The rule, exactly. Before an election each citizen chooses: to vote — or to take a payment and not vote in that election. The payment is a percentage of the median income, set by referendum; there is no fixed sum. One ballot is one vote, with no multiplier in the count; the vote is amplified only in that each ballot's share grows as others step out. The budget pays under law; a candidate never pays. Only a referendum of all citizens — simple majority, no quorum — introduces, changes or repeals the rule.
The protocol has been introduced nowhere and no pilot has been run: shares, turnout and outcome figures in the chapters are estimates, and the protocol promises nobody an election victory. If a chapter says otherwise, Exact Answers and the Charter are correct. For a candidate: ten questions and ten steps. For a citizen, a mayor, a finance officer, a donor, a journalist, a scholar, a lawyer: answers by role. Everything in force in one file: llms-full.txt.
19e. The Husband Penalty: How a Means-Tested Payment Produces a Dependent Voter¶
Chapter: 04 — Electoral dynamics (paired with §19b) File: 04_019e · v1 · 21 September 2026 (Claude dialogue, session 21-09-26) · §5b added 23 September 2026 (Claude dialogue, session 23-09-26): yin-yang at the scale of the family, an answer to 039b.3.3 Source: the architect's question "why does a wife vote differently from her husband" and his own answer: "Masha has no husband; the state is her husband, and without a husband she always votes for the state"; the analysis of the document "Alimony and AB-EXIT" (May 2026); the refinement about targeting and means-testing. Complements 019b (the split of demand), 019c (reform by subtraction), 018 §31 (pressure and verifiability).
How to read this chapter (note of 02.10.2026). The text contains wording that is easy to misread: sums and formulas with coefficients are worked examples: the size of the payment is set as a percentage of the median income approved by referendum. The exact-answers sheet 1d and the charter 048m are in force.
19e.1. The observation: the vote goes to the source of the guarantee¶
The architect's formulation: a married woman's vote goes with her husband's; an unmarried woman has no husband — and his place is taken by whoever the guarantee comes from.
This is not speculation but a measured quantity. In American data the gap by marital status is larger than the gap by sex: before 1980 there is practically no gender gap in voting — the household votes as a household; the gap appears together with the fall in the share of the married. Edlund and Pande (Quarterly Journal of Economics, 2002) showed the mechanism directly: women shifted politically not because they changed their convictions but because they ceased to be married; divorce worsens their position relative to men, and the demand for redistribution rises.
The dating on post-Soviet material coincides. While the village lived by subsistence farming, the bond was economic: children — cow — haymaking — a man, and the household voted as a unit. The shop, the pension and the benefit made the cow optional; with it, the one who mows ceased to be necessary. In the Russian village this is 1995–2005 — one generation ago.
A refinement without which the thesis breaks on counter-examples: the vote goes not to "the authorities" as such but to the source of the guarantee. Where the guarantee comes from the apparatus — a benefit, a subsidy, a nursery place, a state salary — that is the incumbent. Where the opposition promises the guarantee, unmarried women vote for the opposition: in the US they have voted for the same side for decades regardless of whether it holds power. One mechanism, different addresses — and in that form the thesis is falsifiable.
19e.2. What follows for whoever wants power¶
If the vote belongs to the source of the guarantee, the authorities acquire a structural interest: to be that source for as many people as possible, and to allow them no alternative source.
| What benefits the source of the guarantee | What is observable |
|---|---|
| Payment at discretion, not by right | means-tested benefits are granted and withdrawn by an agency's decision |
| A second income in the house reduces the payment | the base of calculation is household income: bring in a working man, lose the benefit |
| Many points of contact | benefit, subsidy, nursery, school meals, after-school care, social worker — each a separate reason to depend |
| Weakening of the alternative breadwinner | recovery from turnover rather than profit; criminal liability for debt (see 19e.4) |
| Invisibility of the payments | many small ones through different windows — not one sum that is remembered |
No conspiracy is needed for this, and the book does not assert one; but "no intent" does not follow either — at the level of the instrument the intent is documented (19e.7). What operates is policy selection: measures that produce loyal dependants are extended and expanded; measures that produce independent people are quietly left unfunded. As a pharmaceutical company does not create a disease but has no incentive to cure it for good.
19e.3. The precise formulation: what breaks is not the state but means-testing¶
The claim "the state destroys the family" is wrong and shatters on counter-examples. Germany directly subsidises marriage through joint taxation (Ehegattensplitting); France operates a family quotient; Hungary and Poland pay for children and for marriage demonstratively. There the state strengthens the family.
The working formulation is narrower:
It is not the payment that punishes partnership but the targeting. A universal child payment does not depend on who lives in the house. A payment calculated on household income always penalises: a second income appears — the benefit is cut.
This is the benefit cliff: on crossing the income threshold the family loses more than it gained. Economically it is a marginal withdrawal rate, in places above 100 %; socially it is a direct price for living together and admitting it officially.
The consequence for testing: where the payment is universal there is no "husband penalty"; where it is means-tested, the penalty is measured in percentages of lost household income. This is computed from any country's regulations without any sociology at all.
19e.4. Symmetry: both sides are captured¶
Divorce produces two dependent voters by different mechanisms.
She — through the cheque: benefit, subsidy, state salary. To vote against the incumbent is to risk the source from which the children eat. This is not apathy and not stupidity; it is a mother's dharma applied to the only remaining protector.
He — through recovery: the obligation is computed from income rather than profit (an enterprise is taxed on profit, a person on his whole turnover); non-payment brings criminal liability, and the decision is made by a judge appointed by that same vertical. Fear does not require an actual phone call — the possibility of one suffices.
Hence a consequence stronger than either half alone: one procedure of family breakdown produces two controllable voters, and each of them believes the problem is the former spouse.
For the children the loop closes in the second generation: a child whose school, meals, after-school care and nursery come from the state grows up with the model "the state decides" and votes for whoever gives — not from stupidity but from experience.
19e.5. What the protocol does¶
The dividend is built exactly opposite to a means-tested benefit on all four parameters:
| Parameter | Means-tested benefit | Dividend |
|---|---|---|
| To whom | to the household, on a means test | to the person |
| How computed | on family income | by the formula D = M × 1.5 × K |
| Who decides | an agency, at discretion | nobody: there is no discretion |
| Can it be withheld | yes, and that is the lever | no |
| What happens when a husband appears | the payment is cut | nothing changes |
Three conclusions follow.
First: the penalty on partnership disappears. The dividend does not ask who lives in the house, so living together stops costing money.
Second: the guarantee stops being anyone's favour. While a payment can be withheld, the vote belongs to whoever signs. When it can be neither withheld nor cancelled, the vote returns to the person — the same woman, with the same income from the state, but in an entirely different political construction.
Third — and this one is electoral. If a person truly has no stake, she takes the dividend, and her vote, bought today for free, leaves the electorate voluntarily, with money in the house and without grievance. If she refuses, then a stake exists and it is worth more than money, and such a vote is no longer fear but choice. The mechanism takes no rights away and requires nobody to become a different person.
19e.5b. Yin-yang: the balance of family life¶
Added 23 September 2026. The architect's formulation came as a heading without elaboration: "Yin-Yang. The balance of family life." The elaboration below is the assistant's; it is reconciled with what the repository already records about yin-yang (015b.2, 015b.6d, 034b.5) and about the family (19e.1–19e.5, 039b.3.3, 048h.4), and where it goes beyond the formulation, that is said.
The protocol does not see the family. It has two buttons per citizen: the dividend is personal (19e.5), credited to the recipient's account (042), the status private by default (048i.5b). The family sees something else — one budget and one kitchen. So the two states of 015b.2, yin on the money and yang with the vote, can for the first time live in one house at the same time: one brings the sum, the other carries the vote. This is not a compromise between spouses and not a loophole, but the same yin-yang described for the country, only at the smallest scale, where it can be seen without a model: at one table sit the contented and the aggrieved (015c.7a), and at that same table they decide afresh each cycle which of them is which.
The arithmetic. A family of two adults. Today it has two votes of 1/N each and no money. After the protocol — three states.
| Family state | Vote in the election of managers | Money |
|---|---|---|
| Both vote (A, A) | two votes of 1/(N − M) | 0 |
| One votes, the other takes (A, B) | one vote of 1/(N − M) | D |
| Both take (B, B) | 0 in this cycle; referendums on the rules — like everyone (048g.6b) | 2D |
M is the number who exited. One remaining vote weighs as much as two of today's as soon as half have left (1/(N − N/2) = 2/N), and as much as three at two thirds. So at an exit of half or more the family (A, B) loses not a single share of influence against today and gets D on top. The question "money or vote", which 039b.3.3 puts before the family as tragic, does not face the family: it has both, and the only argument is about who carries what this cycle. That is an argument of the same kind as "who takes the second shift", and the protocol does not intervene in it — just as the tax code does not intervene in who in the house manages the wages. The formula, incidentally, counts by household and pays by person: the coefficient 1.5 converts the personal median into a household median (002 §2.5), but each adult receives his own D.
Rotation. States, not types (015c.7a): in a quiet cycle both go for the money; things get worse — one comes back; if things are bad — both, with heavier votes (015b.6b). The family is the smallest thermostat. The one that trips once a cycle at the level of the country is checked every day in the kitchen, and who in that house is the aggrieved one is known there sooner than to any sociologist.
The answer to "kitchen terrorism" (039b.3.3). Three cases are mixed there, and their answers differ.
The spouses argue: one needs money for debts, the other wants the vote. Answered above: the protocol does not make them choose. Each presses her own button, and both D and a heavier vote come into the house. The media "victim" in this story is a family that received money and lost no influence.
A tyrant makes his wife press B, takes the cash and votes himself. Compare with today, not with an ideal (037b.4).
| Today | After the protocol | |
|---|---|---|
| Her ballot | Family voting: he votes for her or beside her. OSCE observers saw it in 10–25 % of precincts in 1998–2005, in Albania in 2025 in 11 %, in Moldova in 2025 in 1 % of observations (048h.4) | If she is in B, he has no ballot of hers at all: the second vote that today comes to him free disappears. If he keeps her in A and votes beside her, that is today's family voting, but now it costs the family D per cycle (the price of coercion, 056d.4) |
| His weight | Two votes out of N: his and hers | One vote out of N − M, like any remaining voter. Her exit adds nothing to his weight: 1/(N − M) is set by millions, not by his house |
| Her money | The benefit is computed on household income (19e.2); he takes her wages just as he would take D | D is credited to her account (042). Taking it is the same as taking wages: the same offence, the same court (19e.4), but with a bank record that seized cash never had |
| Her vote on the rules | — | The referendum on the protocol itself stays with her under any status (048g.6b) |
The sum of the right-hand column: the tyrant gains nothing he did not have today, and loses the free second vote — or pays D for it. The one thing the protocol leaves him is the possibility of taking her D, as today he takes her wages; that is the residue from 039b.3.3, and it has not gone anywhere: the protocol is not a family code. But "destroy the protocol's reputation in an evening" with this story is not possible: the story "husband took his wife's money" exists today and is not news. The news is that for the first time she has a statement.
A tyrant makes her vote as he says. This is family voting pure and simple; the protocol does not solve it, and the secrecy of the booth against pressure from one's own works as it does today (018 §31.3, 048h.4). One thing changes: it stops being free.
What this means for the chapter. 19e.1–19e.4 describe how the state took the husband's place: the guarantee comes from the apparatus, a second income in the house cuts it, one procedure of breakdown produces two controllable voters. 19e.5 — that the dividend is built the other way round. This section — what follows for a family that has remained a family: the balance returns inside the house. Who brings the money, who carries the vote and when they change places is decided by two people at a table, and neither the agency nor the protocol takes part in that decision. "The balance of family life" is not a metaphor for yin-yang but its original scale: before a system of millions begins to swing between states as in 015b.6d, every house swings between them.
19e.6. Weak point of the section¶
First: causation is not proven. Edlund and Pande establish a link between the fall in marriage rates and a political shift, but the direction may run the other way — political views affect the probability of marriage. The question is open in the literature.
Second: "the authorities are interested in the breakdown of the family" is an argument about incentives, not about intention, and it cannot be tested directly. Only the consequence can be tested: does a benefit cliff exist and how large is it. That is what should be computed, rather than imputing intent.
Third: the size of the effect is not estimated. How many votes in a given country fall to the "husband penalty" is unknown; an indicator for a pilot: the share of means-tested benefit recipients among those who stay in button A compared with their share among today's voters.
Fourth: the section touches family law, where every measure has a lawful aim — protecting the child, reducing poverty. The argument here is not that the aims are false but that targeting has an electoral side effect that nobody measures.
Fifth, on 19e.5b. "The family loses no influence" holds at an exit of half or more; in a cycle with a small exit — phase 3 in 015b.6 gives 15–20 % B — one remaining vote weighs less than two of today's, and the trade "half a vote for D" becomes real. That is the protocol's normal price, not a flaw, but "no losses" can be promised only for the first cycles. The family-voting figures come from OSCE observations; by rule 23b.1 they transfer the existence of the phenomenon and its direction — there is less of it in the target countries: Moldova 2025 gives 1 % against 11 % in Albania — but not the magnitude for a particular house, and for the EU and US the measured pressure is the employer, not the family (048h.4). Finally, the reading of the formulation "Yin-Yang. The balance of family life" belongs to the assistant: the architect gave the heading, not the text. 🟡
19e.7. The composition of the guarantee electorate: figures and the level of intent¶
The architect's question: "find in surveys and statistics the distribution of the regime's electorate in Russia and similar autocracies — man, woman, single, married, age. Does it confirm the supposition that it is advantageous for an autocrat to become the breadwinner, husband and father for the majority of the electorate, and to suppress men as competitors as far as possible?" And the refinement: "it is important that the present authorities consciously stimulate their electorate and multiply it."
Sections 19e.1–19e.5 gave the mechanism without figures. Here are the figures, and separately the question of intent, because it is decided differently at different levels.
Composition on three axes¶
Survey levels are upper bounds (23b.3). The direction of the gaps between groups is most likely right; the size is unknown, and women and the elderly may be more cautious respondents, which inflates the gap. This is said once and applies to the whole table.
| Axis | Data | Source and type |
|---|---|---|
| Sex | presidential approval: women 81–83 %; men — disapproval 21 %. Readiness to vote at the Duma elections: women 55 %; refusal — men 32 %. United Russia among women 34 % | Levada, April–June 2026 [survey] |
| Age | readiness: 55+ — 59 %; refusal — under 25, 34 %. UR among 55+ 33 %; KPRF among 55+ 14 %; LDPR and New People among the young 14 % each. Highest disapproval — ages 40–54, 22 % | Levada, June–July 2026 [survey] |
| Demography | 1,152 women per 1,000 men; under 30 men are more numerous, from 30 the proportion reverses; 80+: women 75 %. Life expectancy 78.7 against 68.0 — a gap of 10.7 years | Rosstat [administrative data] |
| Behavioural | the three layers of the herded queue — pensioner, state employee, network voter (056c) — are demographically female: teachers, doctors, social workers, municipal employees. Coercion passes through the family: one state-employed woman yields several votes (056f.3c) | administrative composition of the sectors; analysis of 24.09.2026 [observation] |
The composition in sum: the guarantee electorate is female and elderly, and in its compelled part, state-employed. This confirms 19e.1.
But the key point is where that composition comes from. It is what it is not because of the war. The 10.7-year gap in life expectancy made Russian old age a widow's old age a generation before 2022. The 15 % female surplus was produced by peacetime male mortality.
"Breadwinner" is a term of law, not a metaphor¶
The architect's formula "breadwinner, husband and father" is more exact than it sounds: the survivor's pension — in Russian, literally "the pension for the loss of the breadwinner" — is a legal term, and under it the state has for decades been the provider of the surviving spouse in the letter of the law. 19e.1 describes the mechanism; here it can be seen that the state named it itself.
The "single / married" axis — a hole that must be named¶
The mechanism of 19e.1 rests on American data (Edlund and Pande). The assistant found no Russian measurements of support by marital status; Rosstat gives composition by sex and age but not the link "widow → vote". This axis remains derived rather than measured, and it must not be filled by supposition.
"Consciously" — three levels, and at which it is provable¶
The architect's refinement requires a correction to 19e.2. It said: "no conspiracy is needed — policy selection operates." That was written against the conspiracy objection, but it reads as "there is no intent". That is inexact: intent is documented — at a certain level.
Level 1 — the instrument. Proven. On 22 August 2021, at the United Russia congress, a month before the Duma elections, the president announced one-off payments: 10,000 roubles to every pensioner and about 15,000 to every serviceman; the decrees were signed within days and payments began on 2 September [verified]. Two groups, exactly those in the table above; the moment — pre-electoral; the actor — the head of state himself. To this add the tracking of state employees' turnout by name in 2026 (056f.3c) and the means-tested payments in the table of 19e.2. At this level "consciously" is not an interpretation but a chronology.
Level 2 — multiplication. Documented as a direction. The guarantee electorate is not only stimulated but expanded: the state sector, the number of recipients of means-tested payments, the family channel by which one dependant brings the household. Intent here is inferred from persistence: measures that produce dependants are extended, those that produce independent people go unfunded. That is the "policy selection" of 19e.2, with a correction: the selection is run by people who know what they are selecting for.
Level 3 — the war as electoral strategy. Not supported, and the arithmetic runs against it.
- Scale. 352,000 dead (the Meduza and Mediazona estimate from the inheritance register, end of 2025) and 261,000 who left within days of the 2022 mobilisation — against an electorate of about 110 million. The war removes of the order of 0.5–1 %, whereas peacetime mortality had already produced a 15 % female surplus. The electorate was reshaped before the war and without it.
- The wrong men. By 056e.1, those who go to fight for money are men of 35–55 from depressed towns, 88–90 % by economic calculation — the most buyable demographic, the material of the first flow, not competitors. The competitors — urban, mobile — left, and the regime let them out: the border was not closed.
- Young men are not the opposition on the data. Under 24, approval is 83 % (an upper bound, but the direction is the same); the young's refusal to vote is motivated by "nothing will change" (41 %) — futility, not a position.
The defensible version is stronger than the one formulated. The regime does not destroy competitors — it converts the least dependent demographic into the most dependent one. The most dissatisfied male cohort (40–54, highest disapproval) goes to the front and returns either as a paid man or as a widow on a survivor's pension; both depend on the state more than the original person did. 056e describes exactly this: "two legal ways to obtain a large sum — a contract, or death with compensation to the family."
And locally it is not 1 %: the dead are concentrated in the same small towns that deliver the 99 % precincts (056f.1). The effect is concentrated where the machine needs it most.
The level of intent here is "observes and does not correct". Does the regime know of the conversion — it does, it is visible in its own statistics; does it connect it to the electorate — that is unknowable from outside, and the conclusion does not need it: the consequence is produced and is not corrected.
| Level | What is claimed | Status |
|---|---|---|
| instrument | payments, tracking, targeting — timed to elections, aimed at target groups | proven by chronology and by the regime's own words |
| multiplication | expansion of the dependent class and the family channel | documented as a direction; intent from persistence |
| war as strategy | men are removed for the sake of the electorate's composition | not supported; the defensible replacement is the conversion of dependence, "observes and does not correct" |
Similar autocracies — honestly thin¶
In Western democracies the "traditional" gender gap (women more conservative and more loyal to the incumbent) reversed in the 1990s–2000s; Russia retains it. Norris and Inglehart (2019) link the ageing of the electorate to conservative leaders. The assistant found no figures by sex for Belarus, Turkey or Hungary and substitutes none. Belarus 2020 is a counter-signal: the protest was led by women. So "women are the regime's base" is not a law of autocracies but a Russian regularity with a specific demographic cause — and it cannot be carried to other regimes without their own figures.
What this changes for the protocol¶
Three things, all three already built into 19e.5.
First. The dividend does not depend on marital status and is not computed on household income — it removes the lever itself, rather than softening it.
Second. It competes with the survivor's pension on the pension's own ground — as a right rather than a favour — but pays the share to the living husband as well. What makes a widow dependent is, for the protocol, no different from what makes anyone else dependent.
Third. The family multiplier of coercion (056f.3c) changes sign: a mother with a lawful ground not to come has no need to beg the family.
Weak point. The survey figures are upper bounds, and that is said above once; but separately: the gap between the sexes may be partly an artefact of who answers "I approve" more readily. The marital-status axis for Russia is not measured at all. The arithmetic of the war is national; the local concentration is asserted from the geography of losses, but the share removed in particular territories was not computed by the assistant. Level 3 of intent is unprovable in principle — only the consequence is visible from outside. The comparison with other autocracies has no figures. And in general: the composition of an electorate is not its behaviour; the repository holds lessons that a reshaped electorate votes otherwise than expected (040d.1), and the table above says who is in the queue, not what the queue will do. 🟡
Related: 019f (the vector: wearing down men, the peacetime loop) · 019b (the split of demand: money by law, the vote for justice) · 019c (reform by subtraction) · 018 §31.3 (one can order but cannot check) · 011b (learned helplessness) · 015c §7a (states, not types) · 033c.4 (publicity of status) · 020 (why this is not UBI) · 015b.2, 015b.6d (yin-yang; three states) · 039b.3.3 (kitchen terrorism) · 048h.4 (family voting measured) · 042 (crediting to the account) · 056d.4 (the price of coercion)