coincidence account leaves unexplained. And — the deepest problem — the 90 percent agreement is measured over proposed policies, while the veto operates partly upstream of proposal. Non-decisions never enter the denominator: the alternatives that organized money keeps off the agenda entirely are invisible to a dataset built from survey questions about live proposals. This is the second face of power (Bachrach & Baratz 1962), and Schattschneider’s older formulation — that the definition of the alternatives is the supreme instrument of power — is a claim about exactly the stage of the process responsiveness studies cannot see. The agreement zone is real, but its boundaries are partly a product of the mechanism it is offered as evidence against. And the claim is testable rather than merely deflationary, because the proposal gate has a partial bypass that varies by jurisdiction: the ballot initiative. Initiative states let preferences reach the agenda without passing through the brokered proposal stage; if the non-decision argument is right, initiative and pure-legislature states should show systematically different policy mixes precisely on the issues where mass and affluent preferences diverge — minimum wages, marijuana, Medicaid expansion are the visible candidates. If the policy mixes don’t differ, the counterfactual agenda was a phantom and the rebuttal falls.

The capture reading. The strong Gilens-Page interpretation — economic elites dominate, full stop — fails against the same evidence from the other side: the social-welfare responsiveness channel, the election-proximity effect, the takeover ruptures, and the Dobbs case, in which an organized non-elite bloc extracted forty years of performance and, eventually, the policy. Capture predicts none of these; a market with an organization floor predicts all of them.

The institutional-design reading. Perhaps this is an American pathology of money-saturated, weak-party majoritarianism, and democracy is fine elsewhere. This was the strongest available objection until the four-country European data substantially closed it: institutional design moves magnitude — that is the seller-cartel variable, and the model needs it — but the direction is now uniform across rich democracies, and the historical exception decayed on the schedule the model predicts.

What This Changes

If the analysis is right, the standard reform program is aimed at the wrong side of the market. “Getting money out of politics” attacks buyer-side purchasing power, which a century of campaign-finance whack-a-mole suggests is the most adaptive variable in the system. The binding variable on the evidence above is seller-side organization — the floor below which preferences are not tradable at all. The European data’s grim addendum is that even strong seller cartels decayed once capital could exit the jurisdiction; any organizational answer that ignores the exit option is re-running 1998.

Evidence Framework

Documented in Public Records (Tier 1):

Near-zero preference-policy link for low/middle-income Americans when diverging from the affluent; pattern holds within economic, foreign policy, social welfare, and moral/religious domains; largest 10th-percentile penalty on religious-values issues: Gilens, Affluence and Influence (Princeton, 2012); Gilens, Swiss Political Science Review (2015).
Social-welfare partial exception attributed to alignment with organized interest groups, not influence of the poor; responsiveness rises with presidential-election proximity and even partisan division; parties behave as policy maximizers oriented to activists and interest groups: Gilens (2012).
Rich and middle agree ~90% of the time; rich win disagreements 53%; observed inequality operates mostly through negative power (blocking): Branham, Soroka & Wlezien, “When Do the Rich Win?”, Political Science Quarterly 132:1 (2017).
Affluent-skewed responsiveness in Germany, Netherlands, Norway, Sweden; left-government responsiveness to lower incomes in the economic/welfare domain present before 1998, since disappeared: Lupu & Pontusson (eds.), Unequal Democracies (Cambridge, 2023), ch. 2.
Investment theory of party competition: Ferguson, Golden Rule (Chicago, 1995). Cited as published theory.
Staggered Fox News cable entry measurably shifted Republican vote share: DellaVigna & Kaplan, Quarterly Journal of Economics (2007).
QAnon belief predictors — no college, rural, lower income, narrow far-right media diet: PRRI (May 2021).
Moral Majority founded 1979; Dobbs v. Jackson (June 2022); national polling majorities consistently opposed overturning Roe: Gallup/Pew archives.
Union membership 2.8M (1933) → 14M (1945), ~36% of the workforce; density 12.8% (1935) → 34.2% peak (1945); 1934 strike wave (Toledo, Minneapolis, San Francisco): BLS Handbook of Labor Statistics historical series; CRS R47596; standard labor histories.
Wagner Act (1935) statutorily lowered organizing costs; Taft-Hartley (1947) raised them (closed-shop ban, state right-to-work authorization); union density decline dates from the late 1940s–50s plateau: statutory texts; CRS R47596.
Capital controls were the postwar norm under Bretton Woods (1945–early 1970s), dismantled by identifiable legislation and treaty (US 1974; UK 1979; EU directives through Maastricht): standard international political economy record.
“There is no alternative” as the era’s self-description: Thatcher, attributed slogan, 1980s; cited as period rhetoric, not analysis.
2024 US federal cycle total cost ~$15.9–16 billion, a nominal record: OpenSecrets (Oct.–Nov. 2024 reports).
Federalist No. 10’s enumeration of target factions (paper money, abolition of debts, equal division of property): primary text. Economic reading of the founding architecture: Beard (1913); Klarman, The Framers’ Coup (2016).
Cold War foreign-policy stake in civil rights progress: Dudziak, Cold War Civil Rights (2000).
Agenda-stage power and non-decisions: Bachrach & Baratz, APSR (1962); Schattschneider, The Semisovereign People (1960). Cited as theory with its own empirical literature.

Pattern Inferences from Tier 1 Facts (Tier 2):

The organization floor: preferences become tradable only when coupled to a bloc with warrantable leadership, punishment capability, and delivery infrastructure (from: the social-welfare interest-group mechanism, the worst-of-all performance of unorganized moral preferences, the Olson coordination logic, and the Dobbs arc).
Veto asymmetry as cost asymmetry: blocking is cheaper than enacting in a chokepoint-rich system, so inequality concentrates in negative power (from: Branham et al. plus US institutional structure).
Bloc transactions are brokerage, not sale: leadership warrants delivery rather than owning votes, importing broker-defection and intra-bloc agency conflict as failure modes (from: the documented structure of the religious-right arrangement).
The New Deal/Bretton Woods settlement as elite-drafted repricing under two separable pressures — domestic seller cartelization and external regime competition (from: documented authorship, verified union growth and strike capacity, documented framing as preserving capitalism).
The post-1998 European reversal tracks the post-Cold War, post-liberalization collapse in labor’s vote price (from: documented timing against the liberalization timeline).
Elections as low-variance elite rotation, maintained (not originated) because losing factions survive (from: documented cost/boundedness of electoral removal vs. historical variance of non-electoral succession).
Measured rich-middle agreement is conditioned on the proposal agenda, which the veto partly controls; non-decisions are excluded from the denominator (from: the documented method of the responsiveness studies plus the documented blocking asymmetry).

Structural Hypotheses Requiring Additional Evidence (Tier 3):

Information asymmetry as price suppression: the event-driven information environment lacks the features that would let voters price policy stakes, a necessary condition for persistently cheap votes. (Up-tier with: content analysis linking structural omission to stake-mispricing; natural experiments on environment change. Falsified by: equal vote-pricing across structurally rich vs. event-only environments.)
The scale ceiling, scoped: polity size and internal regulatory/capital arbitrage bound economy-wide seller cartels, while organized blocs remain viable at any scale. (Up-tier with: cross-national regression of responsiveness on union density × scale × capital openness. Falsified by: large polities reaching Nordic-gradient responsiveness at low organization cost.)
Post-delivery bloc devaluation: evangelical leverage across candidate selection, agenda-setting, judicial pipeline, and turnout bargaining declines materially within two presidential cycles of Dobbs. (Falsified by: undiminished leverage across those channels through 2028/2032.)
Elite error as information feedback: incumbent mispricing of mass grievance, and hence takeover ruptures, cluster where elite narrative environments are most closed. (Up-tier with: comparative measures of elite information-source diversity against rupture incidence.)
The veto surface is growing: purchasable chokepoints (routinized filibuster, debt-ceiling brinksmanship, nationwide injunctions) have accreted beyond the constitutional design, so the blocking discount is widening over time. (Up-tier with: longitudinal count of effective veto points × usage rates. Falsified by: stable or shrinking effective chokepoint counts.)
Suppressed vote prices may be load-bearing: there exists a structural-information threshold above which mass consent cannot be purchased at rates the system will pay. (Conceptual as much as empirical; the named test is the structural-context threshold measurement in Unresolved Questions.)

Unresolved Questions

First, is the 90 percent rich-middle preference agreement exogenous, or partly produced by the information regime and the proposal agenda jointly? The essay uses the agenda half (documented method limitation) and declines the information half (unfalsifiable as usually deployed) — but the information half is testable in principle: preference movement following documented environment shifts is measurable, and nobody has measured it well.

Second, can economy-wide seller cartels form at continental scale under open capital accounts, or is the post-1998 convergence an equilibrium rather than a trough? The answer determines whether the organizational program in “What This Changes” is a strategy or a eulogy.

Third, the brokerage frame raises a question this essay opens but does not settle: when bloc leadership is paid in access and members are paid in policy, on what schedule do the two payment streams diverge, and what does member defection look like when they do? The evangelical case will generate the data either way.

Fourth — and this is the darkest open question the analysis generates — the information section assumed throughout that correcting voter mispricing would be an improvement. But there is a threshold question underneath it: is there a level of structural understanding at which votes become too expensive for the system to clear at all? The postwar settlement suggests fully priced consent can be purchased, but only at rates the buyers abandoned at the first opportunity. If persistently suppressed prices are not a defect of the system but a load-bearing component of it — if mass consent at full information is unaffordable — then the reform program in “What This Changes” is not a repair but a stress test, and nobody knows the failure mode. The honest statement is that the structural-context threshold for accurate stake estimation has never been measured, and neither has what happens when an electorate crosses it.

The Measurement

There is one institutional fact this essay’s entire evidence base rests on, and it should be stated as the closing demand because it is the smallest possible one. Every responsiveness finding cited above exists because individual academics spent years hand-assembling preference-policy datasets that no state collects. The survey infrastructure exists; the policy coding exists; the method is two decades old and replicable. Yet responsiveness auditing is nowhere an institutional function of any democratic state — the system that names itself for popular rule does not measure, as a matter of public record, whose preferences its policy tracks. Under the model above the omission is legible: the audit would publish the price schedule. But that is precisely why the demand is the right one. It requires no new theory, no constitutional change, and no resolution of anything this essay leaves unresolved. A standing, public, state-funded audit of policy responsiveness by income, organization, and issue domain — the price schedule, published. Refusing it requires explaining why a democracy is better off not knowing. The refusal would be the finding.