What “Vote on Values” Actually Does
Authorization is not verification
Corpus frame
The corpus applies one lens to many domains: what mechanisms produce the outcome? It shares four methodological commitments and one explicit directional commitment. Each linked page argues for its part; the links are derivations and disputes, not evidence inherited by every page. The directional commitment does not by itself settle system boundary, distribution, sacrifice, or institutional authority.
- Mechanisms are what act. Incentive gradients, selection pressures, feedback loops, and capital stocks produce the distribution of outcomes. Intentions, labels, official categories, and stated values are evidence about mechanisms, or are themselves coordination mechanisms. They are not causal substitutes. — Mechanism Realism · Only Selection
- The reference telos is sustained flourishing. The broadest achievable adaptive safety margin over deep time — not the continuity of any incumbent state, coalition, institution, or doctrine. A mechanism's own stated goal can still serve as a local proof obligation — showing that its incentives defeat even the purpose it claims is a bounded finding — but meeting that goal establishes nothing about the margin. — Flourishing Is Maximum Safety Margin
- Law, rights, legitimacy, democracy, markets, and sovereignty are mechanisms under evaluation. They are constraints, carriers, or proxies inside the analysis. None is a terminal value or a boundary of what is real. Treating one as terminal ends the mechanism search before it starts. Evaluation carries current function, replacement cost, path dependence, uncertainty, capture risk, reversibility, and who bears model error into the ledger. — The Stack · Mechanism Space
- Optimization is a system function. A civilization has to build, exercise, and revise metamechanisms that search mechanism-space, discard dominated options, install, observe effects, and repair under uncertainty. Not running that loop leaves margin unrealized, and that is itself the failure. No single component — analyst, model, or institution — is presumed to contain a global optimum; the capacity is a property of the system. — From Telos to Policy · The Three-Layer Architecture
- Uncertainty is preserved, not spent. Partial orders, binding constraints, unknowns, and residuals stay explicit. An unmeasured effect is not a favorable default. — The Compression Paradox · Cargo Cult Epistemology
Each essay bears its own evidence. Links carry definitions, derivations, applications, and disputes; they do not transfer proof. Criticism is answered on its substance.
Voting does not test whether a value works. It tests whether enough voters currently endorse the package that carries it. Authorization is not verification.
I. The Standard Model
The most sophisticated version of democratic theory goes like this: vote on values, bet on beliefs.
Robin Hanson's futarchy separates the two functions of governance: deciding what to want (values — voted on democratically) and deciding how to get it (beliefs — settled by prediction markets). The architecture assumes these are orthogonal axes you can independently vary. Citizens contribute the "what for." Markets contribute the "how."
This is the best proposal available within the standard ontology. It takes seriously both the democratic intuition (people should determine their own goals) and the epistemic problem (crowds are bad at causal reasoning). It is also built on a foundation that dissolves under examination.
The foundation: values are a kind of thing that can be meaningfully voted on.
What does this actually mean? What is the mechanism? What does "voting on values" physically produce? Not what does it intend — what does it do?
II. What the Mechanism Produces
Values can be analyzed as replicators. They transmit, mutate across contexts, compete for attention, and face selection pressure. Reflective revision and agency operate through and against those dynamics.
Voting also represents, authorizes, aggregates, and enables succession. The claim under examination is that preference aggregation can amplify packages with high current transmissibility without a long-horizon host-effect test. That is one output of the mechanism alongside those democratic functions.
The mechanism under examination: Democratic preference aggregation can amplify value packages with high current transmissibility. Transmissibility and adaptiveness are different properties under different selection pressures.
The structural reasons this divergence is systematic rather than occasional: temporal feedback asymmetry (political benefits arrive within one electoral cycle, policy costs accumulate across decades), causal opacity asymmetry (proximate political cause is visible, distal policy failure is obscured by complexity and contestable attribution), and elimination mechanism asymmetry (failed replicators rarely face extinction pressure — Swedish rent control has survived 80 years of documented dysfunction). These three asymmetries mean information and good intentions don't fix the mechanism. Legislators who know rent control destroys supply still pass it; voters who know politicians are self-interested still elect them. The selection pressure operates independently of what individuals know.
The properties that make a replicator transmissible: low cognitive load, high emotional arousal, built-in immunization against critique, identity fusion, and orthogonality to truth. "Everyone should be equal" has built-in majority appeal — most people are below the top. "Sacrifice the present for the future" has built-in minority appeal — most people prefer now to later. The ecology selects for what spreads, not what works.
In this respect, "vote on values" can amplify the most infectious package — which identifies a mechanism that requires countervailing institutions able to distinguish host-supporting from host-consuming effects.
III. The Wrong Unit
There is a deeper problem than transmissibility bias. Democratic preference aggregation takes the individual as the unit of measurement. Each person gets one vote. The aggregate is a sum of individual preferences.
Civilizations have system-level persistence requirements, capital stocks, and failure modes. Constituent rights, agency, welfare, and distribution remain explicit constraints and sites of political choice.
The classic illustration: Desvousges et al. (1993) found that people pay roughly the same (~$80) to save 2,000, 20,000, or 200,000 birds from oil ponds. This is cited as "scope insensitivity" — a cognitive bug. But the deeper problem is that the question itself imports the wrong unit. Bird populations are self-renewing; individual bird deaths don't matter for the ecosystem's sustained complexity unless they threaten population viability. "How many individual units of bad?" is already the wrong question. The human response — "roughly the same regardless of magnitude" — may be tracking the right variable: "is there a systemic problem?" rather than "how many individual units?"
This is the unit-of-concern problem in miniature. The bird study reveals what happens when you try to aggregate individual-level valuations for system-level phenomena. The individual can't produce a meaningful scalar response because the phenomenon isn't scalar at the individual level. "How bad is this for the ecosystem?" is a categorical question — healthy, stressed, threatened, collapsing — not a quantity you can price.
The same structural incapacity applies when you ask individuals to "vote on" civilizational strategy. Civilizational survival, institutional health, intergenerational capital maintenance — these are systemic phenomena that don't decompose into individual preference units. Asking individuals to vote on civilizational strategy is like asking cells to vote on organ function — the wrong level of analysis, producing outputs that are meaningless at the level where the decision actually operates.
This essay uses civilization as one policy-scale unit of analysis, alongside constituent rights, agency, welfare, and distribution.
IV. The Separation Is the Error
Return to futarchy's core move: separate values from beliefs. Vote on what you want, bet on how to get it.
But are values and beliefs separable?
A maximally flourishing civilization IS one that maximally expresses its deepest commitments. A dying civilization ISN'T expressing its commitments regardless of what it claims. "Values" aren't an input to the optimization — they're either identical to flourishing or an output of the flourishing state.
If value claims are typed and host/horizon are bound under a chosen persistence commitment, then "vote on values" raises an operational-boundary question:
- Redundant: everyone votes for flourishing, which is what you should optimize for anyway
- Destructive: people vote for something OTHER than flourishing, and you efficiently optimize for the wrong thing
The hedge: instrumental preferences ARE subjective and vary across civilizations. Language, aesthetics, cultural forms, which side of the road to drive on. A civilization with maximum survival margin has more room to express these. So "vote on values" makes sense at the preference layer — but that's "vote on which flavor of ice cream the government cafeteria serves," not "vote on the objective function of civilization." A much smaller role than the standard model envisions.
"Vote on values, bet on beliefs" embeds the assumption that values are subjective — which is exactly what explicit reference-objective governance rejects. The six claim types — preference reports, causal effects, coordination performance, host-relative viability, standing, explicit commitments — show that futarchy treats all of them as vote-on-able when only preference reports are genuinely subjective. Coordination and viability claims are evaluable by performance once host and horizon are bound.
But the problem is worse than miscategorization. Voting measures along a dimension — replicator fitness — that tracks transmissibility rather than function across the six claim types. The most transmissible preference isn't the best preference. The most transmissible coordination technology isn't the most effective one. The most transmissible viability claim isn't the most host-compatible one. Replicator fitness cross-cuts function: "equality" contains both standing assignments (equal dignity — potentially load-bearing for large-polity stability) and parasitic components (equal outcome — demonstrably substrate-degrading). Voting can't distinguish these because both propagate through the same hosts. It measures which package has the highest viral load, regardless of which components within the package are functional and which are parasitic.
V. The Viral Load Problem
The specific pathology becomes visible when you combine the replicator analysis with democratic mechanics.
Consider which value-replicators win democratic elections:
| Replicator property | Electoral fitness | Civilizational fitness |
|---|---|---|
| Low cognitive load | High — easy to transmit, easy to vote for | Low — simple models of complex systems fail |
| High emotional arousal | High — motivates voting, sharing, organizing | Variable — can motivate or paralyze |
| Identity fusion | High — tribal loyalty drives turnout | Low — prevents updating when wrong |
| Present-optimization | High — benefits now, costs later | Catastrophic — temporal parasitism |
| Majority appeal | High — by definition | Irrelevant — physics doesn't vote |
Electoral fitness and civilizational effects diverge when the values that travel well are not the values that keep the host viable.
This is why democratic consensus drifts toward short-termism. Short-term-reward memes outcompete long-term-investment memes in transmissibility. "Benefits now" wins against "sacrifice now for your grandchildren" as the dominant trajectory in every democracy. The egalitarian ratchet is one instance of this general mechanism: each stage of egalitarian escalation has majority appeal (most people are below the top at each stage), so the ratchet turns in one direction regardless of downstream consequences.
Voting aggregates transmissibility-weighted support during a campaign. It does not by itself measure long-horizon wisdom.
VI. The Containment Response
The standard response to this analysis is itself a replicator move: absorb the critique's language without changing the hard core.
"Yes, voters are imperfect. That's why we need better institutions, better education, better information." This response treats the problem as a calibration error — fix the inputs and the mechanism works. But the mechanism doesn't have a calibration problem. It has a unit of analysis problem. No improvement in individual voter quality fixes the structural incapacity of individual preference aggregation to value civilizational-scale goods. Better-informed cells still can't vote on organ function.
"Deliberative democracy solves this." Deliberation can improve individual belief quality and may alter transmission dynamics. The residual question is whether it supplies an operational boundary between current support and long-horizon system effects.
"Constitutional constraints solve this." Closer. Constitutional constraints remove certain domains from the voting mechanism — you can't vote to abolish free speech, can't vote to establish a state religion. This is the correct structural move: recognize that some things should not be determined by preference aggregation. But existing constitutions constrain process (how decisions are made), not telos (what the system optimizes for). The constitutional constraint says "you must follow these procedures." It doesn't say "the procedures must produce civilizational flourishing." The missing layer.
The futarchy response is more specific and worth addressing directly. Hanson does not claim that voting produces wise values — he is explicit about this. His 2026 position grounds values naturalistically: what people identify as core values are shallow mental structures shaped by biological and cultural evolution; adaptiveness itself is the deepest value. He further acknowledges that selection mechanisms can break — global monoculture has disrupted cultural evolution by eliminating competition between distinct cultural experiments, producing drift toward maladaptive norms.
His proposed fix: tie the "vote on values" half to stable long-term metrics (aggregate population over centuries, total energy usage, "as early as possible a date for our descendants to reach the stars") rather than current shifting preferences. He also acknowledges, however, that futarchy itself fails when "price estimates seem to conflict with deep shared morals, they are denounced" — the markets get overridden, not followed.
His ultimate solution therefore is elite persuasion: "cultural elites to look much more favorably on 'Social Darwinist' values." He concedes that "technocratic solutions cannot substitute for shifted cultural values."
This is a significant partial convergence — Hanson agrees current preference aggregation cannot adequately specify the values half, agrees selection mechanisms can break, and agrees mechanism design requires prior value alignment to work. Three residual problems remain.
First, "adaptation" as a value criterion is circular — what persisted was adaptive, by definition. You cannot engineer toward a tautology. Sustained syntropy is the falsifiable, predictive version: a measurable property of a system's current state that predicts persistence before the selection event, not a label applied after it.
Second, his proposed sacred metrics are proxies, not the underlying principle: aggregate population can be maximized at the expense of complexity; "reaching the stars" is a specific goal, not a general criterion. The metrics he wants to enshrine must themselves be evaluated against something — and that something is what the physics-grounded telos provides.
Third, elite persuasion is itself a replicator contest. The values that win elite buy-in are the most transmissible among elites, not necessarily the most physics-appropriate. Hanson has correctly diagnosed broken selection at two levels (monoculture, current preferences) while his proposed solution — persuade elites to value the right proxies — operates at a third level subject to the same mechanism. The non-arbitrary exit is to derive the criterion from thermodynamics rather than select it by persuasion: sustained generation of organized complexity over the longest time horizon is what distinguishes configurations that persist from those that don't, not as a preference asserted by someone, but as an engineering constraint discovered from first principles.
VII. What Would Work
If "vote on values" produces replicator fitness contests, what produces civilizational alignment?
Constraints from physics, not preferences from polls.
The AI alignment community already has this intuition. You don't align a powerful optimizer by asking it what it wants. You build architectural constraints that make certain behaviors structurally stable and others structurally expensive. The reward signal is designed, not discovered by asking the AI's "preferences."
Applied to governance:
1. Make the reference objective explicit, then contestable. Physics can constrain which configurations persist under specified conditions; it does not eliminate residual political choice. A countervailing institution requires declared commitments, limits, challenge procedures, and political closure.
2. Model mechanisms, measure outcomes. A Fourth Branch that models causal chains, stress-tests incentive structures, and audits whether institutions produce their stated outcomes — not whether citizens are satisfied. Satisfaction is a replicator-mediated signal. Outcomes are physics.
3. Hard constraints on identified failure modes. Switzerland's debt brake is a prototype: a constitutional rule, installed once by referendum, that fires automatically when the system drifts from sustainability — removing future fiscal decisions from continuous preference aggregation. Extend this to demographics, infrastructure maintenance, institutional health. Don't ask whether people want fiscal sustainability. Make fiscal sustainability a structural constraint like gravity.
4. Reserve voting for what voting can handle. Instrumental preferences — where to build the park, which cultural programs to fund, how to allocate the discretionary budget. Things where individual preference is genuinely the relevant input and where the wrong answer doesn't produce civilizational failure. "Vote on ice cream flavors" is a coherent procedure. "Vote on the laws of thermodynamics" is not.
This treats democracy as a set of tools with distinct functions and limits. The question is whether institutions can identify and correct any divergence between current transmissibility and declared long-horizon objectives without displacing rights, agency, or political closure.
VIII. The Meta-Problem
One last structural observation. This essay's argument cannot propagate through the mechanism it critiques.
"Don't vote on values" will never win a vote on values. It pattern-matches to "elitism," "technocracy," "you think you're better than the people." These are antibody responses — kill patterns that fire before evaluation reaches the reasoning layer. The egalitarian replicator's immune system is specifically optimized to neutralize exactly this kind of analysis.
This is not a reason the analysis is wrong. It is a prediction the analysis makes about itself. If "vote on values" is a replicator fitness contest, then the proposition "stop voting on values" should lose every fitness contest — because it reduces the replicator's influence. And it does.
The implication: the path from here to there does not run through democratic persuasion. It runs through crisis, competitive pressure, parallel institutions, or hard constraints installed during constitutional moments. The same paths that every structural reform in history has required, for exactly the same reason: structural capture cannot be reformed through the process it captures.
Your civilization already has a telos. It's the one installed by the fittest replicator in the current ecology. The question is not whether to have a telos, but whether to have one derived from physics or one derived from a popularity contest among viruses.
The underlying physics: The Question Nobody Asks. The replicator dynamics: Values Are Replicators.
Related reading:
- Values Are Replicators — The testable class claim: observable replicator dynamics in some value complexes, with benign, symbiotic, and parasitic cases
- Values Aren't Subjective — Six claim types behind "values" and why conflating them enables evasion
- Telocracy — Governance with an explicit public reference objective, not polls alone
- The Governance Alignment Problem — Why the job of "politician" is structurally misaligned
- The Containment Pattern — How "we already account for that" neutralizes this critique without paradigm change
Sources and Notes
Lerer (2025), "Blind Breeders: The Extended Phenotype of Unnatural Selection in Legislative Systems" (SSRN): The closest empirical treatment of the same mechanism at the legislative level. Lerer documents six case studies (Argentina's rental law, Swedish rent control, Berlin's Mietendeckel, US mandatory minimums, Three Strikes, Prohibition) showing that legislation consistently optimizes for political fitness rather than policy effectiveness. The three structural asymmetries (temporal feedback, causal opacity, elimination mechanism) provide the institutional-level explanation for why the transmissibility-adaptiveness gap is systematic. Lerer's framework is downstream of this essay's: it documents what happens at the legislative level but doesn't address the value-specification problem at the civilizational level or provide the physics-grounded alternative.
Futarchy and cultural drift: Robin Hanson, "Shall We Vote on Values, But Bet on Beliefs?" (2000/2013) — the foundational proposal. His more recent posts directly relevant to Section VI: "Your Deepest Value Is Adaption"; "How to Fix Cultural Drift"; "Culture as a Cruel Master"; "We Need Elites to Value Adaption".
These posts show Hanson acknowledging that current preference aggregation drifts toward maladaptive outcomes, that futarchy requires prior elite alignment to work (markets get overridden when they conflict with "deep shared morals"), and that his ultimate solution is elite persuasion rather than mechanism design — a concession that technocratic solutions alone are insufficient. Section VI addresses what remains: elite persuasion is itself a replicator contest, and his proposed sacred metrics are proxies that still require a first-principles criterion to evaluate them against.
Scope insensitivity: Desvousges et al. (1993), "Measuring Nonuse Damages Using Contingent Valuation." The bird study. Kahneman extended the analysis: "scope insensitivity" reflects not a calibration error but a structural feature of how human valuation works — you value the prototype (a bird dying in oil), not the quantity. This is correctly calibrated for individual-scale action. It is structurally incapable of producing civilizational-scale valuation.
Blackmore (1999), The Meme Machine: Extended Dawkins's meme concept to imitation and voting behavior — arguing that the decision to vote is often memetically determined rather than deliberate, and that political campaigns function as replicator fitness contests exploiting cognitive biases. The specific claim here (that the aggregate mechanism structurally measures transmissibility, orthogonal to civilizational fitness) goes beyond this application, as does the futarchy critique, the unit-of-concern argument, and the physics-grounded alternative.
Replicator dynamics applied to values: The framework extends Dawkins's original meme concept (1976) with two additions: (1) the replicator-level properties that distinguish values from preferences (transmissibility pressure, identity fusion, social enforcement, immune system, self-model construction), and (2) the physics-based evaluation criterion (substrate persistence, not transmissibility) that enables ranking replicators objectively.
Vosoughi et al. (Science, 2018): "The spread of true and false news online." Falsehoods spread faster, deeper, and broader than truth on social media. This is what the replicator framework predicts: truth-tracking is not a fitness advantage in the current information ecology. The properties that maximize spread (novelty, emotional arousal, identity-validation) are orthogonal to truth.
Caplan, The Myth of the Rational Voter (2007): Voters hold systematically biased beliefs about economics. They are not randomly wrong but predictably wrong in directions that favor present consumption, visible benefits, and anti-foreign bias. This is consistent with replicator-mediated belief formation: the beliefs that colonize most voters are the ones with highest transmissibility in the current ecology, not the ones that model reality most accurately.
NECSI Sum Rule (conservation of complexity): Yaneer Bar-Yam's formalization showing that large-scale coherence physically requires suppressed local variance. Applied here: individual preference aggregation at civilization scale requires either (a) individuals whose preferences track civilizational needs (implausible given scope insensitivity) or (b) an aggregation mechanism that transforms individual-scale inputs into civilization-scale outputs (no such mechanism exists in current democratic architecture).
The "nobody takes civilization as terminal unit" observation surveys: Peter Turchin (Cliodynamics — descriptive, no engineering), Oswald Spengler and Arnold Toynbee (descriptive, no mechanism), Singapore (practiced, never theorized), Daniel Schmachtenberger / Game B (closest in spirit, vague on mechanism), Nick Bostrom / longtermism (civilization instrumentally valuable for individual welfare, not terminally valuable as telic system), Balaji Srinivasan / Network State (libertarian individualist ontology underneath).