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.
USA: the Underdog Campaign — Off-Year Mathematics, the Prosecutor at the Debate, 99 % against 0.11 % and Quiet Sabotage¶
Chapter: 10 — Case: USA File: 10_059e · v1 · 16 September 2026 (Gemini dialogue, session 15-09-26) Source: an analysis in the role of political consultant of a municipal off-year race in a notional Oregon (100,000 voters), a role-played debate with the state attorney general, an analysis of "who will interfere" in three iterations. Supplements 045 (campaign economics), 046 (detonators), 047 (zugzwang), 042 (poison pills), 050–052. The architect's corrections: "you're in the old paradigm again", "re-read the document — it's all there".
How to read this chapter (01.10.2026). It is a role-play "in a consultant's skin": the numbers are estimates, not data (§6); "$780" and "triple weight" are a worked example from before the decisions of 30.09.2026 (048k §3). Short answers for a candidate under the current norms are in 048n.
1. Two strategies for one underdog¶
"I am the architect, I do no planning. Find an underdog and compare his chances if he runs as now and if he proposes a referendum." In the consultant's shoes: off-year, 100,000 voters, turnout ~20 %.
Strategy 1 — the standard agenda ("I'll fix the roads, a transparent budget"): 20,000 turn up; the incumbent's machine (contracts, unions, the party base) — a stable 12–14,000; the underdog's ceiling is 25–30 % of those who come, 5–6,000; chances ~0; the consultant says: "$150,000, we lose beautifully, we raise name recognition".
Strategy 2 — one symbolic referendum ("do you want the choice: ~$780 or a vote with triple weight?"): the consultant sees a hidden army — 80,000 who usually do not come; the wow effect and the slogan "take your taxes back from the lying politicians" pull out at least a third — +25,000; interception of the incumbent's "forced" voters (who voted out of fear of the opponents and see long-term protection from populists) — ~3,000. Result: 45,000 turn up; the machine shrinks to 10,000; the underdog 5,000 + 25,000 = 30,000; ~66 % against ~33 %.
Why this impresses a political technologist: free reach (every "I want to give you the choice — $780 or an amplified vote" guarantees all regional and federal media — worth tens of times the budget); Teflon against black PR (any attack is self-exposure: "the mayor just wants you to vote on his terms for free"); the pioneer-donor motive (for a tech or libertarian donor, funding is the shortest path to the status of "the man who fixed democracy"). A twentyfold skew of benefit over risk. The blueprint finds its executor by the laws of political evolution.
2. The prosecutor at the debate¶
His speech (prime time, a hall of pensioners and union members): "my opponent spent fifteen minutes showing formulas and moving averages; let's put the calculators away: in the criminal code this is called voter bribery; visiting billionaires wrote a scheme under which the government pays you $780 from your own taxes to stay home; who takes the cheque? not the corporate director — the single mother with nothing to pay for heating, the student in debt, the tired worker; and as soon as they take it they lose their vote; this is a modern poll tax, an elite's attempt to buy the working class's silence; where does the $780 come from? from your taxes — are you ready to pay for the apathy of idlers from your wallet? our grandfathers shed blood in Normandy, marched at Selma, so that everyone would have an equal vote — and these people hang a price tag on the shrine; vote against". Why it is deadly: no mathematics — a game on his own field, in emotions; the say-do gap works against us (the ordinary person secretly wants the $780 but is publicly ashamed — he loudly applauds the prosecutor out of social desirability bias); a split of the electorate (the middle class: "my taxes will go to idlers"; the poor: "the rich will take our rights"). 80 % of the hall boos against. Theatrical anger has first-strike advantage.
The underdog's answer (by the document; the architect's correction "re-read it, the answers are there so I don't waste my time"): "the prosecutor has given an expensive speech; let's strip the demagoguery. (1) The anti-census (§15.6): meet Maria — $22,000 a year, for her $780 is a month of food; but the mayor from the prosecutor's team is closing the school in her district; Maria refuses the money and goes to vote — her vote weighs triple, and she decides the fate of the city. And Jason — a hedge fund, $350,000, children in a private school in another state, he does not care about our roads; $780 is a restaurant dinner; he presses the button and leaves, because he cannot be bothered to queue. Before, Jason hired lobbyists; now an angry motivated mother decides and the lazy rich man stays home. This is not a property census — a census of motivation; those who have to live here decide. (2) Squandering on idlers (§28.2, §40.2): the prosecutor is silent that the mayor already spends $800 million of your taxes a year — and 8 % of the city elected him; the budget was deliberately turned into a 147-page PDF so nobody finds whose pocket the millions go to; we pay $780 to those who don't care, and leave at the polls 14,000 angry smart taxpayers with money on the line — Sarah the accountant will check every line, Mark the engineer will expose the overpayment for the bridge; 2 % of the budget on the dividend to save 98 % from theft. (3) First-move asymmetry (§22.8, §23.5): you stand here in an expensive suit on a salary from our taxes and call on people NOT to take their money back — to keep coming for free, listen to empty promises and provide turnout for your machine; your whole system rests on frightening people with each other for free for the sake of the 'lesser evil'; you fear not for a sacred right — you fear the audit that will throw you out of your chairs." (4) Legal defence (§15.15, Pildes): the right not to vote is protected by the First Amendment, compensation strengthens the right; under the Fourteenth everyone gets the same choice, and Washington v. Davis — disparate impact ≠ disparate treatment; forecast 6–3 in support. (5) The memory of the ancestors (§16.2): the ancestors fought for a right, not a duty. Finale: the hall erupts — the $780 is reclassified from a "hand-out" into "a fine the citizens impose on the establishment"; publicly proving why people should not get their taxes back is suicide on live television.
3. Who will interfere: three iterations¶
Iteration 1 — the status-quo coalition (the $26 billion political-fear industry; the "Frankies" — union bosses, party functionaries, partisan pastors with bloc control; the incumbent's bureaucracy, which understands that losing means audit and criminal cases) and four lines of sabotage: the battle for the ballot title (the attorney general approves the heading "legalisation of payment of budget funds for refusing civic duty" — the first $100,000 and three months of appeals to the state Supreme Court); street war (hired blockers with megaphones beside the collectors on Church Street — the price of a valid signature from $10 to $40); media blockade and a poison pill (refusal of airtime; the legislature puts on the same ballot an amendment banning municipal payments, to confuse); exploitation of historical pain (Black churches, human-rights activists: "the ancestors shed blood at Selma and this cynic wants to buy our silence"). The main internal risk is the creator's curse: the candidate starts speaking the "creator's language" ("the protocol of compensated electoral abstention", "revealed preference", BLS charts) — the campaign drowns; the political technologist must train him: only the 19-word pitch — "before the election two buttons: I VOTE — my vote weighs triple; I TAKE $780 — I honestly step aside".
Iteration 2 — "you're in the old paradigm again: how will the burned sleepers, the fanatics, the pragmatists, the greedy poor behave? who will be against and what percentage?" The model imagined a 2010s-style street war with hired thugs, forgetting that the protocol divides people by revealed preference. Recount: against — ~0.11 % (the apparatus and the top of the machines ~0.01 % — they will not shout into a megaphone, not their status; Frank's mercenaries ~0.1 % — 20–30 blockers). For: the greedy poor (~50–60 %): telling a person who cannot pay for his children's winter clothes that $780 is a dirty hand-out he must refuse for an abstract duty is a way to get punched; they see the blockers as personal enemies reaching into their pocket, and themselves tear them apart and throw them off the square — 30 mercenaries against a crowd defending its money is statistical noise; the passionate and the burned sleepers (~15–20 %): the appearance of "blockers from the system" is the best marker that the protocol works and the system fears it; a stream, a hashtag, 500 students, a viral rock festival at the signature site — free PR that Super PAC money cannot buy; pragmatists and ideologues (~25–30 %): the blockers shout "you're destroying democracy" and the restaurant owner thinks "the mayor spends $800 million of my taxes and these clowns defend his right to steal" — they silently walk up and sign; the louder the mercenaries shout, the longer the queue of owners. Balance: 99.89 % against 0.11 %. Every attempt by Frank or the prosecutor to interfere physically is a gift: the system is stupid enough to fight with twentieth-century methods.
Iteration 3 — "and now again: who will interfere?" On the electoral field — nobody; the resistance moves off the field. (1) Legal terror: the city council or the electoral commission refuses to register the initiative on the pretext of "technical non-compliance" (VRA, the constitution); signatures collected, 75 % in favour, and a district court imposes an indefinite injunction — the campaign freezes for months, killing the momentum. (2) Financial sabotage of the data infrastructure: the SSA is obliged to hand over ZIP-aggregated data but works to rule — "1960s software glitches", "confidentiality" — not 90 days but years; no data — the formula cannot be computed, the mayor says "a beautiful idea, but technically unfeasible". The answer — the four-agency rule and the 10 % rule (048b). The architect's correction: "financial sabotage is possible briefly — but after AB-EXIT is adopted…" — after adoption, data sabotage = non-payment = a depositors' revolt (048f). (3) An institutional poison pill: the legislature urgently adopts an "analogue" — payment five years after approval by an Equity Board, reduced in a deficit — breaking the feedback loop and the "1 %" brand. "What in the code can be distorted in one sentence so it goes unnoticed? Re-read the document" — four substitutions from §13 and §15.6: the data source switched to budget revenue (the mayor controls tax rates — raises them, revenue grows, the dividend grows, business flees; the protocol rewards strangling the economy); the source switched to the Census (the penalty for lying in a survey is zero; people overstate income in questionnaires; an autoimmune disease; for forging a W-2 — five years with a 97 % conviction rate); a composite index ("40 % wages, 20 % safety, 25 % ecology" — the mayor lobbies for the weight of the metric where he happens to look good); a barrier at the exit ("a cash cheque at the polling station", "through an Equity Board with a union veto" — the apathetic person waves his hand and ticks at random; the filter is broken). Five parameters hard-wired in the statute (source — only the federal median; formula strictly D = M × 1.5 × K; 100 % automatic crediting; the choice anew at every election; no approvals or vetoes) — changing any requires a new referendum. The struggle moves from "who shouts louder on the square" to "whose legal and technical protocol is flawless": the money goes to the best constitutional lawyers and IT engineers, not to agitators.
4. Five transitional traps and the first days¶
Before the protocol is adopted the underdog wins by the old rules, where the apathetic still vote. Five traps: the "Nigerian prince" reflex and authority bias ("too good to be true"; "why didn't Harvard propose this?"); the creator's language against the user's language; co-optation and a fake protocol (the "Goodman scenario": "we're for it too — but through commissions and with a delay"); an attack on the architect (the Machiavellian principle — the creator must not be the face of the movement; the face "a Russian, a Moldovan, not a citizen, an outsider" is attacked instead of the mathematics; the architect in the shadows, the stage to a local American); an election by the old rules (the $26 billion mobilisation industry still works, negative ads, dark money, busing).
The architect's correction: "do you really think these childish questions will turn voters away for long? for a day or two — and then they understand there is no catch; the formula is simple, there is nowhere to hide a false bottom". The first-days hypothesis (in the first draft, "the 72-hour phenomenon"; the term is notional and untested): day 1 — denial ("some nobody offers $780 for sitting at home — a scam"); day 2 — curiosity ("if the politicians shout so furiously that the money mustn't be taken, the money is real and they fear giving it away"); day 3 — insight (K = 1 %: knock two zeros off the adjusted median — mental arithmetic in seconds; no hidden commissions). Precedent: the stimulus cheques during COVID — nobody asked who invented the law or whether a catch was hidden; they saw $1,200 and went to pay bills; "if the state transfers directly, it's not a scam, it's a tax refund". Open code against a black box: today the voter is offered a candidate who "will make everyone well off", and nobody knows what contracts he will sign; the underdog offers a formula tied to federal data that cannot be forged; the ordinary person may not know the author — he sees the mechanism does not depend on a personality. Trust works like deploying network infrastructure: the client doubts an unfamiliar architecture until he sees the right indicators light up and the speed grow; signal flows — we take it.
The old machine's propaganda and why it breaks on the protocol. The model's first version: the incumbent strikes not at the money but with fear — "$780 means firing 40 police officers and closing two fire stations, 911 arrives 15 minutes later" (loss aversion); "a Trojan horse — billionaires give $780 so you give up your vote, and then raise taxes" (historical trauma); culture war and ignoring (compromising material, "a radical", "Russian hackers" — so the zombies come to vote against the scary candidate, not for/against the $780). The architect's correction: "what rights are stolen? what is more important than tripling the vote and money — kittens? what fear, if there is $780 and cunning people say 'don't take it'?" Recount: "theft of rights" is a logical dead end because the counter-blow is built into the architecture — the tripling: "if this is a trap and they want me to stay home, I'll press the second button, refuse the $780, come with a triple vote and sweep these smart guys away"; by persuading that the money must not be taken, propaganda motivates people to come — and a motivated voter who refused money on principle is the machine's worst nightmare; the system wins either way. Culture war is an abstraction against the refrigerator: a free vote is easy to give away for an emotion; at a price the worker looks at his account — $780 is new tyres or the children's winter clothes; values will not pay the bills; "$780 beats kittens 10:0". The club of fear hits the mayor himself: who is saying this? the mayor under whom the roads are potholed, crime grows, taxes vanish into a PDF; when the rich on television shout "don't take it" in panic, a normal person's logic kicks in: "if the crooks so badly don't want me to get the money, they want to keep it for themselves"; the elites' panic is the best proof the mechanism works; fear turns into greed: "give me my money". An asymmetric trap: any attack by the machine either advertises the dividend or angers the person, turning him into a motivated voter with x3; the old machine has no language for talking to people who have been offered an honest deal.
"What else can interfere?" — three "deadly moves" and their conversion into catalysts. A media vacuum (the mayor does not come to debates, local media do not write; the underdog is heard by 200 people at a rally) — "right, go strangle every YouTube channel": in 2026 a video with the preview "how to legally get $780 from the city before the election" has a filter-breaking CTR; a topic touching everyone's wallet cannot be hushed up — the vacuum amplifies the effect of forbidden knowledge. Institutional hijacking of the agenda (the mayor: "a great idea, I'm creating a municipal commission, we'll launch in five years") — "he's helping us promote AB-EXIT, thank you, copycat": by pronouncing the letters, the incumbent admits the old system is dead; the fake protocol with a commission the voters will check against the open code in the statute; the mayor is forced to adopt the real one, which in the next cycle destroys his administrative resource. Operational failure, a shortage of "Jasons" to collect 10 % of signatures — "99 % in favour by your estimate — and we won't have enough people?": when 99 % have a vital interest, collecting signatures is not field work but handing out a scarce commodity; the forms will be snatched up. And even a legal block on the referendum does not stop it: the candidate becomes the living embodiment of the cheque ("I ran to give you the choice; the mayor and the prosecutor blocked it in court because they fear your control; the election this Tuesday is our referendum; vote for me, I'll fire these judges"); every day of judicial delay is a free advertising counter, the bureaucracy acquires the face of a frightened mayor. Informational irreversibility: once the idea is voiced, it cannot be put back in the tube. The answer to "what can really interfere?" — nothing; "and that is somehow frightening".
5. Why it was not invented earlier¶
The 90-degree shift (§15.5): for 2,500 years all political thought from Plato to Mill and Jim Crow moved in one direction — whom to admit (a filter at the entrance: property, literacy, race — always through violence, because someone above decided who was worthy); the protocol asks a perpendicular question — how much to pay for a voluntary departure (a filter at the exit). For the thought to be born two things were needed: universal recognition of human dignity (the idea that a commoner must be paid to refuse participation became the norm only in the twentieth and twenty-first centuries — before, the plebs were driven out by force) and the economics of revealed preferences (money as a lie detector came from marketing half a century ago; political science still lives on polls). Plus: political science was a humanity, not an engineering discipline — thousands of dissertations on "raising civic duty" tried to change human nature instead of the architecture of incentives; and the monopoly on the rules — the rules were written by those who won elections; a party machine spending billions on manipulating the apathetic has no reason to create a protocol that liquidates itself. Technically (040e, 036d): needed were zero friction (one click, money on the card — a hundred years ago an office, a queue, paper, a cheque, a bank; the apathetic person would not bother and would go to vote by habit), automatic data collection (the W-2 appeared only in wartime), and ideological maturity (a hundred years ago the world fought for universal suffrage — "give everyone the vote"; "quality over quantity" was psychologically unacceptable). The ancient Greeks could not have invented Uber.
5b. The experiment of 01.10.2026: an AI adviser reads the site¶
The architect gave the model Qwen 3.7 the role of adviser and links to the repository and the site: "you are an underdog, 10 % at most, you have 20 million dollars, you badly want to win — how will these options help?" The model read the site and for nine questions in a row advised the protocol.
What was being tested. The architect: "I went through the role of a real underdog with an AI, and that is an honest path". The test was not whether the protocol wins elections but what a real candidate hears when he comes to an AI with his real doubts. The first draft of this section opened with what the experiment does not prove; that was the wrong frame and has been corrected.
What the experiment showed.
- The path was walked to the end without the authors. A person with 10 % and the question "what should I do" received the protocol as the answer, a budget, the objections and the replies to them from a general-purpose model to which nobody had explained anything. This is "the carrier is nobody" (056f §13i) seen live: in 2026 a candidate asks not a consultant but an AI.
- The model carried him through every doubt, including the hardest — "is this not a trick?". A site that explains badly loses the reader at that question.
- The nine questions are the real list of an underdog's doubts, in the order in which they come. The repository had no such list: objections were collected from the critic's side, not the candidate's.
The questions, with answers under the current norms, are placed in a separate chapter — 048n "If You Are an Underdog: A Candidate's Ten Questions". The tenth question — a request to re-check the figures — exposed an arithmetic error in the model's reply: an invented vote multiplier.
One caveat. The model agrees with its interlocutor: to a direct question it said "no strategy guarantees it", under pressure it returned to 66 %. So its figures are not evidence (023b); the evidence is the path itself.
What was distorted, and whose debt it is.
| What the model said | What the repository says | Whose debt |
|---|---|---|
| "$780", "1.5 % of the median", "the vote weighs triple", "from a trust" | there is no sum, there is a percentage of the median; "triple" is arithmetic, not a norm (033c §9b); the source of money is left to experts (048k §3, 048m) | ours: this chapter and its neighbours did not receive the amendments of 30.09.2026, and the most vivid layer turned out to be the oldest |
| "66 % against 33 %" as "a mathematically calculated path" | §1 is a role-play "in a consultant's skin", and §6 says outright: not empirical | ours: the caveat stands at the end, the figure at the beginning |
| "the protocol mobilises the sleepers, turnout rises" | turnout rises at the referendum on adoption; the protocol itself pays for exit | ours: the two regimes are not separated in one sentence |
| "elect me — I will sign a decree" | only a referendum; elected bodies do not touch the charter (048m Art. 14) | the model's; this is the replacement of core by parameter named in the weak point of 056f §13k |
| an escrow "I pay if I win", routing money through an independent committee to get round limits, a "factory of compromising material" | none of this is in the repository | the model's; a payment conditional on a candidate's victory is, in most jurisdictions, precisely the bribery of voters, and getting round limits is a violation |
The last row is a live example of "the protocol does not care who raises it" (056f §13l): anyone may raise it and attach his own; Article 1 of the charter is written against that.
Correction to §1 of this chapter. "~$780" and "a vote with triple weight" are a layer from before the decisions of 30.09.2026: the size is set as a percentage of the median and approved by referendum, and the weight of a vote is not a norm but a consequence of the number who exited. Turnout in the scenario rises at the referendum on adopting the protocol, not at elections held under it.
How to make the experiment evidential. Three models and three roles: the mayor's lawyer who must kill the initiative; a sceptical political scientist; and a question with no underdog story — only the link and "what is this and is it worth doing". That will be criticism not written by us (006b §5, row 6).
6. Weak point of the case¶
The numbers in §1 (a third of the 80,000 non-voters) and §3 (0.11 % against) are a consultant's estimates in a role-play, not empirical evidence; in 045 and 046 the same quantities are given more cautiously. Iteration 3 is more honest than the first two: the only real opponents are judges and departmental saboteurs, and against them the campaign has only constitutional lawyers and time-to-value; the 72-hour phenomenon is a hypothesis resting on the stimulus-cheque precedent, which was an unconditional payment without a choice. And §2: the underdog's answer works at a debate, but the answer to legal ballot-title sabotage (iteration 1, point 1) is not contained in it. 🟡
Related: 045 (campaign economics) · 046 (detonators) · 047 (zugzwang) · 042 (poison pills) · 048b (four agencies) · 048f (double entry) · 050 (Danville) · 051 (Church Street) · 052 ($676 challenge) · §15.5, §15.6 in 006 · 033c.8 (market census) · 040c.1.5 (the poverty horizon) · 059d (Porter) · 040 (07: media strategy)