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.
Rollback after Launch: Probabilities, Four Paths, Who Votes to Repeal¶
Chapter: 08 — Implementation File: 08_048g · v1 · 17 September 2026 (Grok dialogue, session 15-09-26) Source: a series of analyses "rollback in percentages", "who would vote to repeal and why", a comparison of durability with mass-payment institutions; the architect's correction "voting for rollback is voting against your own prize". Supplements 036.85.7 (irreversibility as the sixth parameter — qualitatively, five reasons) and 042 (poison pills). Here — numbers, paths and the composition of rollback supporters. All percentages are model estimates.
An early-draft layer (note of 01.10.2026 after audit 040m). "$1,000 or a x3 vote" is an example. The shares for rollback (7–12 %) and for retention (85–90 %) are unsourced estimates; the rows of the table in §4 add up to more than the total and need recalculating (040m, finding 9). The exact-answers sheet 1d and the charter 048m are in force.
1. Base estimate¶
After launch, rolling back is harder than not adopting, but not impossible.
| Scenario | First estimate | Recount by the protocol's logic |
|---|---|---|
| Holds, not rolled back | 70–80 % | 85–90 % |
| A rollback attempt will happen | 55–70 % | high |
| Successful full rollback | 15–25 % | 10–15 % after 1 cycle; 5–10 % after 2–3 |
| Partial hollowing of parameters | 25–35 % | the main risk (§3) |
Why a full rollback is not the most likely: after launch a large group appears that benefits from keeping the system — those who take the money (a direct minus from cancelling the payment), those who stay to vote (a direct minus from losing weight), and politicians who have already embedded themselves in the new rules and do not want the old uncertainty back. The first estimate drew its figures from a world where the fork does not yet exist; the recount proceeds from the fact that rollback hits two large groups at once: "why would I cancel my own $1,000?" and "why would I cancel my own x3 vote?". It is hard to assemble a majority against that.
2. By cycle¶
| Cycle | Rollback risk |
|---|---|
| First | Medium: much fear and many attacks, but also a strong novelty effect |
| Second | Lower: people have already seen the payment and the amplified vote |
| Third and beyond | Lower still for full repeal, higher for attempts to tweak the rules |
3. Four rollback paths¶
| Path | Chance of success |
|---|---|
| A new referendum "repeal completely" | 10–20 % |
| Court, ban | 15–25 % |
| Substitution of the formula, the fund, the data source | 25–35 % |
| Political hollowing without formal repeal | 30–40 % |
A correction to the last row: 30–40 % is an estimate for a world without watchmen; after the protocol, with two watchmen and a press whose audience votes, durable success of quiet spoiling is put at about 10 % (§3c). The most dangerous path is still not an honest rollback but quiet spoilage of the mechanism: keep the signboard and ruin the formula. That is why protecting the parameters matters more than speeches about irreversibility: the five statute parameters (059e.3), the four-agency rule and the 10 % rule (048b), the budget identity (048f.3.2). Full rollback is roughly one in eight to one in ten, not "fifty-fifty". But quiet spoilage is one in three, and that is the most underrated row of the table.
3b. The Georgian lesson: a personal sum versus an abstract parameter¶
In 2011 Georgia wrote a lock into its constitution: new taxes and rate increases only by referendum, spending no higher than 30 % of GDP, the deficit no higher than 3 %, debt no higher than 60 %; the norm kept a proviso "except for excise tax", and the rollback came through it (019c.6h). The architect: "all Georgia's parameters are impersonal and abstract to a villager; that is why there was a rollback. How would they roll back AB-EXIT?"
The diagnosis is exact. Nobody receives "thirty per cent of GDP" in his account; a breach of such a norm is noticed by no one but an economist, and there is nobody to defend it. The sum D is received by everyone who took it, remembered to the leu and compared with the last one (033c.9). Any spoiling of the mechanism that reduces that sum, delays it or narrows the circle of recipients hits the majority personally and is visible the same day. That is the difference between a lock on paper and a lock with a watchman.
So the honest answer to "how would they roll it back" is: not through the sum, but around it. Gaps of the same kind as Georgia's brackets:
| Gap | What it looks like | What closes it |
|---|---|---|
| Deduct on the other side | D is paid in full but counted as income: benefits, heating subsidies and concessions are cut; or a charge is introduced that falls mainly on recipients | A statute norm: D is disregarded in assessing any payments and is untaxed; this must be written in advance, because this gap is the most probable |
| Pay in something other than money | Certificates, bonds, offsetting utility debts | "In money to the account" — in the text of the statute, not in a regulation |
| Narrow the circle | "Except debtors", "except those living abroad", "except those not registered" | Non-selectivity of the payment as one of the five parameters (059e.3); any "except" only by referendum |
| Spoil the median | Change the methodology, the source, the publishing agency | The real median by a fixed methodology, the rule of four agencies (048b) |
| Make the choice difficult | Payment only on personal appearance, a short window, a complicated form; or making the return to voting difficult | The deadline and method of filing in the statute; one and the same channel for both buttons |
| Do not hold the election | A state of emergency, a postponement: no election, no payment | The payment is tied to the calendar of the cycle, not to the fact of voting |
Who exactly would go through the gaps. The architect's question about the table above: "and who, in your account, would be trying to roll it back?" The table is written in the impersonal, and that is its flaw. The answer splits in two. At a referendum to "repeal", it is the groups of §4 that vote, and they are 7–12 %: the machines and their core, principled opponents, status losers. A rollback does not pass that way, and that is exactly why nobody serious will choose it. Others go through the gaps — those who need no referendum because they have access to the text of the law and the regulations:
| Who | Motive | Which gap is open to them |
|---|---|---|
| The ministry of finance under any government | In a hard year D is a large line one is tempted to "optimise" | Count D as income for benefits; pay in something other than money; shift the date |
| Recipients of the levies that have begun to be repealed (019c) | The owners of the gates lose income and are the best organised of all | Narrow the circle, make the choice difficult — anything that brings the indifferent back to the polls |
| A parliamentary majority facing defeat | To get cheap votes back by the next election | "Technical amendments" to filing procedure and deadlines |
| The agency that computes the median | Pressure from above, not its own interest | The methodology and the data source |
| A government in crisis | To hold on | A state of emergency and postponement of the election |
What they have in common is that none of them would win an open vote, and all of them know it; so the attempt will look not like a rollback but like a clarification, an optimisation or a temporary measure. Georgia's excise of 2026 looked the same — not "repeal of the reform" but "an environmental measure against old cars".
Georgia's place on the scale of §5. The law of durability stated there — reforms are rolled back easily when the benefit is abstract, elite and procedural, and almost never when it is personal, regular, mass and understandable — gets in Georgia a tested example for its first half: a constitutional lock on taxes belongs to the row "procedural reforms" with a rollback chance of 20–40 %, and it lived more than ten years until the first successful gap. By the same law the protocol stands in a different row, but only in the part where the sum serves as watchman; gaps that do not touch the sum remain at the level of a procedural reform, and they must be closed by the text of the statute, not by hope in the watchman.
The general rule, which Georgia demonstrated at the cost of its own reform: do not write a closed list of exceptions at all. Every "except" in the statute is a future door for a rollback, and that is the door they will go through. And second: the first row of the table differs from the rest in that it does not touch the sum itself, so the watchman may not notice it. This is the main place where a personal sum does not defend itself. 🟡
3c. Two watchmen: why spoiling gets caught¶
The architect's objection to the table "who would go through the gaps" (§3b): "so the very government, most of which came to power only thanks to AB-EXIT, will start burying it while the people watch passively? Or perhaps at the next election the people, without changing AB-EXIT, will choose a new government. And the old one will know for certain that it will be caught — by those who did not take the money."
The objection is accepted, and it points to the same error as before: the table of §3b describes the actors as they behave today — a ministry nobody questions and a majority answerable to an indifferent voter. After the protocol these are different people in a different position. The parliamentary majority was elected by those who stayed and rests on them; bringing the indifferent back to the polls means devaluing its own voter's vote. The finance minister is appointed by that majority and answers to it. The mechanism would have to be spoiled by those it brought to power, before the eyes of those who elected them.
Hence a correction to §3b: there are two watchmen, not one.
| Watchman | What he guards | Which gaps he notices |
|---|---|---|
| Those who took the money | The sum: its size, date, circle of recipients | Payment in something other than money, narrowing of the circle, spoiling of the median, delay; and counting D against benefits — a person notices a cut heating subsidy just as personally as a cut dividend |
| Those who stayed to vote | The weight of their vote: that the indifferent are not brought back to the polls for free | Making choice B difficult, "technical amendments" to deadlines and procedure, any measure that restores cheap turnout |
The first version of §3b called counting D against benefits the place where "the watchman may not notice". That is inexact: the loss is noticed; what is unclear is only whom to blame for it. But assigning blame is the opposition's job, and after the protocol it has better conditions for it than ever: the voter who stayed counts money and listens to figures (040b.2b), and the promise "I will return what was quietly taken from you" belongs to the same class of cheap verifiable promises as "repeal the levy" (019c.3). Spoiling done by regulation is undone by the next government's regulation; spoiling that requires changing the rules requires a referendum in which everyone votes (§6b), including those being deprived.
So the estimate in §3 should be read as follows: an attempt at quiet spoiling remains likely — the temptation of the ministry in a hard year and of the levy recipients goes nowhere. But "one in three" referred to success in a world without watchmen. With two watchmen and an election between the attempt and its entrenchment, the assistant first put durable success of spoiling at 10–20 %, and after the arguments about the press lower — at the bottom of that range, about 10 %; the typical outcome is spoiling for one cycle, after which it is reversed and punished at the polls. This is an estimate, not a calculation, and here is what it is made of.
| Step | Estimate of durable success of quiet spoiling | What changed the estimate |
|---|---|---|
| Initial (§3, from the dialogue analysed) | 30–40 % | A world without watchmen: nobody questions the ministry, the majority answers to an indifferent voter |
| Two watchmen | 10–20 % | The spoiling would have to be done by those the protocol brought to power, before the eyes of those who elected them; those who took the money notice any blow to the sum, including offsetting against benefits; those who stayed notice any attempt to restore cheap turnout |
| The press, debates, the opposition | about 10 % | Exposing someone else's gap is the cheapest way to win votes in a parliament elected by people who count money; the reader of investigations is the one who votes; a channel is made in an hour, while a bought television channel broadcasts to those who do not vote |
What this figure does not cover: spoiling that is hard to explain (the median's methodology), a government with nothing left to lose before an election, and cancellation of the election itself. The first two are closed by the text of the statute; the third is not closed by a ballot at all (048d). So "about 10 %" is an estimate for a working system with deadlines, channels and a rule of independent recomputation written into the statute; without them it returns to 10–20 %.
The press and debates. The architect: "you have also left out the press and debates entirely. Any investigation, any case of corruption or of quietly breaking the system will be voiced both by part of the government and, all the more, by the voters who did not take the money — and they are the ones who will watch it. The effect is nothing like today's silence and sleep." The omission is substantial: the two watchmen above are described as if each noticed spoiling by himself. In reality, between the gap and the watchman stands whoever shows it, and after the protocol he has a different audience.
The power of the press is its audience multiplied by the audience's votes. Today an investigation into a swapped methodology or into payments being offset is read by a minority, while elections are decided by a majority that did not read it: hence "silence and sleep" — not because nobody wrote, but because what was written does not turn into votes. After the protocol the proportion is reversed: those who do not read investigations have mostly taken the sum, and those who read are the ones who vote. It is the same change as in 040c.3b, only with the opposite sign: the yellow press loses its lever because its reader left with the sum; the investigative press gains one because its reader stayed. The protocol re-weights not only voters but newsrooms.
The second channel is the government itself. Spoiling will first be named not by a journalist but by an opposition deputy: in a parliament elected by people who count money, exposing someone else's gap is the cheapest way to win votes, and gaps will be hunted professionally, with access to documents. "Part of the government will voice it" is not a hope in conscience but competition.
So spoiling that counts on going unnoticed is, after the protocol, badly calculated: there are those who profit from finding it, those who profit from telling it, and those who will listen and vote. The first version of this paragraph listed as a remainder "a press owned by the same levy recipients". The architect: "the press is a YouTube channel or a Twitter account: made in an hour, gaining subscribers in a week." The correction is accepted: the ownership argument comes from a world where the press was a printing works and a frequency; today entry costs almost nothing, one can buy a television channel but not the possibility of starting a channel, and an investigation into a withdrawn subsidy with the document on screen spreads by itself (048e.6). What is scarce is not the platform but an audience that cares — and that is precisely what the protocol creates. And a second observation of the architect's: "grandma flips through the television channels, while the one who did not take the money does not know where his channels are; he watches broadcast TV only to analyse the old government." The audiences of the two kinds of press part along the same line as the A/B choice: a bought television channel after the protocol broadcasts mostly to those who do not vote. Owning it ceases to be a political asset not because it was banned but because its viewer took the sum. The first version kept a caveat here: the elderly are the most disciplined voters, and some will stay to vote out of habit and duty. The architect: "discipline against money, with a minimum pension the populists themselves set so as to give the poor handouts before elections? Discipline versus half a pension — in my view there is not even a doubt." A check of the figures removes the caveat.
| Moldova | |
|---|---|
| Pensioners | 672 thousand, about 28 % of the population |
| Average old-age pension | about 4,400 lei |
| Minimum old-age pension | 3,055–3,525 lei |
| The sum D by the formula (033c.9c) | 2–2.5 thousand lei: more than half the minimum pension and about half the average |
And there is a market test better than any poll (in detail and with other countries — 019d). In autumn 2024, according to the Moldovan police, about 138 thousand citizens received 39 million dollars through a sanctioned Russian bank for voting in the referendum and the election — about 280 dollars on average; the base rate was 2,000 lei for minimal tasks and 5,000 if the polling station voted the required way. Two thousand lei is exactly D. For that sum people accepted criminal risk, registration through a bot, an account in a foreign bank and the role of executor of someone else's will; the number of recipients is comparable to a tenth of all who came to the polls. The protocol offers the same sum lawfully, from one's own state, with no errands and no shame.
Hence the answer to "are polls needed". For the direction — no: the price of a vote in the country has already been measured, and under the worst conditions for the seller. A poll or a pilot is needed for something else — what share of each group will take the sum, because the budget and the weight of those who stay depend on it. Half a pension or a whole one. The architect: "and is it really half a pension, not a whole pension — for medicines and presents for the grandchildren?" By the formula at K = 1 % it is from a half to three quarters: 2–2.5 thousand lei against a minimum pension of 3–3.5 thousand and an average of 4.4 thousand; the sum reaches a whole minimum pension at K of about 1.5 %, and K is a referendum parameter (033c.9c). But measuring against the whole pension is the wrong yardstick. Almost all of a minimum pension goes on food and utility bills: the subsistence minimum for a pensioner is on the order of 2.5 thousand lei (from the assistant's memory, unchecked), so a few hundred lei a month remain as free money. That is what one should count from: the sum D is a pensioner's free money for roughly half a year, received at once. "Medicines and presents for the grandchildren" is an exact description of what it will go on, because those are precisely the expenses the pension does not stretch to. And the other pan of the scales, in the architect's words: "or they will go and vote for Zyuganov for the tenth time, who never wins." The disciplined elderly voter often votes ritually — for a candidate who has not won in decades; he knows the expected result of such a vote from experience and values it at zero. His choice is not between "duty" and "money" but between a ritual without consequences and half a year of free money.
And the architect's remark on where this readiness comes from is worth recording as it stands: a low pension is not an accident but the condition under which a pre-election top-up works; a system that keeps a pensioner on three thousand lei has itself created the voter for whom half a pension decides everything. The protocol turns this against it: the same sensitivity to a sum is captured not by the one handing out the favour but by the person himself. What really remains: cheap entry is equally cheap for forgery — every investigation will meet counter-noise, and telling one from the other will fall to the same money-counting voter, for whom a verifiable figure matters more than loudness; the blocking of platforms — but that is already autocracy, not quiet spoiling within a working system; and spoiling that is hard to explain — a change in the median's methodology is understood less readily than "your subsidy was taken away", and what it needs is not a news report but a rule of independent recomputation written in advance (048b). 🟡
What remains, and it is no longer about a passive people. First, the lag: up to one cycle passes between the spoiling and the election, and a bad amendment gets to serve out the term. Second, a government that is already losing: the deterrent "you will be caught" does not work on someone with nothing to lose, and he is exactly the one who will try to bring cheap votes back just before the election; the defence here is deadlines and the procedure of choice written in the statute, not in a regulation. Third, cancelling the election itself on an emergency pretext: against that, watchmen with ballots are powerless, and this case already belongs to the scenario of force (048d). 🟡
4. Who would vote for rollback and why¶
| Group | Motive | Share of electorate (first estimate) |
|---|---|---|
| Machines, activists, technologists and their core | Regain control of the apathetic: the protocol takes away the fuel that can be delivered, frightened and mobilised | 3–5 % |
| Principled opponents of monetisation | Remove a morally unacceptable deal: "participation cannot be priced" | 4–7 % |
| Old-type politicians and their status supporters | Restore the rules under which they won with noise rather than results | 2–4 % |
| The disappointed | "They promised more": the payment seemed small, the formula "cheated", the amplification was not felt | 3–5 % |
| Those frightened by the consequences of the first cycle | Budget stress, a fund scandal, polarisation, "not everyone rules the city" | 3–6 % |
The first sum — 15–25 %, base guide 20 %. Note of 02.10.2026: all the shares in the table and in §1 are unsourced estimates; the table's rows add up to 15–27 %; the 7–12 % below results from taking the machines in full and only the hard part of the second and third rows; "10–15 %" in §1 is a different quantity, the probability of a successful rollback, not the share voting for it. The recount by the protocol's logic: those who really vote for rollback are only those who need neither option — the machines and their core, the hard principled, some of the status losers; the disappointed and the frightened, with a working payment, would be voting against their own prize, and that is almost never done. Result: for rollback 7–12 %, against rollback 88–93 %, base guide — one in ten, not one in five. If the first cycle fails (payments glitch, the formula is tweaked, a budget scandal, a sense of illegitimacy) — for rollback 30–40 %. So all the durability hangs on one condition: the money arrived and the vote really got heavier in the first cycle.
5. Comparison of durability¶
| Institution | Chance of full rollback once rooted |
|---|---|
| Social Security (US, 1935) | 1–3 % |
| The secret ballot | 1–2 % |
| Women's suffrage | ~1 % |
| Medicare (US, 1965) | 2–5 % |
| The Alaska PFD | 5–10 % |
| NHS-type mass healthcare | 5–10 % |
| The protocol after 1 working cycle | 10–15 % |
| The protocol after 2–3 cycles | 5–10 % |
| RCV, open primaries, procedural reforms | 20–40 % |
Reading scale: 1–5 % — almost iron; 5–15 % — very durable; 15–30 % — resilient but vulnerable; 30 % and above — easily rolled back. At a repeal referendum: an ordinary procedural reform — 40–55 % for repeal; a mass-payment system — 10–20 %; the protocol — 7–12 %. The law of durability: reforms are rolled back easily when the benefit is abstract, elitist, procedural; almost never when the benefit is personal, regular, mass, comprehensible. By that criterion the protocol is closer to Social Security and the PFD than to RCV — which is what 036.85.7 claims, only here it is expressed in comparable numbers rather than "close to zero".
6. Why the language of the dispute changes after launch¶
Before the protocol the question sounds "why don't you go and vote?". After — "why do you vote for free, if you can either take D or get a x3 vote?". The old moral scheme ("civic duty", "the vote is sacred", "one must participate for free") breaks not by argument but because the person sees a third path, and "for free as before" looks no longer like the norm but like an oddity. Non-attendance is no longer shame, participation no longer a duty without a price, influence an explicit asset. Once this has entered heads en masse, returning people to the world of "just vote" is almost impossible: after $1,000, explaining free voting as the only norm is rhetorically hopeless. After launch the protocol behaves not like an ordinary reform but like a meme that eats the old moral economy of participation.
The architect's correction: this begins before the law. As soon as an underdog has simply hammered in the formula "$1,000 or a x3 vote", both old options — voting for free and not voting for free — lose even before the referendum: the non-voter asks "why do they give nothing for my exit?", the voter — "why does my vote equal the vote of someone who doesn't care?", the systemic politician is forced to argue not with a candidate but with a sense of the old norm's unfairness. That is a different world in the head, even if the mechanism does not yet exist (059e.4: the 72-hour phenomenon; 048e.5: fake elections after people learned the number).
6b. Who votes on the rules: everyone — and why that is a lock¶
The architect's clarification: "everyone votes at the referendum — that is the hack of the system, because it is precisely the apathetic who want the money most, and almost all of them will be for the referendum; the government makes a noose for itself out of its own apathetic electorate. And to repeal it there must be a referendum at which the apathetic voluntarily give up easy money." This removes the uncertainty on which §1–5 silently rested: status B is a waiver of taking part in choosing managers for a cycle, not of the right to decide on the rules. Two levels:
| Level | Who decides | What is decided |
|---|---|---|
| Rules of the game | All citizens, including those who took the money | Adoption of the protocol, its repeal, any change to the five statute parameters (059e.3) |
| Governance | Those who stayed in this cycle | Who governs and how the budget is spent |
Three consequences. First — an asymmetric lock: adoption is helped by precisely the majority on which the old system rested, and to repeal the protocol that same majority has to be persuaded to vote against its own payment. Hence the 7–12 % for rollback in §4. Second — protection of those who took the money from those who stayed: the cohort that governs can neither cut the dividend, nor change the formula, nor take away the return of the right — all that lies one level above its powers. This answers the fear "the rich will remain and rewrite the rules for themselves" (059c.4; 034, strike 3). Third — in the terms of constitutional theory this is the separation of constituent power, which belongs to the people as a whole, from constituted power, which a part exercises; the protocol changes the composition of the second and leaves the first untouched.
The boundary. Other referendums — on policy questions, like the British one of 2016 — are not changed by the protocol: everyone votes in them, as now, and the price of that is shown in 055d. Relative to the order in force this is no worsening; the protocol acts on such referendums indirectly. The architect's clarification: "before the protocol — yes, it changes nothing; but after it, bright and overly dangerous actions make no sense for the electorate, because life is already getting easier for it thanks to the government's increased reasonableness". The working link here is the disappearance of anger, not a financial stake: a check against the British data shows that through the median a shock on the scale of Brexit would have changed the dividend by single pounds per cycle (055d.9).
7. Weak point of the section¶
All figures are projections; a before/after baseline will appear only with a pilot. The durability of §4–5 is derived from a working payment, that is, from a successful first cycle, while §3 shows that the most likely rollback path is precisely spoilage of the first cycle (substitution of the formula, data sabotage, delayed payments — 059e.3 iteration 3), after which the share for rollback rises to 30–40 %. The circle closes on protecting the parameters, not on "irreversibility" as a property. 🟡
Related: 036.85.7 (irreversibility as the sixth parameter) · 042 (poison pills) · 048b (four agencies, the 10 % rule) · 048f (double entry) · 059e.3 (five statute parameters) · 059e.4 (72 hours) · 048e.5 · 013c (the referendum as a game) · 015c.4 (mass-benefit institutions) · §14 (the Alaska PFD) · 037b (Grok)